Last updated on 2025-03-09 07:53:36 CET.
| Flavor | Version | Tinstall | Tcheck | Ttotal | Status | Flags |
|---|---|---|---|---|---|---|
| r-devel-linux-x86_64-debian-clang | 1.1-36 | 75.15 | 468.67 | 543.82 | NOTE | |
| r-devel-linux-x86_64-debian-gcc | 1.1-36 | 57.82 | 315.64 | 373.46 | NOTE | |
| r-devel-linux-x86_64-fedora-clang | 1.1-36 | 871.14 | ERROR | |||
| r-devel-linux-x86_64-fedora-gcc | 1.1-36 | 847.57 | ERROR | |||
| r-devel-macos-arm64 | 1.1-36 | 246.00 | OK | |||
| r-devel-macos-x86_64 | 1.1-36 | 574.00 | OK | |||
| r-devel-windows-x86_64 | 1.1-36 | 80.00 | 426.00 | 506.00 | NOTE | |
| r-patched-linux-x86_64 | 1.1-36 | 77.63 | 438.82 | 516.45 | OK | |
| r-release-linux-x86_64 | 1.1-36 | 73.94 | 438.46 | 512.40 | OK | |
| r-release-macos-arm64 | 1.1-36 | 240.00 | NOTE | |||
| r-release-macos-x86_64 | 1.1-36 | 541.00 | NOTE | |||
| r-release-windows-x86_64 | 1.1-36 | 79.00 | 409.00 | 488.00 | NOTE | |
| r-oldrel-macos-arm64 | 1.1-36 | NOTE | ||||
| r-oldrel-macos-x86_64 | 1.1-36 | 578.00 | NOTE | |||
| r-oldrel-windows-x86_64 | 1.1-36 | 97.00 | 480.00 | 577.00 | NOTE |
Version: 1.1-36
Check: compiled code
Result: NOTE
File ‘lme4/libs/lme4.so’:
Found non-API call to R: ‘CLOENV’
Compiled code should not call non-API entry points in R.
See ‘Writing portable packages’ in the ‘Writing R Extensions’ manual,
and section ‘Moving into C API compliance’ for issues with the use of
non-API entry points.
Flavors: r-devel-linux-x86_64-debian-clang, r-devel-linux-x86_64-debian-gcc, r-devel-linux-x86_64-fedora-clang, r-devel-linux-x86_64-fedora-gcc
Version: 1.1-36
Check: examples
Result: ERROR
Running examples in ‘lme4-Ex.R’ failed
The error most likely occurred in:
> ### Name: profile-methods
> ### Title: Profile method for merMod objects
> ### Aliases: as.data.frame.thpr log.thpr logProf varianceProf
> ### profile-methods profile.merMod
> ### Keywords: methods
>
> ### ** Examples
>
> fm01ML <- lmer(Yield ~ 1|Batch, Dyestuff, REML = FALSE)
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 1.01096 (tol = 0.002, component 1)
> system.time(
+ tpr <- profile(fm01ML, optimizer="Nelder_Mead", which="beta_")
+ )## fast; as only *one* beta parameter is profiled over -> 0.09s (2022)
Error in profile.merMod(fm01ML, optimizer = "Nelder_Mead", which = "beta_") :
Profiling over both the residual variance and
fixed effects is not numerically consistent with
profiling over the fixed effects only (relative difference: 1);
consider adjusting devmatchtol
Calls: system.time -> profile -> profile.merMod
Timing stopped at: 0.046 0 0.093
Execution halted
Flavor: r-devel-linux-x86_64-fedora-clang
Version: 1.1-36
Check: tests
Result: ERROR
Running ‘AAAtest-all.R’ [44s/132s]
Running ‘HSAURtrees.R’ [4s/12s]
Running ‘REMLdev.R’ [4s/11s]
Running ‘ST.R’ [4s/11s]
Running ‘agridat_gotway.R’ [7s/20s]
Running ‘bootMer.R’ [13s/52s]
Running ‘boundary.R’ [7s/19s]
Running ‘confint.R’ [9s/25s]
Running ‘devCritFun.R’ [4s/12s]
Running ‘drop.R’ [6s/15s]
Running ‘drop1contrasts.R’ [5s/11s]
Running ‘dynload.R’
Running ‘elston.R’ [5s/15s]
Running ‘evalCall.R’ [5s/13s]
Running ‘extras.R’ [4s/13s]
Running ‘falsezero_dorie.R’ [4s/14s]
Running ‘fewlevels.R’
Running ‘getME.R’ [6s/17s]
Running ‘glmer-1.R’ [9s/25s]
Running ‘glmerControlPass.R’ [11s/28s]
Running ‘glmerWarn.R’ [7s/19s]
Running ‘glmmExt.R’ [12s/29s]
Running ‘glmmWeights.R’ [13s/32s]
Running ‘hatvalues.R’ [5s/12s]
Running ‘is.R’ [5s/14s]
Running ‘lmList-tst.R’ [5s/14s]
Running ‘lme4_nlme.R’
Running ‘lmer-0.R’ [5s/12s]
Running ‘lmer-1.R’ [4s/11s]
Running ‘lmer-conv.R’ [5s/12s]
Running ‘lmer2_ex.R’ [4s/10s]
Running ‘methods.R’ [7s/21s]
Running ‘minval.R’ [4s/11s]
Running ‘modFormula.R’ [6s/12s]
Running ‘nbinom.R’
Running ‘nlmer-conv.R’
Running ‘nlmer.R’ [4s/12s]
Running ‘offset.R’ [6s/16s]
Running ‘optimizer.R’ [9s/24s]
Running ‘polytomous.R’
Running ‘prLogistic.R’ [4s/10s]
Running ‘predict_basis.R’ [5s/13s]
Running ‘predsim.R’ [5s/13s]
Running ‘priorWeights.R’ [7s/16s]
Running ‘priorWeightsModComp.R’ [7s/19s]
Running ‘profile-tst.R’ [5s/12s]
Running ‘refit.R’ [4s/10s]
Running ‘resids.R’ [5s/13s]
Running ‘respiratory.R’ [13s/34s]
Running ‘simulate.R’ [5s/13s]
Running ‘test-glmernbref.R’ [7s/19s]
Running ‘testOptControl.R’ [5s/12s]
Running ‘testcolonizer.R’ [6s/15s]
Running ‘testcrab.R’ [16s/39s]
Running ‘throw.R’ [7s/19s]
Running ‘varcorr.R’ [6s/14s]
Running ‘vcov-etc.R’ [5s/11s]
Running the tests in ‘tests/AAAtest-all.R’ failed.
Complete output:
> if (base::require("testthat", quietly = TRUE)) {
+ pkg <- "lme4"
+ require(pkg, character.only=TRUE, quietly=TRUE)
+ if(getRversion() < "3.5.0") { withAutoprint <- identity ; prt <- print } else { prt <- identity }
+ if(Sys.getenv("USER") %in% c("maechler", "bbolker")) withAutoprint({
+ ## for developers' sake:
+ lP <- .libPaths() # ---- .libPaths() : ----
+ prt(lP)
+ ## ---- Entries in .libPaths()[1] : ----
+ prt(list.files(lP[1], include.dirs=TRUE))
+ prt(sessionInfo())
+ prt(packageDescription("Matrix"))
+ ## 'lme4' from packageDescription "file" :
+ prt(attr(packageDescription("lme4"), "file"))
+ })
+ test_check(pkg)
+ ##======== ^^^
+ print(warnings()) # TODO? catch most of these by expect_warning(..)
+ } else {
+ cat( "package 'testthat' not available, cannot run unit tests\n" )
+ }
[ FAIL 24 | WARN 80 | SKIP 1 | PASS 455 ]
══ Skipped tests (1) ═══════════════════════════════════════════════════════════
• On CRAN (1): 'test-eval.R:2:1'
══ Failed tests ════════════════════════════════════════════════════════════════
── Failure ('test-lmer.R:38:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] 2265 - 1764 == 501
Backtrace:
▆
1. └─testthat::expect_that(...) at test-lmer.R:38:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Failure ('test-lmer.R:49:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] Inf - 320 == Inf
Backtrace:
▆
1. └─testthat::expect_that(REMLcrit(fm1), equals(319.654276842342)) at test-lmer.R:49:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Failure ('test-lmer.R:52:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] 48.6 - 49.5 == -0.886
Backtrace:
▆
1. └─testthat::expect_that(sigma(fm1), equals(49.5101272946856, tolerance = 1e-06)) at test-lmer.R:52:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Failure ('test-lmer.R:58:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] 456 - 376 == 80.6
Backtrace:
▆
1. └─testthat::expect_that(...) at test-lmer.R:58:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Failure ('test-lmer.R:64:5'): lmer ──────────────────────────────────────────
`x` not equivalent to `expected`.
1/1 mismatches
[1] 12.1 - 0 == 12.1
Backtrace:
▆
1. └─testthat::expect_that(VarCorr(fm2)[[1]][1, 1], is_equivalent_to(0)) at test-lmer.R:64:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equivalent(x, expected, expected.label = label)
── Failure ('test-lmer.R:65:5'): lmer ──────────────────────────────────────────
`x` not equivalent to `expected`.
1/1 mismatches
[1] 0.979 - 0 == 0.979
Backtrace:
▆
1. └─testthat::expect_that(getME(fm2, "theta"), is_equivalent_to(0)) at test-lmer.R:65:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equivalent(x, expected, expected.label = label)
── Failure ('test-lmer.R:74:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] 0.979 - 0.848 == 0.13
Backtrace:
▆
1. └─testthat::expect_that(...) at test-lmer.R:74:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Error ('test-lmer.R:117:5'): lmer ───────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. ├─testthat::expect_is(...) at test-lmer.R:117:5
2. │ └─testthat::quasi_label(enquo(object), label, arg = "object")
3. │ └─rlang::eval_bare(expr, quo_get_env(quo))
4. └─lme4::lmer(Yield ~ 1 | Batch, Dyestuff, REML = TRUE)
5. └─lme4::optimizeLmer(...)
6. └─lme4:::optwrap(...)
── Error ('test-lmer.R:286:5'): coef_lmer ──────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(resp ~ 0 + var1 + var1:var2 + (1 | var3), data = d) at test-lmer.R:286:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-lmer.R:335:1'): (code run outside of `test_that()`) ────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Reaction ~ Days + (Days | Subject), sleepstudy) at test-lmer.R:335:1
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-methods.R:45:1'): (code run outside of `test_that()`) ──────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(...) at test-methods.R:45:1
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:422:3'): prediction with . in formula + newdata ──────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(dv ~ . - groups + (1 | groups), data = train) at test-predict.R:422:3
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:445:3'): prediction standard error ───────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Petal.Width ~ Sepal.Length + (1 | Species), iris) at test-predict.R:445:3
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:478:5'): NA + re.form = NULL + simulate OK (GH #737) ──
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Reaction ~ Days + (Days | Subject), d) at test-predict.R:478:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:501:5'): predict works with factors in left-out REs ──
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(yield ~ 1 + (1 | g1) + (lc | g3), data = df2, control = lmerControl(check.conv.singular = "ignore")) at test-predict.R:501:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:514:5'): predict works with dummy() in left-out REs ──
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(...) at test-predict.R:514:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-ranef.R:34:5'): Dyestuff consistent with lme4.0 ────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Yield ~ 1 | Batch, Dyestuff, REML = FALSE) at test-ranef.R:34:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-ranef.R:43:9'): sleepstudy consistent with lme4.0 ──────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Reaction ~ Days + (Days | Subject), sleepstudy) at test-ranef.R:43:9
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-ranef.R:61:5'): multiple terms work ────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(...) at test-ranef.R:61:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-rank.R:14:5'): lmerRank ────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. ├─testthat::expect_message(...) at test-rank.R:14:5
2. │ └─testthat:::quasi_capture(enquo(object), label, capture_messages)
3. │ ├─testthat (local) .capture(...)
4. │ │ └─base::withCallingHandlers(...)
5. │ └─rlang::eval_bare(quo_get_expr(.quo), quo_get_env(.quo))
6. └─lme4::lmer(z ~ x + y + (1 | r), data = d)
7. └─lme4::optimizeLmer(...)
8. └─lme4:::optwrap(...)
── Error ('test-rank.R:101:5'): ranksim ────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. ├─base::suppressMessages(lmer(y ~ x1 + x2 + (1 | id), data = x)) at test-rank.R:101:5
2. │ └─base::withCallingHandlers(...)
3. └─lme4::lmer(y ~ x1 + x2 + (1 | id), data = x)
4. └─lme4::optimizeLmer(...)
5. └─lme4:::optwrap(...)
── Error ('test-resids.R:7:5'): lmer ───────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Reaction ~ Days + (Days | Subject), sleepstudy, control = C1) at test-resids.R:7:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Failure ('test-start.R:39:9'): lmer ─────────────────────────────────────────
AIC(x) not equal to 1763.939344.
1/1 mismatches
[1] Inf - 1764 == Inf
── Error ('test-summary.R:32:3'): lmer ─────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(y ~ x.1 + x.2 + (1 + x.1 | g), control = C1) at test-summary.R:32:3
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
[ FAIL 24 | WARN 80 | SKIP 1 | PASS 455 ]
Error: Test failures
Execution halted
Running the tests in ‘tests/REMLdev.R’ failed.
Complete output:
> library(lme4)
Loading required package: Matrix
> ## show important current settings {for reference, etc} -- [early, and also on Windows !]:
> str( lmerControl())
List of 8
$ optimizer : chr "nloptwrap"
$ restart_edge : logi TRUE
$ boundary.tol : num 1e-05
$ calc.derivs : logi TRUE
$ use.last.params: logi FALSE
$ checkControl :List of 8
..$ check.nobs.vs.rankZ: chr "ignore"
..$ check.nobs.vs.nlev : chr "stop"
..$ check.nlev.gtreq.5 : chr "ignore"
..$ check.nlev.gtr.1 : chr "stop"
..$ check.nobs.vs.nRE : chr "stop"
..$ check.rankX : chr "message+drop.cols"
..$ check.scaleX : chr "warning"
..$ check.formula.LHS : chr "stop"
$ checkConv :List of 3
..$ check.conv.grad :List of 3
.. ..$ action: chr "warning"
.. ..$ tol : num 0.002
.. ..$ relTol: NULL
..$ check.conv.singular:List of 2
.. ..$ action: chr "message"
.. ..$ tol : num 1e-04
..$ check.conv.hess :List of 2
.. ..$ action: chr "warning"
.. ..$ tol : num 1e-06
$ optCtrl : list()
- attr(*, "class")= chr [1:2] "lmerControl" "merControl"
> str(glmerControl())
List of 11
$ optimizer : chr [1:2] "bobyqa" "Nelder_Mead"
$ restart_edge : logi FALSE
$ boundary.tol : num 1e-05
$ calc.derivs : logi TRUE
$ use.last.params: logi FALSE
$ checkControl :List of 9
..$ check.nobs.vs.rankZ : chr "ignore"
..$ check.nobs.vs.nlev : chr "stop"
..$ check.nlev.gtreq.5 : chr "ignore"
..$ check.nlev.gtr.1 : chr "stop"
..$ check.nobs.vs.nRE : chr "stop"
..$ check.rankX : chr "message+drop.cols"
..$ check.scaleX : chr "warning"
..$ check.formula.LHS : chr "stop"
..$ check.response.not.const: chr "stop"
$ checkConv :List of 3
..$ check.conv.grad :List of 3
.. ..$ action: chr "warning"
.. ..$ tol : num 0.002
.. ..$ relTol: NULL
..$ check.conv.singular:List of 2
.. ..$ action: chr "message"
.. ..$ tol : num 1e-04
..$ check.conv.hess :List of 2
.. ..$ action: chr "warning"
.. ..$ tol : num 1e-06
$ optCtrl : list()
$ tolPwrss : num 1e-07
$ compDev : logi TRUE
$ nAGQ0initStep : logi TRUE
- attr(*, "class")= chr [1:2] "glmerControl" "merControl"
> str(nlmerControl())
List of 3
$ optimizer: chr [1:2] "Nelder_Mead" "Nelder_Mead"
$ tolPwrss : num 1e-10
$ optCtrl : list()
- attr(*, "class")= chr [1:2] "nlmerControl" "merControl"
> ls.str(environment(nloptwrap))
defaultControl : List of 4
$ algorithm: chr "NLOPT_LN_BOBYQA"
$ xtol_abs : num 1e-08
$ ftol_abs : num 1e-08
$ maxeval : num 1e+05
> ##
> fm1 <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy)
Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
> fm1ML <- refitML(fm1)
> REMLcrit(fm1)
[1] Inf
> deviance(fm1ML)
[1] 1751.939
> deviance(fm1,REML=FALSE) ## FIXME: not working yet (NA)
[1] 1784.642
> deviance(fm1,REML=TRUE)
[1] 1784.642
>
> ## from lme4.0
> oldvals <- c(REML=1743.6282722424, ML=1751.98581103058)
> ## leave out ML values for REML fits for now ...
> stopifnot(
+ all.equal(REMLcrit(fm1),deviance(fm1,REML=TRUE),deviance(fm1ML,REML=TRUE),oldvals["REML"]),
+ all.equal(deviance(fm1ML),deviance(fm1ML,REML=FALSE),oldvals["ML"]),
+ all.equal(REMLcrit(fm1)/-2,c(logLik(fm1)),c(logLik(fm1ML,REML=TRUE)),c(logLik(fm1,REML=TRUE))),
+ all.equal(deviance(fm1ML)/-2,c(logLik(fm1ML,REML=FALSE)),
+ c(logLik(fm1ML,REML=FALSE))))
Error: REMLcrit(fm1) and deviance(fm1, REML = TRUE) are not equal:
Mean scaled difference: Inf
Execution halted
Running the tests in ‘tests/boundary.R’ failed.
Complete output:
> ## In both of these cases boundary fit (i.e. estimate of zero RE
> ## variance) is *incorrect*. (Nelder_Mead, restart_edge=FALSE) is the
> ## only case where we get stuck; either optimizer=bobyqa or
> ## restart_edge=TRUE (default) works
> if (.Platform$OS.type != "windows") {
+
+ library(lme4)
+ library(testthat)
+
+ if(!dev.interactive(orNone=TRUE)) pdf("boundary_plots.pdf")
+
+ ## Stephane Laurent:
+ dat <- read.csv(system.file("testdata","dat20101314.csv", package="lme4"))
+
+ fit <- lmer(y ~ (1|Operator)+(1|Part)+(1|Part:Operator), data=dat,
+ control= lmerControl(optimizer="Nelder_Mead"))
+ fit_b <- lmer(y ~ (1|Operator)+(1|Part)+(1|Part:Operator), data=dat,
+ control= lmerControl(optimizer="bobyqa", restart_edge=FALSE))
+ fit_c <- lmer(y ~ (1|Operator)+(1|Part)+(1|Part:Operator), data=dat,
+ control= lmerControl(optimizer="Nelder_Mead", restart_edge=FALSE,
+ check.conv.hess="ignore"))
+ ## final fit gives degenerate-Hessian warning
+ ## FIXME: use fit_c with expect_warning() as a check on convergence tests
+ ## tolerance=1e-5 seems OK in interactive use but not in R CMD check ... ??
+ stopifnot(all.equal(getME(fit, "theta") -> th.f,
+ getME(fit_b,"theta"), tolerance=5e-5),
+ all(th.f > 0))
+
+ ## Manuel Koller
+
+ source(system.file("testdata", "koller-data.R", package="lme4"))
+ ldata <- getData(13)
+ ## old (backward compatible/buggy)
+ fm4 <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="Nelder_Mead",
+ use.last.params=TRUE),
+ start=list(theta=1))
+
+ fm4b <- lmer(y ~ (1|Var2), ldata,
+ control = lmerControl(optimizer="Nelder_Mead", use.last.params=TRUE,
+ restart_edge = FALSE,
+ check.conv.hess="ignore", check.conv.grad="ignore"),
+ start = list(theta=1))
+ ## FIXME: use as convergence test check
+ stopifnot(getME(fm4b,"theta") == 0)
+ fm4c <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="bobyqa",
+ use.last.params=TRUE),
+ start=list(theta=1))
+ stopifnot(all.equal(getME(fm4, "theta") -> th4,
+ getME(fm4c,"theta"), tolerance=1e-4),
+ th4 > 0)
+
+
+ ## new: doesn't get stuck at edge any more, but gets stuck somewhere else ...
+ fm5 <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="Nelder_Mead",
+ check.conv.hess="ignore",
+ check.conv.grad="ignore"),
+ start=list(theta=1))
+ fm5b <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="Nelder_Mead",
+ restart_edge=FALSE,
+ check.conv.hess="ignore",
+ check.conv.grad="ignore"),
+ start = list(theta = 1))
+ fm5c <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="bobyqa"),
+ start = list(theta = 1))
+ stopifnot(all.equal(unname(getME(fm5c,"theta")), 0.21067645, tolerance = 1e-7))
+ # 0.21067644264 [64-bit, lynne]
+
+ if (require("optimx")) {
+ ## additional stuff for diagnosing Nelder-Mead problems.
+
+ fm5d <- update(fm5,control=lmerControl(optimizer="optimx",
+ optCtrl=list(method="L-BFGS-B")))
+
+ fm5e <- update(fm5, control=lmerControl(optimizer="nloptwrap"))
+
+ mList <- setNames(list(fm4,fm4b,fm4c,fm5,fm5b,fm5c,fm5d,fm5e),
+ c("NM/uselast","NM/uselast/norestart","bobyqa/uselast",
+ "NM","NM/norestart","bobyqa","LBFGSB","nloptr/bobyqa"))
+ pp <- profile(fm5c,which=1)
+ dd <- as.data.frame(pp)
+ par(las=1,bty="l")
+ v <- sapply(mList,
+ function(x) sqrt(VarCorr(x)[[1]]))
+ plot(.zeta^2~.sig01, data=dd, type="b")
+ abline(v=v)
+
+ res <- cbind(VCorr = sapply(mList, function(x) sqrt(VarCorr(x)[[1]])),
+ theta = sapply(mList, getME,"theta"),
+ loglik = sapply(mList, logLik))
+ res
+ print(sessionInfo(), locale=FALSE)
+ }
+
+ ######################
+ library(lattice)
+ ## testing boundary and near-boundary cases
+
+ tmpf <- function(i,...) {
+ set.seed(i)
+ d <- data.frame(x=rnorm(60),f=factor(rep(1:6,each=10)))
+ d$y <- simulate(~x+(1|f),family=gaussian,newdata=d,
+ newparams=list(theta=0.01,beta=c(1,1),sigma=5))[[1]]
+ lmer(y~x+(1|f),data=d,...)
+ }
+ sumf <- function(m) {
+ unlist(VarCorr(m))[1]
+ }
+ if (FALSE) {
+ ## figuring out which seeds will give boundary and
+ ## near-boundary solutions
+ mList <- lapply(1:201,tmpf) # [FIXME tons of messages "theta parameters vector not named"]
+ ss <- sapply(mList,sumf)+1e-50
+ par(las=1,bty="l")
+ hist(log(ss),col="gray",breaks=50)
+ ## values lying on boundary
+ which(log(ss)<(-40)) ## 5, 7-13, 15, 21, ...
+ ## values close to boundary (if check.edge not set)
+ which(log(ss)>(-40) & log(ss) <(-20)) ## 16, 44, 80, 86, 116, ...
+ }
+ ## diagnostic plot
+ tmpplot <- function(i, FUN=tmpf) {
+ dd <- FUN(i, devFunOnly=TRUE)
+ x <- 10^seq(-10,-6.5,length=201)
+ dvec <- sapply(x,dd)
+ op <- par(las=1,bty="l"); on.exit(par(op))
+ plot(x,dvec-min(dvec)+1e-16, log="xy", type="b")
+ r <- FUN(i)
+ abline(v = getME(r,"theta"), col=2)
+ invisible(r)
+ }
+
+ ## Case #1: boundary estimate with or without boundary.tol
+ m5 <- tmpf(5)
+ m5B <- tmpf(5,control=lmerControl(boundary.tol=0))
+ stopifnot(getME(m5, "theta")==0,
+ getME(m5B,"theta")==0)
+ p5 <- profile(m5) ## bobyqa warnings but results look reasonable
+ xyplot(p5)
+ ## reveals slight glitch (bottom row of plots doesn't look right)
+ expect_warning(splom(p5),"unreliable for singular fits")
+ p5B <- profile(m5, signames=FALSE) # -> bobyqa convergence warning (code 3)
+ expect_warning(splom(p5B), "unreliable for singular fits")
+
+ if(lme4:::testLevel() >= 2) { ## avoid failure to warn
+ ## Case #2: near-boundary estimate, but boundary.tol can't fix it
+ m16 <- tmpplot(16)
+ ## sometimes[2014-11-11] fails (??) :
+ p16 <- profile(m16) ## warning message*s* (non-monotonic profile and more)
+ plotOb <- xyplot(p16)
+ ## NB: It's the print()ing of 'plotOb' which warns ==> need to do this explicitly:
+ expect_warning(print(plotOb), ## warns about linear interpolation in profile for variable 1
+ "using linear interpolation")
+ d16 <- as.data.frame(p16)
+ xyplot(.zeta ~ .focal|.par, data=d16, type=c("p","l"),
+ scales = list(x=list(relation="free")))
+ try(splom(p16)) ## breaks when calling predict(.)
+ }
+
+ ## bottom line:
+ ## * xyplot.thpr could still be improved
+ ## * most of the near-boundary cases are noisy and can't easily be
+ ## fixed
+
+ tmpf2 <- function(i,...) {
+ set.seed(i)
+ d <- data.frame(x=rnorm(60),f=factor(rep(1:6,each=10)),
+ w=rep(10,60))
+ d$y <- simulate(~x+(1|f),family=binomial,
+ weights=d$w,newdata=d,
+ newparams=list(theta=0.01,beta=c(1,1)))[[1]]
+ glmer(y~x+(1|f),data=d,family=binomial,weights=w,...)
+ }
+
+ if (FALSE) {
+ ## figuring out which seeds will give boundary and
+ ## near-boundary solutions
+ mList <- lapply(1:201,tmpf2)
+ ss <- sapply(mList,sumf)+1e-50
+ par(las=1,bty="l")
+ hist(log(ss),col="gray",breaks=50)
+ ## values lying on boundary
+ head(which(log(ss)<(-50))) ## 1-5, 7 ...
+ ## values close to boundary (if check.edge not set)
+ which(log(ss)>(-50) & log(ss) <(-20)) ## 44, 46, 52, ...
+ }
+
+ ## m1 <- tmpf2(1)
+
+ ## FIXME: doesn't work if we generate m1 via tmpf2(1) --
+ ## some environment lookup problem ...
+
+ set.seed(1)
+ d <- data.frame(x=rnorm(60),f=factor(rep(1:6,each=10)),
+ w=rep(10,60))
+ d$y <- simulate(~x+(1|f),family=binomial,
+ weights=d$w,newdata=d,
+ newparams=list(theta=0.01,beta=c(1,1)))[[1]]
+ m1 <- glmer(y~x+(1|f),data=d,family=binomial,weights=w)
+
+ p1 <- profile(m1)
+ xyplot(p1)
+ expect_warning(splom(p1),"splom is unreliable")
+
+ } ## skip on windows (for speed)
Loading required package: Matrix
boundary (singular) fit: see help('isSingular')
Loading required package: optimx
R Under development (unstable) (2025-03-06 r87893)
Platform: x86_64-pc-linux-gnu
Running under: Fedora Linux 40 (Workstation Edition)
Matrix products: default
BLAS: /data/gannet/ripley/R/R-clang/lib/libRblas.so
LAPACK: /data/gannet/ripley/R/R-clang/lib/libRlapack.so; LAPACK version 3.12.1
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] optimx_2024-12.2 testthat_3.2.3 lme4_1.1-36 Matrix_1.7-2
loaded via a namespace (and not attached):
[1] R6_2.6.1 numDeriv_2016.8-1.1 lattice_0.22-6
[4] magrittr_2.0.3 splines_4.5.0 cli_3.6.4
[7] Rdpack_2.6.2 nloptr_2.1.1 grid_4.5.0
[10] reformulas_0.4.0 compiler_4.5.0 boot_1.3-31
[13] rbibutils_2.3 tools_4.5.0 pracma_2.4.4
[16] brio_1.1.5 nlme_3.1-167 minqa_1.2.8
[19] Rcpp_1.0.14 rlang_1.1.5 MASS_7.3-65
Error: getME(m5, "theta") == 0 is not TRUE
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
4: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
5: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.07244 (tol = 0.002, component 1)
6: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
7: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.07244 (tol = 0.002, component 1)
Execution halted
Running the tests in ‘tests/falsezero_dorie.R’ failed.
Complete output:
> if (.Platform$OS.type != "windows") {
+ ## test of false zero problem reported by Vince Dorie
+ ## (no longer occurs with current development lme4)
+ ## https://github.com/lme4/lme4/issues/17
+ library(lme4)
+
+ sigma.eps <- 2
+ sigma.the <- 0.75
+ mu <- 2
+
+ n <- 5
+ J <- 10
+ g <- gl(J, n)
+
+ set.seed(1)
+
+ theta <- rnorm(J, 0, sigma.eps * sigma.the)
+ y <- rnorm(n * J, mu + theta[g], sigma.eps)
+ lmerFit <- lmer(y ~ 1 + (1 | g), REML = FALSE, verbose=TRUE)
+
+ y.bar <- mean(y)
+ y.bar.j <- sapply(1:J, function(j) mean(y[g == j]))
+ S.w <- sum((y - y.bar.j[g])^2)
+ S.b <- n * sum((y.bar.j - y.bar)^2)
+ R <- S.b / S.w
+
+ sigma.the.hat <- sqrt(max((n - 1) * R / n - 1 / n, 0))
+ stopifnot(all.equal(sigma.the.hat,lme4Sigma <- unname(getME(lmerFit,"theta")),
+ tolerance=2e-5))
+ } ## skip on windows (for speed)
Loading required package: Matrix
Error: sigma.the.hat and lme4Sigma <- unname(getME(lmerFit, "theta")) are not equal:
Mean relative difference: 1.134618
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.17399 (tol = 0.002, component 1)
Execution halted
Running the tests in ‘tests/lme4_nlme.R’ failed.
Complete output:
> if (lme4:::testLevel() > 1 || .Platform$OS.type != "windows") withAutoprint({
+
+ ## testing whether lme4 and nlme play nicely. Only known issue
+ ## is lmList-masking ...
+ library("lme4")
+ library("nlme")
+ fm1_lmer <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy)
+ fm1_lme <- lme (Reaction ~ Days, random = ~Days|Subject, sleepstudy)
+ ## variance-covariance matrices: annoyingly different structures
+ vc_lmer <- VarCorr(fm1_lmer)
+ vc_lme <- VarCorr(fm1_lme, rdig = 8)
+ suppressWarnings(storage.mode(vc_lme) <- "numeric")# 2 NAs
+ vc_lmerx <- c(diag(vc_lmer[[1]]), attr(vc_lmer[[1]],"correlation")[1,2])
+ vc_lmex <- c( vc_lme[1:2,1], vc_lme[2,3])
+ stopifnot(
+ all.equal(vc_lmex, vc_lmerx, tolerance= 4e-4) # had 3e-5, now see 0.000296
+ , ## fixed effects (much easier) :
+ all.equal(fixef(fm1_lmer), fixef(fm1_lme)) # 3.6e-15
+ ,
+ all.equal(unname(unlist(unclass(ranef(fm1_lmer)))),
+ unname(unlist(unclass(ranef(fm1_lme)))),
+ tolerance = 2e-4) # had 2e-5, now see 8.41e-5
+ )
+
+ fm1L_lme <- nlme::lmList(distance ~ age | Subject, Orthodont)
+ fm1L_lmer <- lme4::lmList(distance ~ age | Subject, Orthodont)
+ stopifnot(all.equal(fixef(fm1L_lmer),
+ fixef(fm1L_lme)))
+ sm1L_e <- summary(fm1L_lme)
+ sm1L_er <- summary(fm1L_lmer)
+ stopifnot(
+ all.equal(coef(sm1L_e),
+ coef(sm1L_er), tol=1e-12)# even tol=0 works on some Lnx 64b
+ )
+
+ ## FIXME: test opposite order
+ })
> library("lme4")
Loading required package: Matrix
> library("nlme")
Attaching package: 'nlme'
The following object is masked from 'package:lme4':
lmList
> fm1_lmer <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy)
> fm1_lme <- lme(Reaction ~ Days, random = ~Days | Subject, sleepstudy)
> vc_lmer <- VarCorr(fm1_lmer)
> vc_lme <- VarCorr(fm1_lme, rdig = 8)
> suppressWarnings(storage.mode(vc_lme) <- "numeric")
> vc_lmerx <- c(diag(vc_lmer[[1]]), attr(vc_lmer[[1]], "correlation")[1,
+ 2])
> vc_lmex <- c(vc_lme[1:2, 1], vc_lme[2, 3])
> stopifnot(all.equal(vc_lmex, vc_lmerx, tolerance = 4e-04), all.equal(fixef(fm1_lmer),
+ fixef(fm1_lme)), all.equal(unname(unlist(unclass(ranef(fm1_lmer)))), unname(unlist(unclass(ranef(fm1_lme)))),
+ tolerance = 2e-04))
Error: vc_lmex and vc_lmerx are not equal:
Mean relative difference: 0.8916242
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
Execution halted
Running the tests in ‘tests/lmer-0.R’ failed.
Complete output:
> require(lme4)
Loading required package: lme4
Loading required package: Matrix
> source(system.file("test-tools-1.R", package = "Matrix"))# identical3() etc
Loading required package: tools
>
> ## use old (<=3.5.2) sample() algorithm if necessary
> if ("sample.kind" %in% names(formals(RNGkind))) {
+ suppressWarnings(RNGkind("Mersenne-Twister", "Inversion", "Rounding"))
+ }
>
> ## Check that quasi families throw an error
> assertError(lmer(cbind(incidence, size - incidence) ~ period + (1|herd),
+ data = cbpp, family = quasibinomial))
> assertError(lmer(incidence ~ period + (1|herd),
+ data = cbpp, family = quasipoisson))
> assertError(lmer(incidence ~ period + (1|herd),
+ data = cbpp, family = quasi))
>
> ## check bug found by Kevin Buhr
> set.seed(7)
> n <- 10
> X <- data.frame(y=runif(n), x=rnorm(n), z=sample(c("A","B"), n, TRUE))
> fm <- lmer(log(y) ~ x | z, data=X) ## ignore grouping factors with
Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
> ## gave error inside model.frame()
> stopifnot(all.equal(c(`(Intercept)` = -0.834544), fixef(fm), tolerance=.01))
Error: c(`(Intercept)` = -0.834544) and fixef(fm) are not equal:
Mean relative difference: 0.4954795
Execution halted
Running the tests in ‘tests/minval.R’ failed.
Complete output:
> if (lme4:::testLevel() > 1 || .Platform$OS.type!="windows") {
+ ## example posted by Stéphane Laurent
+ ## exercises bug where Nelder-Mead min objective function value was >0
+ set.seed(666)
+ sims <- function(I, J, sigmab0, sigmaw0){
+ Mu <- rnorm(I, mean=0, sd=sigmab0)
+ y <- c(sapply(Mu, function(mu) rnorm(J, mu, sigmaw0)))
+ data.frame(y=y, group=gl(I,J))
+ }
+
+ I <- 3 # number of groups
+ J <- 8 # number of repeats per group
+ sigmab0 <- 0.15 # between standard deviation
+ sigmaw0 <- 0.15 # within standard deviation
+
+ dat <- sims(I, J, sigmab0, sigmaw0)
+
+ library(lme4)
+ isOldTol <- environment(nloptwrap)$defaultControl$xtol_abs == 1e-6
+
+ fm3 <- lmer(y ~ (1|group), data=dat)
+ stopifnot(all.equal(unname(unlist(VarCorr(fm3))),
+ switch(fm3@optinfo$optimizer,
+ "Nelder_Mead" = 0.029662844,
+ "bobyqa" = 0.029662698,
+ "nloptwrap" =
+ if (isOldTol) 0.029679755 else 0.029662699,
+ stop("need new case here: value is ",unname(unlist(VarCorr(fm3))))
+ ),
+ tolerance = 1e-7))
+ } ## skip on windows (for speed)
Loading required package: Matrix
Error: unname(unlist(VarCorr(fm3))) and switch(fm3@optinfo$optimizer, Nelder_Mead = 0.029662844, bobyqa = 0.029662698, .... are not equal:
Mean relative difference: 0.1265733
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 0.147111 (tol = 0.002, component 1)
Execution halted
Running the tests in ‘tests/modFormula.R’ failed.
Complete output:
> if (.Platform$OS.type != "windows") {
+ library(lme4)
+ library(testthat)
+
+ .get.checkingOpts <- lme4:::.get.checkingOpts
+ stopifnot(identical(
+ .get.checkingOpts(
+ c("CheckMe", "check.foo", "check.conv.1", "check.rankZ", "check.rankX"))
+ , c("check.foo", "check.rankZ")))
+
+ lmod <- lFormula(Reaction ~ Days + (Days|Subject), sleepstudy)
+ devfun <- do.call(mkLmerDevfun, lmod)
+ opt <- optimizeLmer(devfun)
+ cc <- lme4:::checkConv(attr(opt,"derivs"), opt$par, ctrl = lmerControl()$checkConv,
+ lbound=environment(devfun)$lower)
+ fm1 <- mkMerMod(environment(devfun), opt, lmod$reTrms, fr = lmod$fr,
+ lme4conv=cc)
+ fm2 <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy)
+
+ ## basic equivalence
+ fm1C <- fm1
+ fm1C@call <- fm2@call
+ expect_equal(fm2,fm1C)
+ expect_equal(range(residuals(fm1)), c(-101.18, 132.547), tolerance = 1e-5) # these are "outliers"!
+ expect_is(model.frame(fm1),"data.frame")
+ ## formulae
+ mfm1 <- model.frame(fm1)
+ expect_equal(formula(fm1), Reaction ~ Days + (Days | Subject))
+ expect_equal(formula(terms(mfm1)), Reaction ~ Days + (Days + Subject))
+ new_form_modframe <- (getRversion() >= "3.6.0" &&
+ as.numeric(version[["svn rev"]]) >= 75891)
+ expect_equal(formula(mfm1),
+ if(new_form_modframe) {
+ Reaction ~ Days + (Days + Subject)
+ } else
+ Reaction ~ Days + Subject
+ )
+ ## predictions
+ expect_equal(predict(fm1,newdata=sleepstudy[1:10,],re.form=NULL),
+ predict(fm2,newdata=sleepstudy[1:10,],re.form=NULL))
+ expect_equal(predict(fm1,newdata=sleepstudy),
+ predict(fm1))
+
+ lmodOff <- lFormula(Reaction ~ Days + (Days|Subject) + offset(0.5*Days),
+ sleepstudy)
+ devfunOff <- do.call(mkLmerDevfun, lmodOff)
+ opt <- optimizeLmer(devfunOff)
+ fm1Off <- mkMerMod(environment(devfunOff), opt, lmodOff$reTrms, fr = lmodOff$fr)
+ fm2Off <- lmer(Reaction ~ Days + (Days|Subject) + offset(0.5*Days), sleepstudy)
+ expect_equal(predict(fm1Off,newdata=sleepstudy[1:10,],re.form=NULL),
+ predict(fm2Off,newdata=sleepstudy[1:10,],re.form=NULL))
+
+ ## FIXME: need more torture tests with offset specified, in different environments ...
+
+ ## FIXME: drop1(.) doesn't work with modular objects ... hard to see how it
+ ## could, though ...
+ ## drop1(fm1Off)
+ drop1(fm2Off)
+
+ } ## skip on windows (for speed)
Loading required package: Matrix
Error: range(residuals(fm1)) not equal to c(-101.18, 132.547).
2/2 mismatches (average diff: 2.42)
[1] -106 - -101 == -4.708
[2] 133 - 133 == 0.129
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In lme4:::checkConv(attr(opt, "derivs"), opt$par, ctrl = lmerControl()$checkConv, :
unable to evaluate scaled gradient
3: In lme4:::checkConv(attr(opt, "derivs"), opt$par, ctrl = lmerControl()$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
4: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
5: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
6: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
Execution halted
Running the tests in ‘tests/predsim.R’ failed.
Complete output:
> ## compare range, average, etc. of simulations to
> ## conditional and unconditional prediction
> library(lme4)
Loading required package: Matrix
> do.plot <- FALSE
>
> if (.Platform$OS.type != "windows") {
+ ## use old (<=3.5.2) sample() algorithm if necessary
+ if ("sample.kind" %in% names(formals(RNGkind))) {
+ suppressWarnings(RNGkind("Mersenne-Twister", "Inversion", "Rounding"))
+ }
+
+ fm1 <- lmer(Reaction~Days+(1|Subject),sleepstudy)
+ set.seed(101)
+ pp <- predict(fm1)
+ rr <- range(usim2 <- simulate(fm1,1,use.u=TRUE)[[1]])
+ stopifnot(all.equal(rr,c(159.3896,439.1616),tolerance=1e-6))
+ if (do.plot) {
+ plot(pp,ylim=rr)
+ lines(sleepstudy$Reaction)
+ points(simulate(fm1,1)[[1]],col=4)
+ points(usim2,col=2)
+ }
+
+ set.seed(101)
+
+ ## conditional prediction
+ ss <- simulate(fm1,1000,use.u=TRUE)
+ ss_sum <- t(apply(ss,1,quantile,c(0.025,0.5,0.975)))
+ plot(pp)
+ matlines(ss_sum,col=c(1,2,1),lty=c(2,1,2))
+ stopifnot(all.equal(ss_sum[,2],pp,tolerance=5e-3))
+
+ ## population-level prediction
+ pp2 <- predict(fm1, re.form=NA)
+ ss2 <- simulate(fm1,1000,use.u=FALSE)
+ ss_sum2 <- t(apply(ss2,1,quantile,c(0.025,0.5,0.975)))
+
+ if (do.plot) {
+ plot(pp2,ylim=c(200,400))
+ matlines(ss_sum2,col=c(1,2,1),lty=c(2,1,2))
+ }
+
+ stopifnot(all.equal(ss_sum2[,2],pp2,tolerance=8e-3))
+
+ ## predict(...,newdata=...) on models with derived variables in the random effects
+ ## e.g. (f:g, f/g)
+ set.seed(101)
+ d <- expand.grid(f=factor(letters[1:10]),g=factor(letters[1:10]),
+ rep=1:10)
+ d$y <- rnorm(nrow(d))
+ m1 <- lmer(y~(1|f:g),d)
+ p1A <- predict(m1)
+ p1B <- predict(m1,newdata=d)
+ stopifnot(all.equal(p1A,p1B))
+ m2 <- lmer(y~(1|f/g),d)
+ p2A <- predict(m2)
+ p2B <- predict(m2,newdata=d)
+ stopifnot(all.equal(p2A,p2B))
+
+ ## with numeric grouping variables
+ dn <- transform(d,f=as.numeric(f),g=as.numeric(g))
+ m1N <- update(m1,data=dn)
+ p1NA <- predict(m1N)
+ p1NB <- predict(m1N,newdata=dn)
+ stopifnot(all.equal(p1NA,p1NB))
+
+ ## simulate with modified parameters
+ set.seed(1)
+ s1 <- simulate(fm1)
+ set.seed(1)
+ s2 <- simulate(fm1,newdata=model.frame(fm1),
+ newparams=getME(fm1,c("theta","beta","sigma")))
+ all.equal(s1,s2)
+
+ fm0 <- update(fm1,.~.-Days)
+ ##
+ ## sim() -> simulate() -> refit() -> deviance
+ ##
+
+ ## predictions and simulations with offsets
+
+ set.seed(101)
+ d <- data.frame(y=rpois(100,5),x=rlnorm(100,1,1),
+ f=factor(sample(10,size=100,replace=TRUE)))
+ gm1 <- glmer(y~offset(log(x))+(1|f),data=d,
+ family=poisson)
+ s1 <- simulate(gm1)
+ } ## skip on windows (for speed)
Error: rr and c(159.3896, 439.1616) are not equal:
Mean relative difference: 0.007853237
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 1.97927 (tol = 0.002, component 1)
Execution halted
Running the tests in ‘tests/priorWeights.R’ failed.
Complete output:
> ## use old (<=3.5.2) sample() algorithm if necessary
> if ("sample.kind" %in% names(formals(RNGkind))) {
+ suppressWarnings(RNGkind("Mersenne-Twister", "Inversion", "Rounding"))
+ }
>
>
> compFunc <- function(lmeMod, lmerMod, tol = 1e-2){
+ lmeVarCorr <- nlme:::VarCorr(lmeMod)[,"StdDev"]
+ lmeCoef <- summary(lmeMod)$tTable[,-c(3,5)]
+ lmeOut <- c(as.numeric(lmeVarCorr), as.numeric(lmeCoef))
+ keep <- !is.na(lmeOut)
+ lmeOut <- lmeOut[keep]
+ dn <- dimnames(lmeCoef)
+ if(is.null(dn)) dn <- list("", names(lmeCoef))
+ names(lmeOut) <- c(
+ paste(names(lmeVarCorr), "Var"),
+ as.character(do.call(outer, c(dn, list("paste")))))[keep]
+
+ ## get nested RE variances in the same order as nlme
+ ## FIXME: not sure if this works generally
+ vcLmer <- VarCorr(lmerMod)
+ vcLmer <- vcLmer[length(vcLmer):1]
+ ##
+
+ lmerVarCorr <- c(sapply(vcLmer, attr, "stddev"),
+ attr(VarCorr(lmerMod), "sc"))
+ ## differentiate lme4{new} and lme4.0 :
+ lmerCoef <- if(is(lmerMod, "merMod"))
+ summary(lmerMod)$coefficients else summary(lmerMod)@coefs
+ lmerOut <- c(lmerVarCorr, as.numeric(lmerCoef))
+ names(lmerOut) <- names(lmeOut)
+
+ return(list(target = lmeOut, current = lmerOut, tolerance = tol))
+ }
>
> if (.Platform$OS.type != "windows") {
+ set.seed(1)
+ nGroups <- 100
+ nObs <- 1000
+
+ # explanatory variable with a fixed effect
+ explVar1 <- rnorm(nObs)
+ explVar2 <- rnorm(nObs)
+
+ # random intercept among levels of a grouping factor
+ groupFac <- as.factor(rep(1:nGroups,each=nObs/nGroups))
+ randEff0 <- rep(rnorm(nGroups),each=nObs/nGroups)
+ randEff1 <- rep(rnorm(nGroups),each=nObs/nGroups)
+ randEff2 <- rep(rnorm(nGroups),each=nObs/nGroups)
+
+ # residuals with heterogeneous variance
+ residSD <- rpois(nObs,1) + 1
+ residError <- rnorm(nObs,sd=residSD)
+
+ # response variable
+ respVar <- randEff0 + (1+randEff1)*explVar1 + (1+randEff2)*explVar2 + residError
+
+ # rename to fit models on one line
+ y <- respVar
+ x <- explVar1
+ z <- explVar2
+ g <- groupFac
+ v <- residSD^2
+ w <- 1/v
+
+ library("nlme")
+ lmeMods <- list(
+ ML1 = lme(y ~ x, random = ~ 1|g, weights = varFixed(~v), method = "ML"),
+ REML1 = lme(y ~ x, random = ~ 1|g, weights = varFixed(~v), method = "REML"),
+ ML2 = lme(y ~ x, random = ~ x|g, weights = varFixed(~v), method = "ML"),
+ REML2 = lme(y ~ x, random = ~ x|g, weights = varFixed(~v), method = "REML"),
+ ML1 = lme(y ~ x+z, random = ~ x+z|g, weights = varFixed(~v), method = "ML"),
+ REML2 = lme(y ~ x+z, random = ~ x+z|g, weights = varFixed(~v), method = "REML"))
+
+ library("lme4")
+ lmerMods <- list(
+ ML1 = lmer(y ~ x + (1|g), weights = w, REML = FALSE),
+ REML1 = lmer(y ~ x + (1|g), weights = w, REML = TRUE),
+ ML2 = lmer(y ~ x + (x|g), weights = w, REML = FALSE),
+ REML2 = lmer(y ~ x + (x|g), weights = w, REML = TRUE),
+ ML3 = lmer(y ~ x + z + (x+z|g), weights = w, REML = FALSE),
+ REML3 = lmer(y ~ x + z + (x+z|g), weights = w, REML = TRUE))
+
+ comp <- mapply(compFunc, lmeMods, lmerMods, SIMPLIFY=FALSE)
+ stopifnot(all(sapply(comp, do.call, what = all.equal)))
+ ## Look at the relative differences:
+ sapply(mapply(compFunc, lmeMods, lmerMods, SIMPLIFY=FALSE, tol = 0),
+ do.call, what = all.equal)
+
+ ## add simulated weights to the sleepstudy example
+ n <- nrow(sleepstudy)
+ v <- rpois(n,1) + 1
+ w <- 1/v
+ sleepLme <- lme(Reaction ~ Days, random = ~ Days|Subject,
+ sleepstudy, weights = varFixed(~v),
+ method = "ML")
+ sleepLmer <- lmer(Reaction ~ Days + (Days|Subject),
+ sleepstudy, weights = w,
+ REML = FALSE)
+ sleepComp <- compFunc(sleepLme, sleepLmer)
+ stopifnot(do.call(all.equal, sleepComp))
+ ## look at relative differences:
+ sleepComp$tolerance <- 0
+ do.call(all.equal, sleepComp)
+
+ if (require("mlmRev")) {
+ n <- nrow(Chem97)
+ v <- rpois(n,1) + 1
+ w <- 1/v
+ Chem97Lme <- lme(score ~ 1, random = ~ 1|lea/school, Chem97)
+ Chem97Lmer <- lmer(score ~ (1|lea/school), Chem97)
+ Chem97Comp <- compFunc(Chem97Lme, Chem97Lmer)
+ stopifnot(do.call(all.equal, Chem97Comp))
+ ## look at relative differences:
+ Chem97Comp$tolerance <- 0
+ do.call(all.equal, Chem97Comp)
+ }
+
+ set.seed(2)
+ n <- 40
+ w <- runif(n)
+ x <- runif(n)
+ g <- factor(sample(1:10,n,replace=TRUE))
+ Z <- model.matrix(~g-1);
+ y <- Z%*%rnorm(ncol(Z)) + x + rnorm(n)/w^.5
+ m <- lmer(y ~ x + (1|g), weights=w, REML = TRUE)
+
+ ## CRAN-forbidden:
+ ## has4.0 <- require("lme4.0"))
+ has4.0 <- FALSE
+ if(has4.0) {
+ ## m.0 <- lme4.0::lmer(y ~ x + (1|g), weights=w, REML = TRUE)
+ lmer0 <- get("lmer", envir=asNamespace("lme4.0"))
+ m.0 <- lmer0(y ~ x + (1|g), weights=w, REML = TRUE)
+ dput(fixef(m.0)) # c(-0.73065400610675, 2.02895402562926)
+ dput(sigma(m.0)) # 1.73614301673377
+ dput(VarCorr(m.0)$g[1,1]) # 2.35670451590395
+ dput(unname(coef(summary(m.0))[,"Std. Error"]))
+ ## c(0.95070076853232, 1.37650858268602)
+ }
+ fixef_lme4.0 <- c(-0.7306540061, 2.0289540256)
+ sigma_lme4.0 <- 1.7361430
+ Sigma_lme4.0 <- 2.3567045
+ SE_lme4.0 <- c(0.95070077, 1.37650858)
+ if(has4.0) try(detach("package:lme4.0"))
+
+ stopifnot(all.equal(unname(fixef(m)), fixef_lme4.0, tolerance = 1e-3))
+ all.equal(unname(fixef(m)), fixef_lme4.0, tolerance = 0) #-> 1.657e-5
+
+ ## but these are not at all equal :
+ (all.equal(sigma(m), sigma_lme4.0, tolerance = 10^-3)) # 0.4276
+ (all.equal(as.vector(VarCorr(m)$g), Sigma_lme4.0, tolerance = 10^-3)) # 1.038
+ (all.equal(as.vector(summary(m)$coefficients[,2]), SE_lme4.0, tolerance = 10^-3)) # 0.4276
+ ## so, lme4.0 was clearly wrong here
+
+
+ ##' make sure models that differ only in a constant
+ ##' prior weight have identical deviance:
+ fm <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy,REML=FALSE)
+ fm_wt <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy, weights = rep(5, nrow(sleepstudy)),REML=FALSE)
+ all.equal(deviance(fm), deviance(fm_wt))
+ } ## skip on windows (for speed)
Loading required package: Matrix
Attaching package: 'lme4'
The following object is masked from 'package:nlme':
lmList
Error: all(sapply(comp, do.call, what = all.equal)) is not TRUE
In addition: There were 13 warnings (use warnings() to see them)
Execution halted
Running the tests in ‘tests/priorWeightsModComp.R’ failed.
Complete output:
> library(lme4)
Loading required package: Matrix
> n <- nrow(sleepstudy)
> op <- options(warn = 1, # show as they happen ("false" convergence warnings)
+ useFancyQuotes = FALSE)
>
> if (.Platform$OS.type != "windows") {
+ ##' remove all attributes but names
+ dropA <- function(x) `attributes<-`(x, list(names = names(x)))
+ ##' transform result of "numeric" all.equal.list() to a named vector
+ all.eqL <- function(x1, x2, ...) {
+ r <- sub("^Component ", '', all.equal(x1, x2, tolerance = 0, ...))
+ r <- strsplit(sub(": Mean relative difference:", "&&", r),
+ split="&&", fixed=TRUE)
+ setNames(as.numeric(vapply(r, `[`, "1.234", 2L)),
+ ## drop surrounding "..."
+ nm = sub('"$', '', substring(vapply(r, `[`, "nam", 1L), first=2)))
+ }
+ seedF <- function(s) {
+ if(s %in% c(6, 39, 52, 57, 63, 74, 76, 86))
+ switch(as.character(s)
+ , "52"=, "63"=, "74" = 2
+ , "6"=, "39" = 3
+ , "86" = 8 # needs 4 on Lnx-64b
+ , "76" = 70 # needs 42 on Lnx-64b
+ , "57" = 90 # needs 52 on Lnx-64b
+ )
+ else if(s %in% c(1, 12, 15, 34, 36, 41, 42, 43, 49, 55, 59, 67, 80, 85)) ## seeds 41,59, .. 15
+ 1.0
+ else ## seeds 22, 20, and better
+ 0.25
+ }
+ ## be fast, running only 10 seeds by default:
+ sMax <- if(lme4:::testLevel() > 1) 99L else 9L
+ mySeeds <- 0L:sMax
+
+ lapply(setNames(,mySeeds), function(seed) {
+ cat("\n------ random seed =", seed, "---------\n")
+ set.seed(seed)
+ v <- rpois(n,1) + 1
+ w <- 1/v
+ cat("weights w:\n")
+ fm1 <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy, REML=FALSE, weights = w); cat("..2:\n")
+ fm2 <- lmer(Reaction ~ Days + (1 | Subject), sleepstudy, REML=FALSE, weights = w)
+ cat("weights w*10:\n")
+ fm1.10 <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy, REML=FALSE, weights = w*10);cat("..2:\n")
+ fm2.10 <- lmer(Reaction ~ Days + (1 | Subject), sleepstudy, REML=FALSE, weights = w*10)
+ ##
+ ano12... <- dropA(anova(fm1, fm2 ))
+ ano12.10 <- dropA(anova(fm1.10, fm2.10))
+ print(aEQ <- all.eqL(ano12..., ano12.10)) # showing differences
+ if(!exists("notChisq"))
+ notChisq <<-
+ local({ n <- names(ano12...)
+ grep("Chisq", n, value=TRUE, fixed=TRUE, invert=TRUE) })
+ stopifnot(
+ all.equal(ano12...$Chisq,
+ ano12.10$Chisq, tol = 1e-6 * seedF(seed))
+ ,
+ all.equal(ano12...[notChisq],
+ ano12.10[notChisq], tol= 1.5e-8 * seedF(seed))
+ )
+ aEQ
+ }) -> rallEQ
+
+ cat("=====================================\n")
+
+ rallEQ <- t(simplify2array(rallEQ))
+ notChisq <- intersect(notChisq, colnames(rallEQ))
+ ## sort according to "severity":
+ srallEQ <- rallEQ[with(as.data.frame(rallEQ), order(AIC, Chisq)), ]
+ round(log10(srallEQ), 2)
+ saveRDS(srallEQ, "priorWeightsMod_relerr.rds")
+
+ if(!dev.interactive(orNone=TRUE)) pdf("priorWeightsMod_relerr.pdf")
+
+ matplot(mySeeds, log10(srallEQ), type="l", xlab=NA) ; grid()
+ legend("topleft", ncol=3, bty="n",
+ paste(1:6, colnames(srallEQ), sep = ": "), col=1:6, lty=1:6)
+ tolD <- sqrt(.Machine$double.eps) # sqrt(eps_C)
+ abline(h = log10(tolD), col = "forest green", lty=3)
+ axis(4, at=log10(tolD), label=quote(sqrt(epsilon[c])), las=1)
+ LRG <- which(srallEQ[,"AIC"] > tolD)
+ if (length(LRG)>0) {
+ text(LRG, log10(srallEQ[LRG, "AIC"]), names(LRG), cex = .8)
+ }
+
+ ## how close are we ..
+ str(tF <- sapply(mySeeds, seedF))
+ round(sort( rallEQ[, "Chisq"] / (tF * 1e-6 ), decreasing=TRUE), 1)
+ round(sort(apply(rallEQ[,notChisq] / (tF * 1.5e-8), 1, max), decreasing=TRUE), 1)
+ } ## skip on windows (for speed)
------ random seed = 0 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.12755 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 1 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.00449 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 2 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.68252 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 26.1456 (tol = 0.002, component 1)
RUE
NA
------ random seed = 3 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.11241 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 4 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.41542 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 5 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.86338 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 6 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.58993 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 7 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.57508 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 8 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.62162 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 9 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.73043 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
=====================================
Error in eval(substitute(expr), data, enclos = parent.frame()) :
object 'Chisq' not found
Calls: with -> with.default -> eval -> eval -> order
Execution halted
Running the tests in ‘tests/vcov-etc.R’ failed.
Complete output:
> stopifnot(require(lme4))
Loading required package: lme4
Loading required package: Matrix
>
> (testLevel <- lme4:::testLevel())
[1] 1
> source(system.file("testdata", "lme-tst-funs.R", package="lme4", mustWork=TRUE))# -> unn()
>
>
> ## "MEMSS" is just 'Suggest' -- must still work, when it's missing:
> if (suppressWarnings(!require(MEMSS, quietly=TRUE)) ||
+ (data(ergoStool, package="MEMSS") != "ergoStool")) {
+
+ cat("'ergoStool' data from package 'MEMSS' is not available --> skipping test\n")
+ } else {
+
+ fm1 <- lmer (effort ~ Type + (1|Subject), data = ergoStool)
+ ##sp no longer supported since ~ 2012-3:
+ ##sp fm1.s <- lmer (effort ~ Type + (1|Subject), data = ergoStool, sparseX=TRUE)
+ ## was segfaulting with sparseX (a while upto 2010-04-06)
+
+ fe1 <- fixef(fm1)
+ ##sp fe1.s <- fixef(fm1.s)
+
+ print(s1.d <- summary(fm1))
+ ##sp print(s1.s <- summary(fm1.s))
+ Tse1.d <- c(0.57601226, rep(0.51868384, 3))
+ stopifnot(exprs = {
+ ##sp all.equal(fe1, fe1.s, tolerance= 1e-12)
+ all.equal(Tse1.d, unname(se1.d <- coef(s1.d)[,"Std. Error"]),
+ tolerance = 1e-6) # std.err.: no too much accuracy
+ is(V.d <- vcov(fm1), "symmetricMatrix")
+ ##sp all.equal(se1.d, coef(s1.s)[,"Std. Error"])#, tol = 1e-10
+ ##sp all.equal( V.d, vcov(fm1.s))#, tol = 1e-9
+ all.equal(Matrix::diag(V.d), unn(se1.d)^2, tolerance= 1e-12)
+ })
+
+ }## if( ergoStool is available from pkg MEMSS )
Attaching package: 'MEMSS'
The following objects are masked from 'package:datasets':
CO2, Orange, Theoph
Linear mixed model fit by REML ['lmerMod']
Formula: effort ~ Type + (1 | Subject)
Data: ergoStool
REML criterion at convergence: Inf
Scaled residuals:
Min 1Q Median 3Q Max
-1.55001 -0.64996 -0.04112 0.60381 1.64861
Random effects:
Groups Name Variance Std.Dev.
Subject (Intercept) 0.915 0.9565
Residual 1.517 1.2316
Number of obs: 36, groups: Subject, 9
Fixed effects:
Estimate Std. Error t value
(Intercept) 8.5556 0.5198 16.459
TypeT2 3.8889 0.5806 6.698
TypeT3 2.2222 0.5806 3.827
TypeT4 0.6667 0.5806 1.148
Correlation of Fixed Effects:
(Intr) TypeT2 TypeT3
TypeT2 -0.558
TypeT3 -0.558 0.500
TypeT4 -0.558 0.500 0.500
optimizer (nloptwrap) convergence code: -2 (NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).)
Model failed to converge with max|grad| = 1.60567 (tol = 0.002, component 1)
Error: Tse1.d and unname(se1.d <- coef(s1.d)[, "Std. Error"]) are not equal:
Mean relative difference: 0.1134759
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 1.60567 (tol = 0.002, component 1)
Execution halted
Flavor: r-devel-linux-x86_64-fedora-clang
Version: 1.1-36
Check: re-building of vignette outputs
Result: ERROR
Error(s) in re-building vignettes:
--- re-building ‘lmerperf.Rmd’ using rmarkdown
--- finished re-building ‘lmerperf.Rmd’
--- re-building ‘Theory.Rnw’ using knitr
--- finished re-building ‘Theory.Rnw’
--- re-building ‘lmer.Rnw’ using knitr
Quitting from lines 2660-2661 [profile_zeta_plot] (lmer.Rnw)
Error: processing vignette 'lmer.Rnw' failed with diagnostics:
arguments imply differing number of rows: 18, 0
--- failed re-building ‘lmer.Rnw’
--- re-building ‘PLSvGLS.Rnw’ using Sweave
--- finished re-building ‘PLSvGLS.Rnw’
SUMMARY: processing the following file failed:
‘lmer.Rnw’
Error: Vignette re-building failed.
Execution halted
Flavors: r-devel-linux-x86_64-fedora-clang, r-devel-linux-x86_64-fedora-gcc
Version: 1.1-36
Check: examples
Result: ERROR
Running examples in ‘lme4-Ex.R’ failed
The error most likely occurred in:
> ### Name: profile-methods
> ### Title: Profile method for merMod objects
> ### Aliases: as.data.frame.thpr log.thpr logProf varianceProf
> ### profile-methods profile.merMod
> ### Keywords: methods
>
> ### ** Examples
>
> fm01ML <- lmer(Yield ~ 1|Batch, Dyestuff, REML = FALSE)
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 1.01096 (tol = 0.002, component 1)
> system.time(
+ tpr <- profile(fm01ML, optimizer="Nelder_Mead", which="beta_")
+ )## fast; as only *one* beta parameter is profiled over -> 0.09s (2022)
Error in profile.merMod(fm01ML, optimizer = "Nelder_Mead", which = "beta_") :
Profiling over both the residual variance and
fixed effects is not numerically consistent with
profiling over the fixed effects only (relative difference: 1);
consider adjusting devmatchtol
Calls: system.time -> profile -> profile.merMod
Timing stopped at: 0.037 0 0.045
Execution halted
Flavor: r-devel-linux-x86_64-fedora-gcc
Version: 1.1-36
Check: tests
Result: ERROR
Running ‘AAAtest-all.R’ [43s/80s]
Running ‘HSAURtrees.R’
Running ‘REMLdev.R’
Running ‘ST.R’
Running ‘agridat_gotway.R’ [6s/12s]
Running ‘bootMer.R’ [12s/35s]
Running ‘boundary.R’ [7s/19s]
Running ‘confint.R’ [8s/20s]
Running ‘devCritFun.R’
Running ‘drop.R’
Running ‘drop1contrasts.R’ [4s/12s]
Running ‘dynload.R’ [15s/34s]
Running ‘elston.R’ [5s/12s]
Running ‘evalCall.R’ [4s/11s]
Running ‘extras.R’ [4s/10s]
Running ‘falsezero_dorie.R’ [4s/11s]
Running ‘fewlevels.R’
Running ‘getME.R’
Running ‘glmer-1.R’ [7s/17s]
Running ‘glmerControlPass.R’ [11s/28s]
Running ‘glmerWarn.R’ [7s/20s]
Running ‘glmmExt.R’ [13s/31s]
Running ‘glmmWeights.R’ [13s/30s]
Running ‘hatvalues.R’ [4s/11s]
Running ‘is.R’
Running ‘lmList-tst.R’ [4s/11s]
Running ‘lme4_nlme.R’ [4s/11s]
Running ‘lmer-0.R’ [4s/11s]
Running ‘lmer-1.R’
Running ‘lmer-conv.R’ [4s/10s]
Running ‘lmer2_ex.R’ [4s/11s]
Running ‘methods.R’ [7s/16s]
Running ‘minval.R’ [4s/10s]
Running ‘modFormula.R’ [6s/16s]
Running ‘nbinom.R’
Running ‘nlmer-conv.R’
Running ‘nlmer.R’
Running ‘offset.R’ [6s/12s]
Running ‘optimizer.R’ [7s/17s]
Running ‘polytomous.R’ [4s/11s]
Running ‘prLogistic.R’
Running ‘predict_basis.R’ [5s/11s]
Running ‘predsim.R’
Running ‘priorWeights.R’ [6s/11s]
Running ‘priorWeightsModComp.R’ [6s/11s]
Running ‘profile-tst.R’
Running ‘refit.R’
Running ‘resids.R’
Running ‘respiratory.R’ [12s/21s]
Running ‘simulate.R’
Running ‘test-glmernbref.R’ [6s/13s]
Running ‘testOptControl.R’
Running ‘testcolonizer.R’ [5s/10s]
Running ‘testcrab.R’ [15s/28s]
Running ‘throw.R’
Running ‘varcorr.R’
Running ‘vcov-etc.R’
Running the tests in ‘tests/AAAtest-all.R’ failed.
Complete output:
> if (base::require("testthat", quietly = TRUE)) {
+ pkg <- "lme4"
+ require(pkg, character.only=TRUE, quietly=TRUE)
+ if(getRversion() < "3.5.0") { withAutoprint <- identity ; prt <- print } else { prt <- identity }
+ if(Sys.getenv("USER") %in% c("maechler", "bbolker")) withAutoprint({
+ ## for developers' sake:
+ lP <- .libPaths() # ---- .libPaths() : ----
+ prt(lP)
+ ## ---- Entries in .libPaths()[1] : ----
+ prt(list.files(lP[1], include.dirs=TRUE))
+ prt(sessionInfo())
+ prt(packageDescription("Matrix"))
+ ## 'lme4' from packageDescription "file" :
+ prt(attr(packageDescription("lme4"), "file"))
+ })
+ test_check(pkg)
+ ##======== ^^^
+ print(warnings()) # TODO? catch most of these by expect_warning(..)
+ } else {
+ cat( "package 'testthat' not available, cannot run unit tests\n" )
+ }
[ FAIL 24 | WARN 80 | SKIP 1 | PASS 455 ]
══ Skipped tests (1) ═══════════════════════════════════════════════════════════
• On CRAN (1): 'test-eval.R:2:1'
══ Failed tests ════════════════════════════════════════════════════════════════
── Failure ('test-lmer.R:38:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] 2265 - 1764 == 501
Backtrace:
▆
1. └─testthat::expect_that(...) at test-lmer.R:38:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Failure ('test-lmer.R:49:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] Inf - 320 == Inf
Backtrace:
▆
1. └─testthat::expect_that(REMLcrit(fm1), equals(319.654276842342)) at test-lmer.R:49:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Failure ('test-lmer.R:52:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] 48.6 - 49.5 == -0.886
Backtrace:
▆
1. └─testthat::expect_that(sigma(fm1), equals(49.5101272946856, tolerance = 1e-06)) at test-lmer.R:52:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Failure ('test-lmer.R:58:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] 456 - 376 == 80.6
Backtrace:
▆
1. └─testthat::expect_that(...) at test-lmer.R:58:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Failure ('test-lmer.R:64:5'): lmer ──────────────────────────────────────────
`x` not equivalent to `expected`.
1/1 mismatches
[1] 12.1 - 0 == 12.1
Backtrace:
▆
1. └─testthat::expect_that(VarCorr(fm2)[[1]][1, 1], is_equivalent_to(0)) at test-lmer.R:64:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equivalent(x, expected, expected.label = label)
── Failure ('test-lmer.R:65:5'): lmer ──────────────────────────────────────────
`x` not equivalent to `expected`.
1/1 mismatches
[1] 0.979 - 0 == 0.979
Backtrace:
▆
1. └─testthat::expect_that(getME(fm2, "theta"), is_equivalent_to(0)) at test-lmer.R:65:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equivalent(x, expected, expected.label = label)
── Failure ('test-lmer.R:74:5'): lmer ──────────────────────────────────────────
`x` not equal to `expected`.
1/1 mismatches
[1] 0.979 - 0.848 == 0.13
Backtrace:
▆
1. └─testthat::expect_that(...) at test-lmer.R:74:5
2. └─testthat (local) condition(object)
3. └─testthat::expect_equal(x, expected, ..., expected.label = label)
── Error ('test-lmer.R:117:5'): lmer ───────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. ├─testthat::expect_is(...) at test-lmer.R:117:5
2. │ └─testthat::quasi_label(enquo(object), label, arg = "object")
3. │ └─rlang::eval_bare(expr, quo_get_env(quo))
4. └─lme4::lmer(Yield ~ 1 | Batch, Dyestuff, REML = TRUE)
5. └─lme4::optimizeLmer(...)
6. └─lme4:::optwrap(...)
── Error ('test-lmer.R:286:5'): coef_lmer ──────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(resp ~ 0 + var1 + var1:var2 + (1 | var3), data = d) at test-lmer.R:286:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-lmer.R:335:1'): (code run outside of `test_that()`) ────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Reaction ~ Days + (Days | Subject), sleepstudy) at test-lmer.R:335:1
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-methods.R:45:1'): (code run outside of `test_that()`) ──────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(...) at test-methods.R:45:1
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:422:3'): prediction with . in formula + newdata ──────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(dv ~ . - groups + (1 | groups), data = train) at test-predict.R:422:3
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:445:3'): prediction standard error ───────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Petal.Width ~ Sepal.Length + (1 | Species), iris) at test-predict.R:445:3
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:478:5'): NA + re.form = NULL + simulate OK (GH #737) ──
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Reaction ~ Days + (Days | Subject), d) at test-predict.R:478:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:501:5'): predict works with factors in left-out REs ──
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(yield ~ 1 + (1 | g1) + (lc | g3), data = df2, control = lmerControl(check.conv.singular = "ignore")) at test-predict.R:501:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-predict.R:514:5'): predict works with dummy() in left-out REs ──
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(...) at test-predict.R:514:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-ranef.R:34:5'): Dyestuff consistent with lme4.0 ────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Yield ~ 1 | Batch, Dyestuff, REML = FALSE) at test-ranef.R:34:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-ranef.R:43:9'): sleepstudy consistent with lme4.0 ──────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Reaction ~ Days + (Days | Subject), sleepstudy) at test-ranef.R:43:9
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-ranef.R:61:5'): multiple terms work ────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(...) at test-ranef.R:61:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Error ('test-rank.R:14:5'): lmerRank ────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. ├─testthat::expect_message(...) at test-rank.R:14:5
2. │ └─testthat:::quasi_capture(enquo(object), label, capture_messages)
3. │ ├─testthat (local) .capture(...)
4. │ │ └─base::withCallingHandlers(...)
5. │ └─rlang::eval_bare(quo_get_expr(.quo), quo_get_env(.quo))
6. └─lme4::lmer(z ~ x + y + (1 | r), data = d)
7. └─lme4::optimizeLmer(...)
8. └─lme4:::optwrap(...)
── Error ('test-rank.R:101:5'): ranksim ────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. ├─base::suppressMessages(lmer(y ~ x1 + x2 + (1 | id), data = x)) at test-rank.R:101:5
2. │ └─base::withCallingHandlers(...)
3. └─lme4::lmer(y ~ x1 + x2 + (1 | id), data = x)
4. └─lme4::optimizeLmer(...)
5. └─lme4:::optwrap(...)
── Error ('test-resids.R:7:5'): lmer ───────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(Reaction ~ Days + (Days | Subject), sleepstudy, control = C1) at test-resids.R:7:5
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
── Failure ('test-start.R:39:9'): lmer ─────────────────────────────────────────
AIC(x) not equal to 1763.939344.
1/1 mismatches
[1] Inf - 1764 == Inf
── Error ('test-summary.R:32:3'): lmer ─────────────────────────────────────────
Error in `optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower,
control = control, adj = FALSE, verbose = verbose, ...)`: (converted from warning) convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Backtrace:
▆
1. └─lme4::lmer(y ~ x.1 + x.2 + (1 + x.1 | g), control = C1) at test-summary.R:32:3
2. └─lme4::optimizeLmer(...)
3. └─lme4:::optwrap(...)
[ FAIL 24 | WARN 80 | SKIP 1 | PASS 455 ]
Error: Test failures
Execution halted
Running the tests in ‘tests/REMLdev.R’ failed.
Complete output:
> library(lme4)
Loading required package: Matrix
> ## show important current settings {for reference, etc} -- [early, and also on Windows !]:
> str( lmerControl())
List of 8
$ optimizer : chr "nloptwrap"
$ restart_edge : logi TRUE
$ boundary.tol : num 1e-05
$ calc.derivs : logi TRUE
$ use.last.params: logi FALSE
$ checkControl :List of 8
..$ check.nobs.vs.rankZ: chr "ignore"
..$ check.nobs.vs.nlev : chr "stop"
..$ check.nlev.gtreq.5 : chr "ignore"
..$ check.nlev.gtr.1 : chr "stop"
..$ check.nobs.vs.nRE : chr "stop"
..$ check.rankX : chr "message+drop.cols"
..$ check.scaleX : chr "warning"
..$ check.formula.LHS : chr "stop"
$ checkConv :List of 3
..$ check.conv.grad :List of 3
.. ..$ action: chr "warning"
.. ..$ tol : num 0.002
.. ..$ relTol: NULL
..$ check.conv.singular:List of 2
.. ..$ action: chr "message"
.. ..$ tol : num 1e-04
..$ check.conv.hess :List of 2
.. ..$ action: chr "warning"
.. ..$ tol : num 1e-06
$ optCtrl : list()
- attr(*, "class")= chr [1:2] "lmerControl" "merControl"
> str(glmerControl())
List of 11
$ optimizer : chr [1:2] "bobyqa" "Nelder_Mead"
$ restart_edge : logi FALSE
$ boundary.tol : num 1e-05
$ calc.derivs : logi TRUE
$ use.last.params: logi FALSE
$ checkControl :List of 9
..$ check.nobs.vs.rankZ : chr "ignore"
..$ check.nobs.vs.nlev : chr "stop"
..$ check.nlev.gtreq.5 : chr "ignore"
..$ check.nlev.gtr.1 : chr "stop"
..$ check.nobs.vs.nRE : chr "stop"
..$ check.rankX : chr "message+drop.cols"
..$ check.scaleX : chr "warning"
..$ check.formula.LHS : chr "stop"
..$ check.response.not.const: chr "stop"
$ checkConv :List of 3
..$ check.conv.grad :List of 3
.. ..$ action: chr "warning"
.. ..$ tol : num 0.002
.. ..$ relTol: NULL
..$ check.conv.singular:List of 2
.. ..$ action: chr "message"
.. ..$ tol : num 1e-04
..$ check.conv.hess :List of 2
.. ..$ action: chr "warning"
.. ..$ tol : num 1e-06
$ optCtrl : list()
$ tolPwrss : num 1e-07
$ compDev : logi TRUE
$ nAGQ0initStep : logi TRUE
- attr(*, "class")= chr [1:2] "glmerControl" "merControl"
> str(nlmerControl())
List of 3
$ optimizer: chr [1:2] "Nelder_Mead" "Nelder_Mead"
$ tolPwrss : num 1e-10
$ optCtrl : list()
- attr(*, "class")= chr [1:2] "nlmerControl" "merControl"
> ls.str(environment(nloptwrap))
defaultControl : List of 4
$ algorithm: chr "NLOPT_LN_BOBYQA"
$ xtol_abs : num 1e-08
$ ftol_abs : num 1e-08
$ maxeval : num 1e+05
> ##
> fm1 <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy)
Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
> fm1ML <- refitML(fm1)
> REMLcrit(fm1)
[1] Inf
> deviance(fm1ML)
[1] 1751.939
> deviance(fm1,REML=FALSE) ## FIXME: not working yet (NA)
[1] 1784.642
> deviance(fm1,REML=TRUE)
[1] 1784.642
>
> ## from lme4.0
> oldvals <- c(REML=1743.6282722424, ML=1751.98581103058)
> ## leave out ML values for REML fits for now ...
> stopifnot(
+ all.equal(REMLcrit(fm1),deviance(fm1,REML=TRUE),deviance(fm1ML,REML=TRUE),oldvals["REML"]),
+ all.equal(deviance(fm1ML),deviance(fm1ML,REML=FALSE),oldvals["ML"]),
+ all.equal(REMLcrit(fm1)/-2,c(logLik(fm1)),c(logLik(fm1ML,REML=TRUE)),c(logLik(fm1,REML=TRUE))),
+ all.equal(deviance(fm1ML)/-2,c(logLik(fm1ML,REML=FALSE)),
+ c(logLik(fm1ML,REML=FALSE))))
Error: REMLcrit(fm1) and deviance(fm1, REML = TRUE) are not equal:
Mean scaled difference: Inf
Execution halted
Running the tests in ‘tests/boundary.R’ failed.
Complete output:
> ## In both of these cases boundary fit (i.e. estimate of zero RE
> ## variance) is *incorrect*. (Nelder_Mead, restart_edge=FALSE) is the
> ## only case where we get stuck; either optimizer=bobyqa or
> ## restart_edge=TRUE (default) works
> if (.Platform$OS.type != "windows") {
+
+ library(lme4)
+ library(testthat)
+
+ if(!dev.interactive(orNone=TRUE)) pdf("boundary_plots.pdf")
+
+ ## Stephane Laurent:
+ dat <- read.csv(system.file("testdata","dat20101314.csv", package="lme4"))
+
+ fit <- lmer(y ~ (1|Operator)+(1|Part)+(1|Part:Operator), data=dat,
+ control= lmerControl(optimizer="Nelder_Mead"))
+ fit_b <- lmer(y ~ (1|Operator)+(1|Part)+(1|Part:Operator), data=dat,
+ control= lmerControl(optimizer="bobyqa", restart_edge=FALSE))
+ fit_c <- lmer(y ~ (1|Operator)+(1|Part)+(1|Part:Operator), data=dat,
+ control= lmerControl(optimizer="Nelder_Mead", restart_edge=FALSE,
+ check.conv.hess="ignore"))
+ ## final fit gives degenerate-Hessian warning
+ ## FIXME: use fit_c with expect_warning() as a check on convergence tests
+ ## tolerance=1e-5 seems OK in interactive use but not in R CMD check ... ??
+ stopifnot(all.equal(getME(fit, "theta") -> th.f,
+ getME(fit_b,"theta"), tolerance=5e-5),
+ all(th.f > 0))
+
+ ## Manuel Koller
+
+ source(system.file("testdata", "koller-data.R", package="lme4"))
+ ldata <- getData(13)
+ ## old (backward compatible/buggy)
+ fm4 <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="Nelder_Mead",
+ use.last.params=TRUE),
+ start=list(theta=1))
+
+ fm4b <- lmer(y ~ (1|Var2), ldata,
+ control = lmerControl(optimizer="Nelder_Mead", use.last.params=TRUE,
+ restart_edge = FALSE,
+ check.conv.hess="ignore", check.conv.grad="ignore"),
+ start = list(theta=1))
+ ## FIXME: use as convergence test check
+ stopifnot(getME(fm4b,"theta") == 0)
+ fm4c <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="bobyqa",
+ use.last.params=TRUE),
+ start=list(theta=1))
+ stopifnot(all.equal(getME(fm4, "theta") -> th4,
+ getME(fm4c,"theta"), tolerance=1e-4),
+ th4 > 0)
+
+
+ ## new: doesn't get stuck at edge any more, but gets stuck somewhere else ...
+ fm5 <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="Nelder_Mead",
+ check.conv.hess="ignore",
+ check.conv.grad="ignore"),
+ start=list(theta=1))
+ fm5b <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="Nelder_Mead",
+ restart_edge=FALSE,
+ check.conv.hess="ignore",
+ check.conv.grad="ignore"),
+ start = list(theta = 1))
+ fm5c <- lmer(y ~ (1|Var2), ldata, control=lmerControl(optimizer="bobyqa"),
+ start = list(theta = 1))
+ stopifnot(all.equal(unname(getME(fm5c,"theta")), 0.21067645, tolerance = 1e-7))
+ # 0.21067644264 [64-bit, lynne]
+
+ if (require("optimx")) {
+ ## additional stuff for diagnosing Nelder-Mead problems.
+
+ fm5d <- update(fm5,control=lmerControl(optimizer="optimx",
+ optCtrl=list(method="L-BFGS-B")))
+
+ fm5e <- update(fm5, control=lmerControl(optimizer="nloptwrap"))
+
+ mList <- setNames(list(fm4,fm4b,fm4c,fm5,fm5b,fm5c,fm5d,fm5e),
+ c("NM/uselast","NM/uselast/norestart","bobyqa/uselast",
+ "NM","NM/norestart","bobyqa","LBFGSB","nloptr/bobyqa"))
+ pp <- profile(fm5c,which=1)
+ dd <- as.data.frame(pp)
+ par(las=1,bty="l")
+ v <- sapply(mList,
+ function(x) sqrt(VarCorr(x)[[1]]))
+ plot(.zeta^2~.sig01, data=dd, type="b")
+ abline(v=v)
+
+ res <- cbind(VCorr = sapply(mList, function(x) sqrt(VarCorr(x)[[1]])),
+ theta = sapply(mList, getME,"theta"),
+ loglik = sapply(mList, logLik))
+ res
+ print(sessionInfo(), locale=FALSE)
+ }
+
+ ######################
+ library(lattice)
+ ## testing boundary and near-boundary cases
+
+ tmpf <- function(i,...) {
+ set.seed(i)
+ d <- data.frame(x=rnorm(60),f=factor(rep(1:6,each=10)))
+ d$y <- simulate(~x+(1|f),family=gaussian,newdata=d,
+ newparams=list(theta=0.01,beta=c(1,1),sigma=5))[[1]]
+ lmer(y~x+(1|f),data=d,...)
+ }
+ sumf <- function(m) {
+ unlist(VarCorr(m))[1]
+ }
+ if (FALSE) {
+ ## figuring out which seeds will give boundary and
+ ## near-boundary solutions
+ mList <- lapply(1:201,tmpf) # [FIXME tons of messages "theta parameters vector not named"]
+ ss <- sapply(mList,sumf)+1e-50
+ par(las=1,bty="l")
+ hist(log(ss),col="gray",breaks=50)
+ ## values lying on boundary
+ which(log(ss)<(-40)) ## 5, 7-13, 15, 21, ...
+ ## values close to boundary (if check.edge not set)
+ which(log(ss)>(-40) & log(ss) <(-20)) ## 16, 44, 80, 86, 116, ...
+ }
+ ## diagnostic plot
+ tmpplot <- function(i, FUN=tmpf) {
+ dd <- FUN(i, devFunOnly=TRUE)
+ x <- 10^seq(-10,-6.5,length=201)
+ dvec <- sapply(x,dd)
+ op <- par(las=1,bty="l"); on.exit(par(op))
+ plot(x,dvec-min(dvec)+1e-16, log="xy", type="b")
+ r <- FUN(i)
+ abline(v = getME(r,"theta"), col=2)
+ invisible(r)
+ }
+
+ ## Case #1: boundary estimate with or without boundary.tol
+ m5 <- tmpf(5)
+ m5B <- tmpf(5,control=lmerControl(boundary.tol=0))
+ stopifnot(getME(m5, "theta")==0,
+ getME(m5B,"theta")==0)
+ p5 <- profile(m5) ## bobyqa warnings but results look reasonable
+ xyplot(p5)
+ ## reveals slight glitch (bottom row of plots doesn't look right)
+ expect_warning(splom(p5),"unreliable for singular fits")
+ p5B <- profile(m5, signames=FALSE) # -> bobyqa convergence warning (code 3)
+ expect_warning(splom(p5B), "unreliable for singular fits")
+
+ if(lme4:::testLevel() >= 2) { ## avoid failure to warn
+ ## Case #2: near-boundary estimate, but boundary.tol can't fix it
+ m16 <- tmpplot(16)
+ ## sometimes[2014-11-11] fails (??) :
+ p16 <- profile(m16) ## warning message*s* (non-monotonic profile and more)
+ plotOb <- xyplot(p16)
+ ## NB: It's the print()ing of 'plotOb' which warns ==> need to do this explicitly:
+ expect_warning(print(plotOb), ## warns about linear interpolation in profile for variable 1
+ "using linear interpolation")
+ d16 <- as.data.frame(p16)
+ xyplot(.zeta ~ .focal|.par, data=d16, type=c("p","l"),
+ scales = list(x=list(relation="free")))
+ try(splom(p16)) ## breaks when calling predict(.)
+ }
+
+ ## bottom line:
+ ## * xyplot.thpr could still be improved
+ ## * most of the near-boundary cases are noisy and can't easily be
+ ## fixed
+
+ tmpf2 <- function(i,...) {
+ set.seed(i)
+ d <- data.frame(x=rnorm(60),f=factor(rep(1:6,each=10)),
+ w=rep(10,60))
+ d$y <- simulate(~x+(1|f),family=binomial,
+ weights=d$w,newdata=d,
+ newparams=list(theta=0.01,beta=c(1,1)))[[1]]
+ glmer(y~x+(1|f),data=d,family=binomial,weights=w,...)
+ }
+
+ if (FALSE) {
+ ## figuring out which seeds will give boundary and
+ ## near-boundary solutions
+ mList <- lapply(1:201,tmpf2)
+ ss <- sapply(mList,sumf)+1e-50
+ par(las=1,bty="l")
+ hist(log(ss),col="gray",breaks=50)
+ ## values lying on boundary
+ head(which(log(ss)<(-50))) ## 1-5, 7 ...
+ ## values close to boundary (if check.edge not set)
+ which(log(ss)>(-50) & log(ss) <(-20)) ## 44, 46, 52, ...
+ }
+
+ ## m1 <- tmpf2(1)
+
+ ## FIXME: doesn't work if we generate m1 via tmpf2(1) --
+ ## some environment lookup problem ...
+
+ set.seed(1)
+ d <- data.frame(x=rnorm(60),f=factor(rep(1:6,each=10)),
+ w=rep(10,60))
+ d$y <- simulate(~x+(1|f),family=binomial,
+ weights=d$w,newdata=d,
+ newparams=list(theta=0.01,beta=c(1,1)))[[1]]
+ m1 <- glmer(y~x+(1|f),data=d,family=binomial,weights=w)
+
+ p1 <- profile(m1)
+ xyplot(p1)
+ expect_warning(splom(p1),"splom is unreliable")
+
+ } ## skip on windows (for speed)
Loading required package: Matrix
boundary (singular) fit: see help('isSingular')
Loading required package: optimx
R Under development (unstable) (2025-03-07 r87894)
Platform: x86_64-pc-linux-gnu
Running under: Fedora Linux 40 (Workstation Edition)
Matrix products: default
BLAS: /data/gannet/ripley/R/R-devel/lib/libRblas.so
LAPACK: /usr/lib64/liblapack.so.3.12.0 LAPACK version 3.12.0
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] optimx_2024-12.2 testthat_3.2.3 lme4_1.1-36 Matrix_1.7-2
loaded via a namespace (and not attached):
[1] R6_2.6.1 numDeriv_2016.8-1.1 lattice_0.22-6
[4] magrittr_2.0.3 splines_4.5.0 cli_3.6.4
[7] Rdpack_2.6.2 nloptr_2.1.1 grid_4.5.0
[10] reformulas_0.4.0 compiler_4.5.0 boot_1.3-31
[13] rbibutils_2.3 tools_4.5.0 pracma_2.4.4
[16] brio_1.1.5 nlme_3.1-167 minqa_1.2.8
[19] Rcpp_1.0.14 rlang_1.1.5 MASS_7.3-65
Error: getME(m5, "theta") == 0 is not TRUE
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
4: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
5: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.07244 (tol = 0.002, component 1)
6: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
7: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.07244 (tol = 0.002, component 1)
Execution halted
Running the tests in ‘tests/falsezero_dorie.R’ failed.
Complete output:
> if (.Platform$OS.type != "windows") {
+ ## test of false zero problem reported by Vince Dorie
+ ## (no longer occurs with current development lme4)
+ ## https://github.com/lme4/lme4/issues/17
+ library(lme4)
+
+ sigma.eps <- 2
+ sigma.the <- 0.75
+ mu <- 2
+
+ n <- 5
+ J <- 10
+ g <- gl(J, n)
+
+ set.seed(1)
+
+ theta <- rnorm(J, 0, sigma.eps * sigma.the)
+ y <- rnorm(n * J, mu + theta[g], sigma.eps)
+ lmerFit <- lmer(y ~ 1 + (1 | g), REML = FALSE, verbose=TRUE)
+
+ y.bar <- mean(y)
+ y.bar.j <- sapply(1:J, function(j) mean(y[g == j]))
+ S.w <- sum((y - y.bar.j[g])^2)
+ S.b <- n * sum((y.bar.j - y.bar)^2)
+ R <- S.b / S.w
+
+ sigma.the.hat <- sqrt(max((n - 1) * R / n - 1 / n, 0))
+ stopifnot(all.equal(sigma.the.hat,lme4Sigma <- unname(getME(lmerFit,"theta")),
+ tolerance=2e-5))
+ } ## skip on windows (for speed)
Loading required package: Matrix
Error: sigma.the.hat and lme4Sigma <- unname(getME(lmerFit, "theta")) are not equal:
Mean relative difference: 1.134618
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.17399 (tol = 0.002, component 1)
Execution halted
Running the tests in ‘tests/lme4_nlme.R’ failed.
Complete output:
> if (lme4:::testLevel() > 1 || .Platform$OS.type != "windows") withAutoprint({
+
+ ## testing whether lme4 and nlme play nicely. Only known issue
+ ## is lmList-masking ...
+ library("lme4")
+ library("nlme")
+ fm1_lmer <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy)
+ fm1_lme <- lme (Reaction ~ Days, random = ~Days|Subject, sleepstudy)
+ ## variance-covariance matrices: annoyingly different structures
+ vc_lmer <- VarCorr(fm1_lmer)
+ vc_lme <- VarCorr(fm1_lme, rdig = 8)
+ suppressWarnings(storage.mode(vc_lme) <- "numeric")# 2 NAs
+ vc_lmerx <- c(diag(vc_lmer[[1]]), attr(vc_lmer[[1]],"correlation")[1,2])
+ vc_lmex <- c( vc_lme[1:2,1], vc_lme[2,3])
+ stopifnot(
+ all.equal(vc_lmex, vc_lmerx, tolerance= 4e-4) # had 3e-5, now see 0.000296
+ , ## fixed effects (much easier) :
+ all.equal(fixef(fm1_lmer), fixef(fm1_lme)) # 3.6e-15
+ ,
+ all.equal(unname(unlist(unclass(ranef(fm1_lmer)))),
+ unname(unlist(unclass(ranef(fm1_lme)))),
+ tolerance = 2e-4) # had 2e-5, now see 8.41e-5
+ )
+
+ fm1L_lme <- nlme::lmList(distance ~ age | Subject, Orthodont)
+ fm1L_lmer <- lme4::lmList(distance ~ age | Subject, Orthodont)
+ stopifnot(all.equal(fixef(fm1L_lmer),
+ fixef(fm1L_lme)))
+ sm1L_e <- summary(fm1L_lme)
+ sm1L_er <- summary(fm1L_lmer)
+ stopifnot(
+ all.equal(coef(sm1L_e),
+ coef(sm1L_er), tol=1e-12)# even tol=0 works on some Lnx 64b
+ )
+
+ ## FIXME: test opposite order
+ })
> library("lme4")
Loading required package: Matrix
> library("nlme")
Attaching package: 'nlme'
The following object is masked from 'package:lme4':
lmList
> fm1_lmer <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy)
> fm1_lme <- lme(Reaction ~ Days, random = ~Days | Subject, sleepstudy)
> vc_lmer <- VarCorr(fm1_lmer)
> vc_lme <- VarCorr(fm1_lme, rdig = 8)
> suppressWarnings(storage.mode(vc_lme) <- "numeric")
> vc_lmerx <- c(diag(vc_lmer[[1]]), attr(vc_lmer[[1]], "correlation")[1,
+ 2])
> vc_lmex <- c(vc_lme[1:2, 1], vc_lme[2, 3])
> stopifnot(all.equal(vc_lmex, vc_lmerx, tolerance = 4e-04), all.equal(fixef(fm1_lmer),
+ fixef(fm1_lme)), all.equal(unname(unlist(unclass(ranef(fm1_lmer)))), unname(unlist(unclass(ranef(fm1_lme)))),
+ tolerance = 2e-04))
Error: vc_lmex and vc_lmerx are not equal:
Mean relative difference: 0.8916242
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
Execution halted
Running the tests in ‘tests/lmer-0.R’ failed.
Complete output:
> require(lme4)
Loading required package: lme4
Loading required package: Matrix
> source(system.file("test-tools-1.R", package = "Matrix"))# identical3() etc
Loading required package: tools
>
> ## use old (<=3.5.2) sample() algorithm if necessary
> if ("sample.kind" %in% names(formals(RNGkind))) {
+ suppressWarnings(RNGkind("Mersenne-Twister", "Inversion", "Rounding"))
+ }
>
> ## Check that quasi families throw an error
> assertError(lmer(cbind(incidence, size - incidence) ~ period + (1|herd),
+ data = cbpp, family = quasibinomial))
> assertError(lmer(incidence ~ period + (1|herd),
+ data = cbpp, family = quasipoisson))
> assertError(lmer(incidence ~ period + (1|herd),
+ data = cbpp, family = quasi))
>
> ## check bug found by Kevin Buhr
> set.seed(7)
> n <- 10
> X <- data.frame(y=runif(n), x=rnorm(n), z=sample(c("A","B"), n, TRUE))
> fm <- lmer(log(y) ~ x | z, data=X) ## ignore grouping factors with
Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
> ## gave error inside model.frame()
> stopifnot(all.equal(c(`(Intercept)` = -0.834544), fixef(fm), tolerance=.01))
Error: c(`(Intercept)` = -0.834544) and fixef(fm) are not equal:
Mean relative difference: 0.4954795
Execution halted
Running the tests in ‘tests/minval.R’ failed.
Complete output:
> if (lme4:::testLevel() > 1 || .Platform$OS.type!="windows") {
+ ## example posted by Stéphane Laurent
+ ## exercises bug where Nelder-Mead min objective function value was >0
+ set.seed(666)
+ sims <- function(I, J, sigmab0, sigmaw0){
+ Mu <- rnorm(I, mean=0, sd=sigmab0)
+ y <- c(sapply(Mu, function(mu) rnorm(J, mu, sigmaw0)))
+ data.frame(y=y, group=gl(I,J))
+ }
+
+ I <- 3 # number of groups
+ J <- 8 # number of repeats per group
+ sigmab0 <- 0.15 # between standard deviation
+ sigmaw0 <- 0.15 # within standard deviation
+
+ dat <- sims(I, J, sigmab0, sigmaw0)
+
+ library(lme4)
+ isOldTol <- environment(nloptwrap)$defaultControl$xtol_abs == 1e-6
+
+ fm3 <- lmer(y ~ (1|group), data=dat)
+ stopifnot(all.equal(unname(unlist(VarCorr(fm3))),
+ switch(fm3@optinfo$optimizer,
+ "Nelder_Mead" = 0.029662844,
+ "bobyqa" = 0.029662698,
+ "nloptwrap" =
+ if (isOldTol) 0.029679755 else 0.029662699,
+ stop("need new case here: value is ",unname(unlist(VarCorr(fm3))))
+ ),
+ tolerance = 1e-7))
+ } ## skip on windows (for speed)
Loading required package: Matrix
Error: unname(unlist(VarCorr(fm3))) and switch(fm3@optinfo$optimizer, Nelder_Mead = 0.029662844, bobyqa = 0.029662698, .... are not equal:
Mean relative difference: 0.1265733
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 0.147111 (tol = 0.002, component 1)
Execution halted
Running the tests in ‘tests/modFormula.R’ failed.
Complete output:
> if (.Platform$OS.type != "windows") {
+ library(lme4)
+ library(testthat)
+
+ .get.checkingOpts <- lme4:::.get.checkingOpts
+ stopifnot(identical(
+ .get.checkingOpts(
+ c("CheckMe", "check.foo", "check.conv.1", "check.rankZ", "check.rankX"))
+ , c("check.foo", "check.rankZ")))
+
+ lmod <- lFormula(Reaction ~ Days + (Days|Subject), sleepstudy)
+ devfun <- do.call(mkLmerDevfun, lmod)
+ opt <- optimizeLmer(devfun)
+ cc <- lme4:::checkConv(attr(opt,"derivs"), opt$par, ctrl = lmerControl()$checkConv,
+ lbound=environment(devfun)$lower)
+ fm1 <- mkMerMod(environment(devfun), opt, lmod$reTrms, fr = lmod$fr,
+ lme4conv=cc)
+ fm2 <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy)
+
+ ## basic equivalence
+ fm1C <- fm1
+ fm1C@call <- fm2@call
+ expect_equal(fm2,fm1C)
+ expect_equal(range(residuals(fm1)), c(-101.18, 132.547), tolerance = 1e-5) # these are "outliers"!
+ expect_is(model.frame(fm1),"data.frame")
+ ## formulae
+ mfm1 <- model.frame(fm1)
+ expect_equal(formula(fm1), Reaction ~ Days + (Days | Subject))
+ expect_equal(formula(terms(mfm1)), Reaction ~ Days + (Days + Subject))
+ new_form_modframe <- (getRversion() >= "3.6.0" &&
+ as.numeric(version[["svn rev"]]) >= 75891)
+ expect_equal(formula(mfm1),
+ if(new_form_modframe) {
+ Reaction ~ Days + (Days + Subject)
+ } else
+ Reaction ~ Days + Subject
+ )
+ ## predictions
+ expect_equal(predict(fm1,newdata=sleepstudy[1:10,],re.form=NULL),
+ predict(fm2,newdata=sleepstudy[1:10,],re.form=NULL))
+ expect_equal(predict(fm1,newdata=sleepstudy),
+ predict(fm1))
+
+ lmodOff <- lFormula(Reaction ~ Days + (Days|Subject) + offset(0.5*Days),
+ sleepstudy)
+ devfunOff <- do.call(mkLmerDevfun, lmodOff)
+ opt <- optimizeLmer(devfunOff)
+ fm1Off <- mkMerMod(environment(devfunOff), opt, lmodOff$reTrms, fr = lmodOff$fr)
+ fm2Off <- lmer(Reaction ~ Days + (Days|Subject) + offset(0.5*Days), sleepstudy)
+ expect_equal(predict(fm1Off,newdata=sleepstudy[1:10,],re.form=NULL),
+ predict(fm2Off,newdata=sleepstudy[1:10,],re.form=NULL))
+
+ ## FIXME: need more torture tests with offset specified, in different environments ...
+
+ ## FIXME: drop1(.) doesn't work with modular objects ... hard to see how it
+ ## could, though ...
+ ## drop1(fm1Off)
+ drop1(fm2Off)
+
+ } ## skip on windows (for speed)
Loading required package: Matrix
Error: range(residuals(fm1)) not equal to c(-101.18, 132.547).
2/2 mismatches (average diff: 2.42)
[1] -106 - -101 == -4.708
[2] 133 - 133 == 0.129
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In lme4:::checkConv(attr(opt, "derivs"), opt$par, ctrl = lmerControl()$checkConv, :
unable to evaluate scaled gradient
3: In lme4:::checkConv(attr(opt, "derivs"), opt$par, ctrl = lmerControl()$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
4: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
5: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
6: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
Execution halted
Running the tests in ‘tests/predsim.R’ failed.
Complete output:
> ## compare range, average, etc. of simulations to
> ## conditional and unconditional prediction
> library(lme4)
Loading required package: Matrix
> do.plot <- FALSE
>
> if (.Platform$OS.type != "windows") {
+ ## use old (<=3.5.2) sample() algorithm if necessary
+ if ("sample.kind" %in% names(formals(RNGkind))) {
+ suppressWarnings(RNGkind("Mersenne-Twister", "Inversion", "Rounding"))
+ }
+
+ fm1 <- lmer(Reaction~Days+(1|Subject),sleepstudy)
+ set.seed(101)
+ pp <- predict(fm1)
+ rr <- range(usim2 <- simulate(fm1,1,use.u=TRUE)[[1]])
+ stopifnot(all.equal(rr,c(159.3896,439.1616),tolerance=1e-6))
+ if (do.plot) {
+ plot(pp,ylim=rr)
+ lines(sleepstudy$Reaction)
+ points(simulate(fm1,1)[[1]],col=4)
+ points(usim2,col=2)
+ }
+
+ set.seed(101)
+
+ ## conditional prediction
+ ss <- simulate(fm1,1000,use.u=TRUE)
+ ss_sum <- t(apply(ss,1,quantile,c(0.025,0.5,0.975)))
+ plot(pp)
+ matlines(ss_sum,col=c(1,2,1),lty=c(2,1,2))
+ stopifnot(all.equal(ss_sum[,2],pp,tolerance=5e-3))
+
+ ## population-level prediction
+ pp2 <- predict(fm1, re.form=NA)
+ ss2 <- simulate(fm1,1000,use.u=FALSE)
+ ss_sum2 <- t(apply(ss2,1,quantile,c(0.025,0.5,0.975)))
+
+ if (do.plot) {
+ plot(pp2,ylim=c(200,400))
+ matlines(ss_sum2,col=c(1,2,1),lty=c(2,1,2))
+ }
+
+ stopifnot(all.equal(ss_sum2[,2],pp2,tolerance=8e-3))
+
+ ## predict(...,newdata=...) on models with derived variables in the random effects
+ ## e.g. (f:g, f/g)
+ set.seed(101)
+ d <- expand.grid(f=factor(letters[1:10]),g=factor(letters[1:10]),
+ rep=1:10)
+ d$y <- rnorm(nrow(d))
+ m1 <- lmer(y~(1|f:g),d)
+ p1A <- predict(m1)
+ p1B <- predict(m1,newdata=d)
+ stopifnot(all.equal(p1A,p1B))
+ m2 <- lmer(y~(1|f/g),d)
+ p2A <- predict(m2)
+ p2B <- predict(m2,newdata=d)
+ stopifnot(all.equal(p2A,p2B))
+
+ ## with numeric grouping variables
+ dn <- transform(d,f=as.numeric(f),g=as.numeric(g))
+ m1N <- update(m1,data=dn)
+ p1NA <- predict(m1N)
+ p1NB <- predict(m1N,newdata=dn)
+ stopifnot(all.equal(p1NA,p1NB))
+
+ ## simulate with modified parameters
+ set.seed(1)
+ s1 <- simulate(fm1)
+ set.seed(1)
+ s2 <- simulate(fm1,newdata=model.frame(fm1),
+ newparams=getME(fm1,c("theta","beta","sigma")))
+ all.equal(s1,s2)
+
+ fm0 <- update(fm1,.~.-Days)
+ ##
+ ## sim() -> simulate() -> refit() -> deviance
+ ##
+
+ ## predictions and simulations with offsets
+
+ set.seed(101)
+ d <- data.frame(y=rpois(100,5),x=rlnorm(100,1,1),
+ f=factor(sample(10,size=100,replace=TRUE)))
+ gm1 <- glmer(y~offset(log(x))+(1|f),data=d,
+ family=poisson)
+ s1 <- simulate(gm1)
+ } ## skip on windows (for speed)
Error: rr and c(159.3896, 439.1616) are not equal:
Mean relative difference: 0.007853237
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 1.97927 (tol = 0.002, component 1)
Execution halted
Running the tests in ‘tests/priorWeights.R’ failed.
Complete output:
> ## use old (<=3.5.2) sample() algorithm if necessary
> if ("sample.kind" %in% names(formals(RNGkind))) {
+ suppressWarnings(RNGkind("Mersenne-Twister", "Inversion", "Rounding"))
+ }
>
>
> compFunc <- function(lmeMod, lmerMod, tol = 1e-2){
+ lmeVarCorr <- nlme:::VarCorr(lmeMod)[,"StdDev"]
+ lmeCoef <- summary(lmeMod)$tTable[,-c(3,5)]
+ lmeOut <- c(as.numeric(lmeVarCorr), as.numeric(lmeCoef))
+ keep <- !is.na(lmeOut)
+ lmeOut <- lmeOut[keep]
+ dn <- dimnames(lmeCoef)
+ if(is.null(dn)) dn <- list("", names(lmeCoef))
+ names(lmeOut) <- c(
+ paste(names(lmeVarCorr), "Var"),
+ as.character(do.call(outer, c(dn, list("paste")))))[keep]
+
+ ## get nested RE variances in the same order as nlme
+ ## FIXME: not sure if this works generally
+ vcLmer <- VarCorr(lmerMod)
+ vcLmer <- vcLmer[length(vcLmer):1]
+ ##
+
+ lmerVarCorr <- c(sapply(vcLmer, attr, "stddev"),
+ attr(VarCorr(lmerMod), "sc"))
+ ## differentiate lme4{new} and lme4.0 :
+ lmerCoef <- if(is(lmerMod, "merMod"))
+ summary(lmerMod)$coefficients else summary(lmerMod)@coefs
+ lmerOut <- c(lmerVarCorr, as.numeric(lmerCoef))
+ names(lmerOut) <- names(lmeOut)
+
+ return(list(target = lmeOut, current = lmerOut, tolerance = tol))
+ }
>
> if (.Platform$OS.type != "windows") {
+ set.seed(1)
+ nGroups <- 100
+ nObs <- 1000
+
+ # explanatory variable with a fixed effect
+ explVar1 <- rnorm(nObs)
+ explVar2 <- rnorm(nObs)
+
+ # random intercept among levels of a grouping factor
+ groupFac <- as.factor(rep(1:nGroups,each=nObs/nGroups))
+ randEff0 <- rep(rnorm(nGroups),each=nObs/nGroups)
+ randEff1 <- rep(rnorm(nGroups),each=nObs/nGroups)
+ randEff2 <- rep(rnorm(nGroups),each=nObs/nGroups)
+
+ # residuals with heterogeneous variance
+ residSD <- rpois(nObs,1) + 1
+ residError <- rnorm(nObs,sd=residSD)
+
+ # response variable
+ respVar <- randEff0 + (1+randEff1)*explVar1 + (1+randEff2)*explVar2 + residError
+
+ # rename to fit models on one line
+ y <- respVar
+ x <- explVar1
+ z <- explVar2
+ g <- groupFac
+ v <- residSD^2
+ w <- 1/v
+
+ library("nlme")
+ lmeMods <- list(
+ ML1 = lme(y ~ x, random = ~ 1|g, weights = varFixed(~v), method = "ML"),
+ REML1 = lme(y ~ x, random = ~ 1|g, weights = varFixed(~v), method = "REML"),
+ ML2 = lme(y ~ x, random = ~ x|g, weights = varFixed(~v), method = "ML"),
+ REML2 = lme(y ~ x, random = ~ x|g, weights = varFixed(~v), method = "REML"),
+ ML1 = lme(y ~ x+z, random = ~ x+z|g, weights = varFixed(~v), method = "ML"),
+ REML2 = lme(y ~ x+z, random = ~ x+z|g, weights = varFixed(~v), method = "REML"))
+
+ library("lme4")
+ lmerMods <- list(
+ ML1 = lmer(y ~ x + (1|g), weights = w, REML = FALSE),
+ REML1 = lmer(y ~ x + (1|g), weights = w, REML = TRUE),
+ ML2 = lmer(y ~ x + (x|g), weights = w, REML = FALSE),
+ REML2 = lmer(y ~ x + (x|g), weights = w, REML = TRUE),
+ ML3 = lmer(y ~ x + z + (x+z|g), weights = w, REML = FALSE),
+ REML3 = lmer(y ~ x + z + (x+z|g), weights = w, REML = TRUE))
+
+ comp <- mapply(compFunc, lmeMods, lmerMods, SIMPLIFY=FALSE)
+ stopifnot(all(sapply(comp, do.call, what = all.equal)))
+ ## Look at the relative differences:
+ sapply(mapply(compFunc, lmeMods, lmerMods, SIMPLIFY=FALSE, tol = 0),
+ do.call, what = all.equal)
+
+ ## add simulated weights to the sleepstudy example
+ n <- nrow(sleepstudy)
+ v <- rpois(n,1) + 1
+ w <- 1/v
+ sleepLme <- lme(Reaction ~ Days, random = ~ Days|Subject,
+ sleepstudy, weights = varFixed(~v),
+ method = "ML")
+ sleepLmer <- lmer(Reaction ~ Days + (Days|Subject),
+ sleepstudy, weights = w,
+ REML = FALSE)
+ sleepComp <- compFunc(sleepLme, sleepLmer)
+ stopifnot(do.call(all.equal, sleepComp))
+ ## look at relative differences:
+ sleepComp$tolerance <- 0
+ do.call(all.equal, sleepComp)
+
+ if (require("mlmRev")) {
+ n <- nrow(Chem97)
+ v <- rpois(n,1) + 1
+ w <- 1/v
+ Chem97Lme <- lme(score ~ 1, random = ~ 1|lea/school, Chem97)
+ Chem97Lmer <- lmer(score ~ (1|lea/school), Chem97)
+ Chem97Comp <- compFunc(Chem97Lme, Chem97Lmer)
+ stopifnot(do.call(all.equal, Chem97Comp))
+ ## look at relative differences:
+ Chem97Comp$tolerance <- 0
+ do.call(all.equal, Chem97Comp)
+ }
+
+ set.seed(2)
+ n <- 40
+ w <- runif(n)
+ x <- runif(n)
+ g <- factor(sample(1:10,n,replace=TRUE))
+ Z <- model.matrix(~g-1);
+ y <- Z%*%rnorm(ncol(Z)) + x + rnorm(n)/w^.5
+ m <- lmer(y ~ x + (1|g), weights=w, REML = TRUE)
+
+ ## CRAN-forbidden:
+ ## has4.0 <- require("lme4.0"))
+ has4.0 <- FALSE
+ if(has4.0) {
+ ## m.0 <- lme4.0::lmer(y ~ x + (1|g), weights=w, REML = TRUE)
+ lmer0 <- get("lmer", envir=asNamespace("lme4.0"))
+ m.0 <- lmer0(y ~ x + (1|g), weights=w, REML = TRUE)
+ dput(fixef(m.0)) # c(-0.73065400610675, 2.02895402562926)
+ dput(sigma(m.0)) # 1.73614301673377
+ dput(VarCorr(m.0)$g[1,1]) # 2.35670451590395
+ dput(unname(coef(summary(m.0))[,"Std. Error"]))
+ ## c(0.95070076853232, 1.37650858268602)
+ }
+ fixef_lme4.0 <- c(-0.7306540061, 2.0289540256)
+ sigma_lme4.0 <- 1.7361430
+ Sigma_lme4.0 <- 2.3567045
+ SE_lme4.0 <- c(0.95070077, 1.37650858)
+ if(has4.0) try(detach("package:lme4.0"))
+
+ stopifnot(all.equal(unname(fixef(m)), fixef_lme4.0, tolerance = 1e-3))
+ all.equal(unname(fixef(m)), fixef_lme4.0, tolerance = 0) #-> 1.657e-5
+
+ ## but these are not at all equal :
+ (all.equal(sigma(m), sigma_lme4.0, tolerance = 10^-3)) # 0.4276
+ (all.equal(as.vector(VarCorr(m)$g), Sigma_lme4.0, tolerance = 10^-3)) # 1.038
+ (all.equal(as.vector(summary(m)$coefficients[,2]), SE_lme4.0, tolerance = 10^-3)) # 0.4276
+ ## so, lme4.0 was clearly wrong here
+
+
+ ##' make sure models that differ only in a constant
+ ##' prior weight have identical deviance:
+ fm <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy,REML=FALSE)
+ fm_wt <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy, weights = rep(5, nrow(sleepstudy)),REML=FALSE)
+ all.equal(deviance(fm), deviance(fm_wt))
+ } ## skip on windows (for speed)
Loading required package: Matrix
Attaching package: 'lme4'
The following object is masked from 'package:nlme':
lmList
Error: all(sapply(comp, do.call, what = all.equal)) is not TRUE
In addition: There were 13 warnings (use warnings() to see them)
Execution halted
Running the tests in ‘tests/priorWeightsModComp.R’ failed.
Complete output:
> library(lme4)
Loading required package: Matrix
> n <- nrow(sleepstudy)
> op <- options(warn = 1, # show as they happen ("false" convergence warnings)
+ useFancyQuotes = FALSE)
>
> if (.Platform$OS.type != "windows") {
+ ##' remove all attributes but names
+ dropA <- function(x) `attributes<-`(x, list(names = names(x)))
+ ##' transform result of "numeric" all.equal.list() to a named vector
+ all.eqL <- function(x1, x2, ...) {
+ r <- sub("^Component ", '', all.equal(x1, x2, tolerance = 0, ...))
+ r <- strsplit(sub(": Mean relative difference:", "&&", r),
+ split="&&", fixed=TRUE)
+ setNames(as.numeric(vapply(r, `[`, "1.234", 2L)),
+ ## drop surrounding "..."
+ nm = sub('"$', '', substring(vapply(r, `[`, "nam", 1L), first=2)))
+ }
+ seedF <- function(s) {
+ if(s %in% c(6, 39, 52, 57, 63, 74, 76, 86))
+ switch(as.character(s)
+ , "52"=, "63"=, "74" = 2
+ , "6"=, "39" = 3
+ , "86" = 8 # needs 4 on Lnx-64b
+ , "76" = 70 # needs 42 on Lnx-64b
+ , "57" = 90 # needs 52 on Lnx-64b
+ )
+ else if(s %in% c(1, 12, 15, 34, 36, 41, 42, 43, 49, 55, 59, 67, 80, 85)) ## seeds 41,59, .. 15
+ 1.0
+ else ## seeds 22, 20, and better
+ 0.25
+ }
+ ## be fast, running only 10 seeds by default:
+ sMax <- if(lme4:::testLevel() > 1) 99L else 9L
+ mySeeds <- 0L:sMax
+
+ lapply(setNames(,mySeeds), function(seed) {
+ cat("\n------ random seed =", seed, "---------\n")
+ set.seed(seed)
+ v <- rpois(n,1) + 1
+ w <- 1/v
+ cat("weights w:\n")
+ fm1 <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy, REML=FALSE, weights = w); cat("..2:\n")
+ fm2 <- lmer(Reaction ~ Days + (1 | Subject), sleepstudy, REML=FALSE, weights = w)
+ cat("weights w*10:\n")
+ fm1.10 <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy, REML=FALSE, weights = w*10);cat("..2:\n")
+ fm2.10 <- lmer(Reaction ~ Days + (1 | Subject), sleepstudy, REML=FALSE, weights = w*10)
+ ##
+ ano12... <- dropA(anova(fm1, fm2 ))
+ ano12.10 <- dropA(anova(fm1.10, fm2.10))
+ print(aEQ <- all.eqL(ano12..., ano12.10)) # showing differences
+ if(!exists("notChisq"))
+ notChisq <<-
+ local({ n <- names(ano12...)
+ grep("Chisq", n, value=TRUE, fixed=TRUE, invert=TRUE) })
+ stopifnot(
+ all.equal(ano12...$Chisq,
+ ano12.10$Chisq, tol = 1e-6 * seedF(seed))
+ ,
+ all.equal(ano12...[notChisq],
+ ano12.10[notChisq], tol= 1.5e-8 * seedF(seed))
+ )
+ aEQ
+ }) -> rallEQ
+
+ cat("=====================================\n")
+
+ rallEQ <- t(simplify2array(rallEQ))
+ notChisq <- intersect(notChisq, colnames(rallEQ))
+ ## sort according to "severity":
+ srallEQ <- rallEQ[with(as.data.frame(rallEQ), order(AIC, Chisq)), ]
+ round(log10(srallEQ), 2)
+ saveRDS(srallEQ, "priorWeightsMod_relerr.rds")
+
+ if(!dev.interactive(orNone=TRUE)) pdf("priorWeightsMod_relerr.pdf")
+
+ matplot(mySeeds, log10(srallEQ), type="l", xlab=NA) ; grid()
+ legend("topleft", ncol=3, bty="n",
+ paste(1:6, colnames(srallEQ), sep = ": "), col=1:6, lty=1:6)
+ tolD <- sqrt(.Machine$double.eps) # sqrt(eps_C)
+ abline(h = log10(tolD), col = "forest green", lty=3)
+ axis(4, at=log10(tolD), label=quote(sqrt(epsilon[c])), las=1)
+ LRG <- which(srallEQ[,"AIC"] > tolD)
+ if (length(LRG)>0) {
+ text(LRG, log10(srallEQ[LRG, "AIC"]), names(LRG), cex = .8)
+ }
+
+ ## how close are we ..
+ str(tF <- sapply(mySeeds, seedF))
+ round(sort( rallEQ[, "Chisq"] / (tF * 1e-6 ), decreasing=TRUE), 1)
+ round(sort(apply(rallEQ[,notChisq] / (tF * 1.5e-8), 1, max), decreasing=TRUE), 1)
+ } ## skip on windows (for speed)
------ random seed = 0 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.12755 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 1 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.00449 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 2 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.68252 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 26.1456 (tol = 0.002, component 1)
RUE
NA
------ random seed = 3 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.11241 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 4 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.41542 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 5 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.86338 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 6 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.58993 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 7 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.57508 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 8 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 3.62162 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
------ random seed = 9 ---------
weights w:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 2.73043 (tol = 0.002, component 1)
weights w*10:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 2 negative eigenvalues
..2:
Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
unable to evaluate scaled gradient
Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
RUE
NA
=====================================
Error in eval(substitute(expr), data, enclos = parent.frame()) :
object 'Chisq' not found
Calls: with -> with.default -> eval -> eval -> order
Execution halted
Running the tests in ‘tests/vcov-etc.R’ failed.
Complete output:
> stopifnot(require(lme4))
Loading required package: lme4
Loading required package: Matrix
>
> (testLevel <- lme4:::testLevel())
[1] 1
> source(system.file("testdata", "lme-tst-funs.R", package="lme4", mustWork=TRUE))# -> unn()
>
>
> ## "MEMSS" is just 'Suggest' -- must still work, when it's missing:
> if (suppressWarnings(!require(MEMSS, quietly=TRUE)) ||
+ (data(ergoStool, package="MEMSS") != "ergoStool")) {
+
+ cat("'ergoStool' data from package 'MEMSS' is not available --> skipping test\n")
+ } else {
+
+ fm1 <- lmer (effort ~ Type + (1|Subject), data = ergoStool)
+ ##sp no longer supported since ~ 2012-3:
+ ##sp fm1.s <- lmer (effort ~ Type + (1|Subject), data = ergoStool, sparseX=TRUE)
+ ## was segfaulting with sparseX (a while upto 2010-04-06)
+
+ fe1 <- fixef(fm1)
+ ##sp fe1.s <- fixef(fm1.s)
+
+ print(s1.d <- summary(fm1))
+ ##sp print(s1.s <- summary(fm1.s))
+ Tse1.d <- c(0.57601226, rep(0.51868384, 3))
+ stopifnot(exprs = {
+ ##sp all.equal(fe1, fe1.s, tolerance= 1e-12)
+ all.equal(Tse1.d, unname(se1.d <- coef(s1.d)[,"Std. Error"]),
+ tolerance = 1e-6) # std.err.: no too much accuracy
+ is(V.d <- vcov(fm1), "symmetricMatrix")
+ ##sp all.equal(se1.d, coef(s1.s)[,"Std. Error"])#, tol = 1e-10
+ ##sp all.equal( V.d, vcov(fm1.s))#, tol = 1e-9
+ all.equal(Matrix::diag(V.d), unn(se1.d)^2, tolerance= 1e-12)
+ })
+
+ }## if( ergoStool is available from pkg MEMSS )
Attaching package: 'MEMSS'
The following objects are masked from 'package:datasets':
CO2, Orange, Theoph
Linear mixed model fit by REML ['lmerMod']
Formula: effort ~ Type + (1 | Subject)
Data: ergoStool
REML criterion at convergence: Inf
Scaled residuals:
Min 1Q Median 3Q Max
-1.55001 -0.64996 -0.04112 0.60381 1.64861
Random effects:
Groups Name Variance Std.Dev.
Subject (Intercept) 0.915 0.9565
Residual 1.517 1.2316
Number of obs: 36, groups: Subject, 9
Fixed effects:
Estimate Std. Error t value
(Intercept) 8.5556 0.5198 16.459
TypeT2 3.8889 0.5806 6.698
TypeT3 2.2222 0.5806 3.827
TypeT4 0.6667 0.5806 1.148
Correlation of Fixed Effects:
(Intr) TypeT2 TypeT3
TypeT2 -0.558
TypeT3 -0.558 0.500
TypeT4 -0.558 0.500 0.500
optimizer (nloptwrap) convergence code: -2 (NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).)
Model failed to converge with max|grad| = 1.60567 (tol = 0.002, component 1)
Error: Tse1.d and unname(se1.d <- coef(s1.d)[, "Std. Error"]) are not equal:
Mean relative difference: 0.1134759
In addition: Warning messages:
1: In optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, :
convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS: Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).
2: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 1.60567 (tol = 0.002, component 1)
Execution halted
Flavor: r-devel-linux-x86_64-fedora-gcc
Version: 1.1-36
Check: compiled code
Result: NOTE
File 'lme4/libs/x64/lme4.dll':
Found non-API call to R: 'CLOENV'
Compiled code should not call non-API entry points in R.
See 'Writing portable packages' in the 'Writing R Extensions' manual,
and section 'Moving into C API compliance' for issues with the use of
non-API entry points.
Flavor: r-devel-windows-x86_64
Version: 1.1-36
Check: installed package size
Result: NOTE
installed size is 27.4Mb
sub-directories of 1Mb or more:
R 1.5Mb
doc 1.6Mb
libs 22.2Mb
testdata 1.3Mb
Flavors: r-release-macos-arm64, r-release-macos-x86_64, r-release-windows-x86_64, r-oldrel-macos-arm64, r-oldrel-macos-x86_64, r-oldrel-windows-x86_64