Added a Markdown README file.
Removed old compiler directives in Fortran sources: they were relevant only for Windows compilers (CVF and Intel).
In Srho.test.ts and Srho.test.ts.p,
when ci.type='mbb' it is now possible to choose the moving
bootstrap block length through the switch blag.
The workhorse function Srho.func and its Fortran
engine have been rewritten as to allow generic non-diagonal bandwidth
selectors. Also, the Fortran code has been vectorized and now the
hcubature routine can take advantage from this. The
routines for computing the bivariate bandwidths have been amended as to
output a matrix. The old routines are still in the source code but not
exported and they will be removed in a future release. A new option
bdiag has been added to Srho.ts and related
files. If bdiag=FALSE, then the bandwitdh matrix is
unstructured.
Fixed two bugs in the Fortran routine
Srhosum.
The help files of the parallel routines have been merged with that of the non-parallel ones and improved.
Minor fixes to the show method of
Srho-class.
Bug fix: in Srho.test the slot @call
now contains the correct call.
In surrogate.AR and surrogate.ARs
order.max is set to NULL, which is the default
in stats::ar.
Added the horizontal line at 0 in the plot of
Srho-class consistently with that of
Srho.test-class.
In surrogate.SA changed the default value
eps.SA to 0.05.
Changed the links to the papers as to use doi instead of url.
The integration routine name adaptIntegrate has been
changed to its alias hcubature to comply with the change in
the cubature package. The documentation has been amended
accordingly.
The ellipsis has been added to all the Srho-related
routined as to allow passing additional arguments, typically to the
hcubature routine. Incidentally, the option maxpts that was
passed to maxEval has been removed.
Fixed a buglet in Srho.ts.uni.
The Fortran code has been made cleaner and more portable by setting the machine-independent precision REAL64 through the module ISO_FORTRAN_ENV.
The permutation/bootstrap versions of the test for discrete data
and the surrogate.SA routine now call the R API C routines
for random number generation. The results are now fully reproducible by
setting set.seed.
Fixed lag.max in the documentation.
Fixed a buglet in plot methods for S4 classes Srho
and Srho.test: the switch ‘main’ was ignored.
Adjusted margins in plot method for S4 class
Srho.test.
nslaves has been replaced with nwork
throughout the package.
Added bandwidth selection methods to Srho.ts from
package ks: least square cross validation (lscv), smoothed
cross validation (scv), plugin (pi). The documentation of
Srho.ts has been updated accordingly.
In Srho.ts.uni and Srho.ts.biv the
bandwidths of the marginal densities are computed only once (this
assumes stationarity).
Added Srho.test.ts.p, the parallel version of
Srho.test.ts.
Added the url of the Biometrika paper to the docs that had not been updated.
Added the reference to the Biometrika paper in CITATION.
Minor fixes to the documentation.
All the native routines have been registered.
First CRAN version. Made the package compliant with the latest CRAN policy.
Added the url of the Biometrika paper to the docs.
Moved the package parallel from Suggests to Imports.
Changelog converted to markdown
Minor cosmetic changes to documentation files.
Added the switch “nor” to Srho,
Srho.test and related internal routines. This allows the
choice over the normalization for Srho in the case of
integer/categorical time series. Added a relevant example to the help
page of Srho. The information has been inserted in the show
method of Srho as a note.
The Fortran subroutines SS* for computing Srho for integer/categorical time series have been modified as to use a unique workhorse function SRHOBIVA.
Fortran subroutines SS,SS2,SSB have been renamed SSUNI,SSUNI2,SSUNIB for consistency.
Simplified the routines Srho.test.uni and
Srho.test.biv as to use directly the switch “stationary” to
call the associated Fortran routines.
Minor improvements to the docs of Srho and
Srho.test.
Reverted to the native Fortran dnorm for performance reasons
The Fortran subroutines have been reorganized in a module, many interfaces have been removed.
Minor modifications to Srho.test.ts.Rd
The Fortran subroutines have been cleaned for portability and to remove some obsolescent constructs.
The Fortran routines now use the C function dnorm from the R API through a C wrapper.
Updated the references in the documentation files.
Modified the f90 function MEVA (variance) for floating point precision accuracy.
Parameter eps becomes tol for consistency with
adaptIntegrate of package cubature. Docs amended
accordingly.
Paramter maxpts is passed to maxEval in
adaptIntegrate, defaults to 0 (unlimited). Docs amended
accordingly.
Paramter minpts is suppressed. Docs amended accordingly.
Rewritten the parallel routines (Srho.test.AR.p,
Trho.test.AR.p, Trho.test.SA.p) as to use the
package parallel. The package no longer depends upon Rmpi and should
work out of the box across platforms (including Windoze).
Set B = 100 in most tests.
Added a CITATION file.
Truncated the lines of several Rd files as to avoid line truncation and notes from –as-cran.
Fixed a bug where the parallel tests did not return the significant lags.
Moved the initial checks of Srho.uni.*,
Srho.biv.* at the level of Srho, those of
Srho.test.uni, Srho.test.biv.* at the level of
Srho.test, those of Srho.ts.uni,
Srho.ts.biv.* at the level of
Srho.ts.
Srho.test.iid has been renamed
Srho.test.ts for consistency.
Fixed p-values names in Srho.test.ts.
Added further examples in Srho.test.ts.Rd.
Built under R 3.0.2
Built under R 3.0.1
Modified files NAMESPACE and DESCRIPTION and the parallel routines to make the package compliant to the latest import/depends directives as specified in the manual.
Cleaned, updated and improved almost all the help files.
surrogate.AR
and surrogate.ARs.Added the computation of the bootstrap p-value for all the tests.
Corrected a buglet on names that caused a warning on all the tests.
Added the computation of the bootstrap p-value for
Srho.test. Added a p.value slot to the class
Srho.test.
Commented out line 132 of the f90 code (reset of the random seed) as it caused problems.
Modified the code of the examples in Srho and
Srho.test.
Added documentation for Srho.test.AR and
Srho.test.AR.p.
Srho.test.AR and
Srho.test.AR.p that have been modified as to use also the
smoothed sieve which is the default.Modified the routine Srho for integer/categorical
data. Now Srho is normalized on the interval [0, 1], the
maximum possible value is computed analytically and taken as a
benchmark. This makes the results of Srho comparable among
different series.
Minor improvements of the Fortran routines SRHOBIVA and SS2.
Fixed some inconsistencies on setMethod(“plot”) and improved setMethod(“show”).
Built under R 2.14.0 with rtools 2.14.
Fixed some ylab inconsistencies in
Trho.test*.
Set na.action=na.pass in acf based estimation of Srho (Srho.cor).
Added Surrogate.ARs, the smoothed sieve
bootstrap.
Added references in Surrogate.ARs.Rd and Surrogate.AR.Rd.
Added the option smoothed to Trho.test.AR*. Modified
the documentation accordingly.
removed Srho.test.SA, Srho.test.AR,
Srho.test.SA.p Srho.test.AR.p.
added Trho.test.SA.p, Trho.test.AR.p
with documentation.
added documentation for Trho.test.SA,
Trho.test.AR.
added safe.Trho in the computation of Trho.test.SA,
Trho.test.AR.
fixed bug on setMethod(“plot”): it was not possible to set lwd.
substituted the integration routine adapt with
adaptIntegrate from package “cubature”.
changed the names Srho.test.**2 in
Trho.test.** to comply with the notation of the
paper.
added Srho.test.SA2 (Experimental).
fixed NAMESPACE for “previous import show” warning.
inserted the missing test type in Srho.test.biv.
fixed several non utf8 characters in Rd files.
added Srho.test.AR2 (Experimental).
set writeout = 200 in safe.Srho.
fixed and reworked Srho.test.iid and its
documentation; included the mbb option into the
routine.
Rmpi now in the Suggests list (was in the Depends one).
added Srho.test.iid.
improved and cleaned the documentation.
modified safe.Srho to deal with the bivariate case.