Movatterモバイル変換

pcalg News

CHANGES IN VERSION 2.7-8 (2022-12-17, svn r528)

• FixidaFast() regression tests, usingall.equal()instead ofround(*) == round(*), notably for ATLAS-usingBLAS/Lapack.

CHANGES IN VERSION 2.7-7 (2022-10-04, svn r526)

• Tiny bug fix for fci.

CHANGES IN VERSION 2.7-6 (2022-04-04, svn r525)

• ‘Proper’ registration of.Call()ed C routines.

• Not usingif(class(.) = "...")

• Adapt to upcoming deprecation of C++str::binary_function()

• dagitty maskingrandomDAG().

CHANGES IN VERSION 2.7-5 (2022-02-21, svn r518)

• fixing bug.

CHANGES IN VERSION 2.7-4 (2021-11-24, svn r517)

• updated links, removed old vignette.

CHANGES IN VERSION 2.7-3 (2021-05-20, svn r516)

• adapt to R-devel's new/changed warning ofmatrix(<non-matching-length>, n,m).

CHANGES IN VERSION 2.4-2 (2016-09-22, svn r403)

• print() andsummary() methods forpc(),fci(), etc.

• newas(*, "amat") methods to get the adjacency matrixof a model, including aprint() method for "amat"s.

  ?amatType documents the different kinds of adjacency matricesinpcalg.

• New functions ......

• DeprecatedLINGAM() in favor of newlingam().

• DeprecatedallDags() in favor ofpdag2allDags().

CHANGES IN VERSION 2.2-4 (2015-07-22, svn r344)

• Bug fix insimulate(<gies>).

• Bug fix inLINGAM(., verbose=FALSE) (the default, now tested).

• Fixplot(pc(..), zvalue.lwd=TRUE), and better documentthat plot method.

• Much fastergSquareDis() andgSquareBin() for|S| in 1:4 or so.

• More test coverage, notably forpcSelect() andmcor().

• r.gauss.pardag() gets an optional argumentlabels.

CHANGES IN VERSION 2.2-3 (2015-07-12, svn r322)

• Adapt to API change inigraph package.

• More test coverage and first speed tweaks.

CHANGES IN VERSION 2.2-2 (2015-06-24, svn r304)

• Functiongac for testing if a set of nodes satisfiesthe Generalized Adjustment Criterion (GAC). GAC is a sufficient andnecessary criterion for estimating causal effects via covariateadjustment in DAGs, CPDAGs, MAGs and PAGs.

CHANGES IN VERSION 2.2-0

• FunctionrandDAG for generating random directed acyclicgraphs (DAGs) with fixed expected number of neighbours. Severaldifferent methods are provided, each intentionally biased towardscertain properties. The methods are based on the analogue functionsin theigraph package.

• FunctionunifDAG andunifDAG.approx for uniformsampling of a labelled directed acyclic graph (DAG) with combinatorialenumeration.

CHANGES IN VERSION 2.1-0

• FunctionLINGAM for estimating the structure of Linearnon-Gaussian Additive Models

CHANGES IN VERSION 2.0-4

• FunctionfciPlus for a fast (but accurate) version offci (note that the already existing functionrfci is afastapproximation offci)

CHANGES IN VERSION 2.0-1

• Functionsgds,gies,ges for score based learning

• Functionbackdoor for generalized backdoor criterion

• Bugfix in functionwgtMatrix; this affects functionstrueCov andrmvDAG; to reproduce old version, use option 'back.compatible =TRUE'

• skeleton,pc,fci andrfci nowhave 'alpha' as third argument,then 'labels, p' where 'p' is optional; namely, derived from 'labels'when these are specified.

• The default for extending the skeleton to a CPDAG is nowudag2pdagRelaxed, which is completely deterministic (wasudag2pdagbefore, which can use random orientations if sampling errors occur).