null_complement now treats a numeric m as a matrix with one column in all cases.
removed a volatile test causing “Additional issues” on CRAN.
removed sim_mcseeds(), it had been superceded by sim_chains() for some time.
doi for Boshnakov (2002) was pointing erroneously to Boshnakov (2007).
numerous small code fixes and documentation edits.
another tunning of DESCRIPTION.
first CRAN version
added examples to user facing functions/classes which didn’t have them.
added references in DESCRIPTION per the CRAN team request.
some documentation clean-up.
moved ‘methods’ and ‘Matrix’ from Depends: to Imports:.
created Org directory; now some files are generated from org sources. Names of R files generated from Org sources start with “auto_” to distinguish from the rest.
renamed mc.0chain.transf() to reduce_chains_simple() - the name was misleading since the algorithm is valid for any eigenvalues and is not specific to mc-matrices.
renamed mc.0chain.triang() to mc_chains_triangulate() - works for any eigenvalue, not only zero.
other renaming.
completed (almost?) the handling of 0chains.
new function Jordan_submatrix.
renamed smc2co to smc_chains.
chains_to_list converts a similarity matrix to list of chains. (mc_chain_to_list still exists but is not mc specific and will be removed).
Streamlining and bug fixes of code for 0chains, non0 chains and related functions. Some arguments were removed. The low level functions dealing with individual mc-chains should not consider the case mo.col < mo - it should be dealt with at the level of functions dealing with all (or a number of chains). The work on this is not finished yet.
now generation of mc-matrices covers all cases involving zero eigenvalues.
introduced class SmallMultiCompanion for complete treatment of the case mo < mo.col.
additional clean-up.
defined S3 method for as.matrix, as.matrix.MultiCompanion, for contexts where as() doesn’t see the S4 method.
defined explicit methods for matrix multiplication of “MultiCompanion” and “matrix”. For some reason an error started to popup (change in package methods or Matrix?).
var2mf was still buggy. Changed the return value to be more consistent.
var2mf now has an argument, “perm”, for specifications of the order of the variables when treating them as seasons.
corrected a bug in mcSpec.
mCfromfilter now works also when the order of the filter is smaller than the periodicity.
bugfix: mfVSform gave wrong value for Phi0inv when form = "I".
mc_ev and the subspace parameters; now all tests from version 0.2-5 are passed.mc_ev now works and is called by mC.ss from package pcts. Partial implementation of subspace parameters.added a keyword “internal” to a number of functions which are rarely used or essentially internal, so that their documentation does not appear in the pdf manual. The documentation is still there and can be viewed with the help command.
continuing consolidaation from 0.2-4, renamed more functions the old names will be valid for some time.
The index in mcompanion-package.Rd now lists only selected functions and does this by topic (needs clean up though).
mc.0chain.structObsolete is now obsolete.
introduced mcev_core(), seems to work but will be developed further and is not used elsewhere yet.
moved sim_real and similar to gbutils.
renamed a number of functions, hopefully with more user friendly names; the old names will be valid for some time.
mC.gen.evecs gets new argument mo.col, the code adapted to handle it properly.
renamed and rewrote mc.chain.transf to mc.chain.scale.
mC.gen.evec and mC.gen.gevec now give error when what.co = "top" for a zero eigenvalue
the functions for Jordan chains corresponding to zero eigenvalues were debugged and somewhat cleaned up.
new slots for mcSpec - mc.col and F0bot.
updated rblockmult() to work with non-square second argument.
rename mC.sim.r to sim_real; mC.sim.c to sim_complex; mCsim.eigval to sim_numbers.
bug fixes and improvements to mcSpec.
renamed mc0chains.struct to mc.0chain.struct, mCchain.struct0 to mc.0chain.structObsolete, mCchain.triang0 to mc.0chain.triang.
other changes.
Import “gbutils” instead of “pad” (the latter is now part of gbutils).
Small corrections to the documentation files.