Raw File
Tip revision: 1cc05f63437c1a519a5bdc24b3cc669980d81d48 authored by Charles J. Geyer on 13 May 2019, 18:20 UTC
version 1.0-3
Tip revision: 1cc05f6
 0.2    first version to CRAN
 0.2-1  fixed bugs in docs found by scripts on CRAN (R-devel check)
 0.3    added data/echinacea.txt
        added test in aster.R
        added package vignette
        added examples on all help pages to be used by users
 0.3-1  changed variable "site" to "pop" in echinacea.txt
        corresponding changes in many other files
        \usepackage[utf8]{inputenc} in tutor.Rnw
        many small changes in wording in tutor.Rnw
 0.4    made trust the default optimizer
 0.4-1  removed the timing.pdf vignette, fixed xrefs to trust in Rd files
 0.4-2  added import(trust) to NAMESPACE
 0.5    added 4-th family "two.trunc.poisson"
        actually implemented code for k-truncated Poisson
        put document describing k-truncated Poisson in inst/doc
        new R function rktp that simulates k-truncated Poisson
        cannot implement k-truncated Poisson for arbitrary family
              until families get (non-exponential family) parameters,
              such as k for k-truncated Poisson, r for negative binomial,
              offset for Poisson, n for binomial(n, p), sigma for
              normal-location.  All of that goes on the to-do list,
              along with multivariate families (as per revised version of
              the paper), especially multinomial and normal-location-scale.
        fixed (?!) or at least worked around nasty validation bug in
 0.5-1  fixed bug in tests/tutor.R caused by bug fix in R.  optim was ignoring
            its maxit argument when method == "CG".  When it stopped ignoring
            it, it failed to converge, because it takes more than 1000
 0.5-2  fixed bug in inst/doc/trunc.Rnw -- must be vignette to pass R CMD check
 0.6    new superfamily system: exponential families with non-exponential
            family hyperparameters
        canonical affine models
 0.6-1  negative binomial distribution
 0.6-2  truncated negative binomial distribution
        fixed bug in 0.6 and 0.6-1 that made them useless with models
            that did not have unrestricted canonical parameter space:
            mlogl could return NaN rather than Inf (which is what it
            is supposed to do) when evaluating at point outside the
            canonical space.  Now hopefully fixed with new code
            in astunco.c and astcond.c.  (TODO: need more tests with
            restricted parameter space.  foobar.R is not enough.  Note
            0.6 and 0.6-1 were not made public, so no one bug CJG saw
            this bug.  Still it does show the tests directory is deficient.)
 0.7    added normal-location families; see ?fam.normal.location
 0.7-1  fixed incorrect signature in as.character.astfam
 0.7-2  new data sets: echin2 chamae2 chamae aphid
        added warning (varb must be factor) to vignette
 0.7-3  unmatched braces in .Rd files chamae2 chamae
 0.7-4  fixed bug in predict.aster.formula documentation.
            amat is three-way array just as for predict.aster
            example already required this.
        removed tests/tutor.R   took too long and tested nothing specific
 0.7-5  added newcoef argument to predict.aster and predict.aster.formula
 0.7-6  fix doc bug found by Rd version 2 parser
 0.7-7  add sim.rda dataset for TR 669 and 674  Also add references to aster.Rd
 0.8    add radish.rda and oats.rda datasets and pseudo function for TR 676.
            Also fixed other references removing TeX ~.  Also fixed
            "make clean" in inst/doc Makefile.
 0.8-1  added pickle, penmlogl, and transform functions and put on
            github in private repo
 0.8-2  change name of transform to astertransform to avoid conflict with
 0.8-3  mods to pickle for pseudo-REML (an idea later abandoned).  Also ...
            arg added to pickle.
 0.8-4  revised pickle to return Inf rather than error when negative
            variance comp
        added tol argument to pickle, deleted tol argument from penmlogl
            Now pickle works with optim method "Nelder-Mead"
        added type argument to pickle
        added cache argument to pickle (but this was later revised)
        renamed type argument of pickle to rescale
        added tests/penmlogl.R
 0.8-5  changes in 0.8-4 now working
        cache argument removed from penmlogl
        revisions to pickle (penmlogl called by
        more revisions to pickle determinant calculated as product of
            eigenvalues using log1p for accuracy
        tests/pickle.R added (penmlogl and pickle functions now have
            full tests)
 0.8-6  bug found by Frank Shaw fixed.  pickle was using global variable nfix
            bug fixed and tested (no global variables now used)
 0.8-7  deleted COPYING file as per instructions from R CMD check
        added cache argument to pickle (this version now works)
        tests/cache.R added to test that caching actually works
 0.8-8  revise oats dataset for new analysis
        delete R/pseudo.R man/pseudo.Rd tests/pseudo.R tests/
        (pseudo idea is now gone)

 0.8-9  deleted fterm = 0 from argument of trust call inside
        pickle.  If users want that behavior, they can specify
        fterm = 0 as argument to optim, which will be pased to pickle.
        And do we really need such finickiness to be the default?

 0.8-10 Add function reaster to hide pickle and penmlogl from ordinary users
        add function penmlogl2 (not exported) for evaluation of hessian

 0.8-11 add functions pickle1, pickle2, and pickle3 to do reaster (not for
        naive users), called by reaster.  Add function summary.reaster.
 0.8-12 add functions quickle and newpickle (the latter not used).  Rewrite
        reaster to use original (not square root) parameterization for
        variance components in computing standard errors.  Totally rewrite
        design doc (re.pdf) to reflect such.

 0.8-13 skipped in honor of Stevie Wonder

 0.8-14 changed examples in aster, predict.aster, anova.aster docs

 0.8-15 use paste0 in R functions
        added chamae3 dataset
        fixed names of components of result of reaster
        fixed bug in penmlogl2 (scale component of result wrong length)

 0.8-16 no DON'T use paste0 in R functions (that requires R-2.15.0)
        changed behavior of reaster.summary when calculated Fisher information
            matrix is not positive definite.  Now not an error but all std.
            err. set to Inf
        fixed test penmlogl.R to correspond to bugfix in previous version

 0.8-17 changed some | to || in reaster.R
        fixed many, many errors [ without drop = FALSE
        fixed several errors diag(x) when x could be length 1
        added tests/reaster1.R to test just one variance component
        added options(digits=4) to several tests in hope of fewer differences


 0.8-19 added tech rept ref to reaster.Rd

 0.8-20 added options(na.action = to aster and reaster

 0.8-21 changed license to "MIT + file LICENSE" as per CRAN instructions

 0.8-22 bug fix.  summary.reaster crashed with inexplicable error message
        when all variance component estimates were zero

 0.8-23 added \dontrun to some examples per instructions from Brian Ripley

 0.8-24 fixed bug in penmlogl2 (needed drop = FALSE) when only one random
        added deviance component to result of reaster function (can now
            so some "anova" tests
        made aster and reaster objects subclass of class "asterOrReaster"
            so one "anova" function can do both and can compare models
            with and without random effects

 0.8-25 fixed inst/doc/Makefile as per instructions from Uwe Ligges
            and "Writing R Extensions"
        No.  That does not work either.  Just removed Makefile

 0.8-26 added .Rbuildignore as per instructions from Brian Ripley

 0.8-27 moved vignettes to vignettes directory as per instructions from
            Uwe Ligges

 0.8-28 bug fix in summary.reaster

 0.8-29 feature change to predict.aster: finally document gradient component
            of output and add new argument "gradient" to request gradent
            even when not computing standard errors

 0.8-30 removed test for convergence of Nelder-Mead step in reaster.default.
            use whether or not converged.  We are going to polish anyway.
            Only do stop if error.
        put warning about negative binomial and random effects in
            documentation for reaster and also warning in reaster function.

 0.8-31 explicit import from non-base packages, as required by R-3.3.0

 0.9    allow C functions that evaluate aster unconditional log likelihood
        to be called from C code in other packages.

        remove styles from C function registration.  Deprecated in R-3.3.3.

        lots of changes to restrict visibility as per Sections 5.4 and 6.15
        of Writing R Extensions

 0.9.1  Minor changes made by Brian D. Ripley to CRAN package.  A bunch changes made by Brian D. Ripley to CRAN package, and
          package is orphaned.

 1.0    package unorphaned.

        lots of fixes to computer arithmetic for zero truncated poisson,
        both bugs (returning NaN when can do better) and inaccuracies.

        fixed check for data validity for k-truncated distributions
        check was if (xpred == 0) x == 0 else x > k
        check is  if (xpred == 0) x == 0 else x >= xpred * (k + 1)

        new argument is.always.parameter added to aster method of
        R generic function predict.  Computes conditional mean value
        parameters using the correct definition.

 1.0-1  Changed http: to https: in cran URLs

 1.0-2  Added dontrun to help for asterOrReaster method of R generic
        function anova (because CRAN computers are slow)

back to top