28/03/2021
Initial implementation.
This package *REPLACES* the package "hse".  Effectively
this is a name-change only.  A referee of a paper that
I wrote, based around this package, objected to the name
"hse" ("hope springs eternal") saying that readers might
find this name too frivilous.

Changelogs for the "hse" predecessors of this package may
be found in the CRAN archives.

Version 0.0-1.

29/03/2021

Added the data set "hrsRcePred" and documented it.

Added the function mleBb() and documented it.

Incremented the version number.
Version 0.0-2

30/03/2021

Added the data set "visRecog" and documented it.

Changed the name of the success probability parameter in mleBB() from "p"
to "m" to be consistent with the notation used in rmutil.

Added names to the vector returned by mleBB().

Incremented the version number.
Version 0.0-3

04/04/2021

Got rid of mleBb(); going with just using the db distribution.

Changed the default value of maxit in mleDb() from 100 to 1000.

Added a maxit argument to simulate.mleDb().

Added maxit arguments to gof().

Modified ddb() to take a "log" argument (defaulting to FALSE).

06/04/2021

Added a "maxit" argument to mcCovMat().

Added the possibility that a warning may be issued (rather
than an error being thrown) in mleDb() or mleBb().  Which
of the two happens is controlled by options()[["maxitErrorOrWarn"]]
which is set to "error" at startup.

Added the function set.eow() to allow the user to switch between
the two options.

Added the function get.eow() to query this option.

07/04/2021

Added functions gradDb() and hessDb() and
revised the grad() and hess() functions from gradHess.R to make use
of them.

Made some adjustments to mleDb() (eliminated an unnecessary
complication) in respect of the way the hessian-cum-covariance
matrix is calculated.

10/04/2021

Fixed a glitche in mleDb() in respect of the names of
the gpar argument used by gfun.  This was mucking everying up
to an incredible extent!

11/04/2021

Added arguments "covmat" and "useGinv" to mleDb().

Amended mcCovMat to call mleDb() with covmat set equal to FALSE.

Corrected sign errors in aHess() and nHess().

Added dimnames attribute to the value returned by nHess().

Added dimnames attribute to the value returned by hessDb().

12/04/2021

Made various other (fairly minor) corrections to the code and to
the help files, fixing errors that were noticed in the course of
fiddling about with them.

13/04/2021

Added a simulate.Vpars() method for simulate.

Added a function makeVpars() to create "Vpars" objects.

Added an example using makeVpars() simulate.Vars() to the help
for ndata.

Incremented the version number.
Version 0.0-4.

23/04/2021

Fixed a stupid glitch w.r.t. setting of seeds in simulate.Dbdpars()
and simulate.mleDb().

Incremented the version number.
Version 0.0-5.

26/04/2021

Amended mcCovMat() so that it works with objects of class "Dbdpars".

Amended finfo() so that its arguments are the same as (in the same order
as) those of makeDbdpars()

Incremented the version number.
Version 0.0-6.

29/04/2021
Uploaded to CRAN.

30/04/2021

Made changes/corrections at the behest of CRAN (explicitly Gregor Seyer).

* wrapped all examples which use data from hmm.discnp in
    if(requireNamespace("hmm.discnp")) { ... }

* changed instances of "if(require(...))" to "if(requireNamespace("..."))".

* adjusted the value returned by logLik.mleDb() to have class
  "logLik" and an attribute "df" (equal to 2).  Made a corresponding
  adjustment to the help file for logLik.mldDb().

Incremented the version number.
Version 0.0-7.

Made a correction (re-set par() values if these are changed in
examples!) which I had neglected to make.

Incremented the version number.
Version 0.0-8.

Uploaded to CRAN

01/05/2021

Fixed a typo "namep-change" (!!!) in this ChangeLog.

Removed spurious links to meDb() (which is undocumented!) from
some help files.

25/05/2021

Removed another spurious link to meDb() (from the help for
mleDb()).

Incremented the version number.
Version 0.0-9.

11/07/2021
At the suggestion of a referee of the accompanying paper, added a
function exactMeBb(), and documented it.

Incremented the version number.
Version 0.0-10.

12/07/2021
Made the zeta argument of exactMeDb() default to FALSE (matching
up with mleDb()).

Incremented the version number.
Version 0.0-11.

12/07/2021
Adjusted the help file eow.Rd to make reference to exactMeBb() as
well as mleDb().  Added "eow" to the aliases.  Fixed a glitch
(start of a sentence had been omitted).  Added an example of
set.eow() --- no effect when options()$eow is "error".

Incremented the version number.
Version 0.0-12.

26/07/2021
Added tools for dealing with the beta-binomial distribution.
These were original created in the "Verdis" ("versatile discrete
distributions") package, into which the dbd package morphed for a
time.  I decided to eschew use of the beta-binomial distribution, but
have now decided to reinstate it, but to have it play a relatively
minor with the db distribution holding prominence.

* added function makeBbdpars() and method simulate.Bbdpars()

* made mcCovMat() generic, with methods
        mcCovMat.mleDb()
        mcCovMat.mlebb()
        mcCovMat.Dbdpars()
        mcCovMat.Bbdpars()
        mcCovMat.default()
  The default method simply throws an error.

* made a number of minor adjustments and corrections to the code
  and the help.

* fixed a stupid glitch in the simulate methods, in respect of the
  "seed' argument, which could have resulted in all simulated data
  sets being identical.

Incremented the version number.
Version 0.0-13.

27/07/2021
Fixed a glitch pertaining to the extraction of "zeta" from the
"object" argument of mcCovMat (the code had been trying to find
zeta as an attibute of the object, whereas, when the object is
of class "Dbdpars" or "Bbdpars", zeta is a *component* of the list
consituting the object).  Also adjusted some 'list$xxx' constructs
to have the form 'list[["xxx"]' --- safer!

Incremented the version number.
Version 0.0-14.

29/07/2021
Replaced the functions finfoDb() and finfoB() by a single
function finfo() the first argument of which is "distr" that
specifies which distribution is to be considered.

Incremented the version number.
Version 0.0-15.

15/08/2021
Added the argument "verb" (defaulting to FALSE) to the gof
methods.  Setting it to TRUE causes "progress reports" to
be issued during the course of the simulations that are conducted
when MC is TRUE.

Added the argument "seed" to the gof methods, for use when MC is
TRUE.  If MC is TRUE the value of the "seed" argument, or the
random value chosen if "seed" is not supplied, is made an attribute
of the returned value.

Incremented the version number.
Version 0.0-16.

16/08/2021
Adjusted the "Parsonnet scores" example in the help for gof().
Added references to the Wittenberg paper and to the spcadjust
package.

Incremented the version number.
Version 0.0-17.

17/08/2021
Changed the plotting functions plotDb(), plotBb(), plot.mleDb()
and plot.mleBb() to return values (invisibly).  These values consist
of the data that were plotted.

Eliminated some redundant calculations in plot.mleDb() and plot.mleBb()
(these calculations also being done by plotDb() and plotBb() respectively.

Added an "incr" argument to plot.mleDb() and plot.mleBb().  This specifies
the gap between the vertical lines representing the fitted probabilities
and those representing the observed proportions.  Previously the value
of incr was hard-wired ... and then ignored (i.e. re-hard-wired in the
call to lines()).

Incremented the version number.
Version 0.0-18.

17/08/2021
Added a "plot" (logical scalar) argument to the plotting functions
plotDb(), plotBb(), plot.mleDb() and plot.mleBb().  Determines
whether a plot is actually produced (rather than having the function
simply return a value).

Incremented the version number.
Version 0.0-19.

17/08/2021
Changed the default value of "incr" to depend on "size"
(plot.mleBb()) or "ntop" (plot.mleDb()).  It now defaults to 0.1
for values of these quantities less than 20 and to 0.5 otherwise.

Incremented the version number.
Version 0.0-20.

17/08/2021
Wrapped some code in \dontrun{ } in makeBbdpars.Rd and hrsRcePred.Rd
because it was taking too long for CRAN.

Changed http to https in mleBb.Rd at the behest of CRAN.

Incremented the version number.
Version 0.0-21.

19/08/2021
Corrected a couple of major bugs in the gof() functions.
(1) Setting of seed for the simulations was incorrectly handled.
(2) Starting values used for fitting models to simulated data sets
should have been set (to the parameters of the model being simulated)
and had not been so set.

Incremented the version number.
Version 0.0-22.
