Sep 14th 2005 build

[Core]

1). Added an automtic MPI likelihood function parallelizer,
by allocateing blocks of site patterns to slave nodes. Rate matrix
exponentiation will be duplicated among nodes (should be a minor expense
in most cases). Works best for long and large (100+ nucleotide sequences,
20+ codon sequences) alignments.

Can be triggered at run time by setting the AUTO_PARALLELIZE_OPTIMIZE
constant to 1.

This allows a batch file to switch between MPI processing of multiple concurrent
jobs and MPI optimization of a single likelihood function.

[GUI] (All platforms)

1). "Input partition" menu item in data panel window now allows
partiton strings to be either 0 or 1 based based on a checkbox in the
dialog.

2). Right clicking (or ctrl-clicking on the Mac) in the data panel window (sequence panel) will
display this new options:

   -- If a single column is selected - sequences can be sorted alphabetically on the character
           in that column.
          
3). Choosing 'Data->Invert Selection' menu option in the data panel window will if nothing is
selected in the data viewer window but at least one partition is selected in the partition
table, will select all sites not in that partition.

[Batch Language]

1). Signifcantly expanded the functionality of matrix indexing command. See the
entry for "[]" in the online reference for complete details. Can now: access matrix
rows (e.g. M[R][-1], to get row R), columns (M[-1][C] to get column C), use a boolean
stencil to extract a subset of elements, extract a rectagular submatrix (M[{{L,T}}][{{R,B}}]
to get rows L through R over columns T through B). See examples in
Examples/BatchLanguage/MatrixIndexing.bf

2). Added a syntax for performing arbitraty element-wise matrix operations (see examples in the
file in Examples/BatchLanguage/MatrixIndexing.bf)

3). Can now use the '+' operation to add a number to every element of a matrix (e.g. M + 1).

4). There is a backlog of over 100 additions over the last few months which I am in the process of cataloging...