DEvAl README
------------

1) Building.

    If you are reading this then you probably have the code (not true anymore).
In order to build said code, you should be able to do the following
(in the top level directory):

    ./buildprep
    make

This will build all of the code and place the binaries (both the executables
and libdeval.so*) in bin/. Assuming no build errors (this code was tested
on SunOS 5 and Fedora Linux, but it should build for any Linux distro; other 
UNIX OSes may work as long as they support the GNU *rand48_r() fuctions) you 
should be ready to run some code.

2) Running.

    In bin/ `root_finder' and `mixture' both implement an evolutionary
algorithm via the libdeval.so library. `root_finder' does as one would expect:
it finds roots of polynomials (or it finds minimas if there are no roots).
`mixture' is more complex. This program maximizes MLE estimates for data 
following multinomial normal distributions.
    There are two other programs for generating distributions of data. `norm.m'
and `norm_mixture.m' can be used to generate data with either 1 or 2 normals.
If more normals are desired then the results from `norm.m' can be concatentated
together. These two commands require octave to be installed in /usr/bin. These
commands should work with matlab as well, though; just change the #! to point
to matlab's binary instead. Some example commands:

    $ cd deval/bin
    $ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:.
    $ ./norm_mixture 5 2 -5 2 .5 10000 >mix.txt # Generate a 2 norm mixture.
    # Make a norm file for the mixture program. For details on how to do that
    # see the comment at the top of that file. Lets say the norms file is
    # called norms.txt
    $ ./mixture --norms norms.txt --data mix.txt

Some sample output:

# Algorithm parameters:
#   Population size:      100
#   Thread count:         1
#   Maximum iterations:   100
#   Gene dispersal:       0.000000
#   Reproduction rate:    0.250000
#   Breed fitness:        0.250000
#   Check for converge:   no
#   Data file:            mix.txt
#   Normal distributions: norms.txt
# Read 2 normal distributions.
#   norm1: mean=[-10.0000,10.0000] stddev=[0.5000,5.0000] var=[ 0.0100 0.0050 ]
#   norm2: mean=[-10.0000,10.0000] stddev=[0.5000,5.0000] var=[ 0.0100 0.0050 ]
# Read 10000 data samples.
# Initializing mixture allocation buckets.
# Initializing param allocation buckets.
# Gene pool made, solutions inited, running...
run time: 4935 ms

    The `root_finder' works much the same way only you don't need to make any
extra files. Just specify the parameters for a polynomial. The parameters are
all described in the comments in the src/algos/root_finder.c file.