The following programs run a genetic algorithm (GA) inversion.  
Both examples are for a volcanic source inversions.  Three c programs 
are necessary for the GA - inversion.c, fit.c, ran1.c.  Inversion.c 
contains the main genetic algorithm; fit.c will contain the forward 
model to which you want to fit the data; ran1.c is a random number 
generator.  In the header to inversion.c you must set the number of 
parameters, the population size, and the mutation and crossover rates.  In fit.c you set the number of generations, the number of each type 
of data point, and the number of each type of volcanic source you 
want to invert for.

Two examples of forward model applications are given here.  In both 
cases, fit.c compares the actual data (contained in motion.in, fit1.
in, and error.in) to those values computed by the model and computes 
a fitness. In the first example, fit_c.tgz, fit.c itself computes the 
modeled values.  In the second, fit_grav.tgz, fit.c uses a system call 
to invoke a separate executable, gravw1, originally written as a 
fortran program.  

Again, the data files motion.in, fit1.in, and error.in should be 
included in the same directory as the GA.  motion.in contains a 
listing of the measured data; error.in contains a listing of the 
standard deviations for each data point as listed in motion.in; and 
fit1.in contains, on the first line, the number of each type of data 
and the location of each data point, again in the same order as in 
motion.in and error.in.  The last file that must be in the same 
directory is gadata.txt.  This contains the range of possible values, 
in pairs of maximum and minimum, for the parameters to be inverted 
for, in the same order as they are listed in the main program.  For 
example, in the elastic-gravitational model there are 6 parameters - x 
and y location, depth, pressure, radius, and mass.  

The example fit_c.tgz works as follows:

Unpack fit_c.tgz in a separate directory.
Set the parameters in inversion.c and fit.c
Compile the main program as follows:  gcc -o inversion -lm inversion.c
Make sure the input files are in the same directory.
Call inversion with a negative seed, as discussed below.

The example fit_grav.zip works slightly differently:
Unpack fit_grav.zip in another directory (many of the file names are 
the same, do not overwrite the other example!!!)
The fortran program, Gravw1.f, runs the forward model for the gravity 
data. 
Compile the fortran program as follows:  f77 -o gravw1 Gravw1.f
Again, compile the main program as follows:  gcc -o inversion -lm 
inversion.c

Output, in both cases, is generated in galog.txt. It outputs the best 
fitness, the average and standard deviation for each generation.  You 
can change output format in the report subroutine.

To run the program, you need a random negative seed.  Call it using

inversion "seed"

Again, seed must be a negative number!    

A longer description of the genetic algorithm, complete with flow 
chart, is contained in the following references:

Tiampo, K.F., Rundle, J.B., Fernandez, J., and Langbein, J.O.  
Spherical and ellipsoidal volcanic sources at Long Valley caldera, 
California, using a genetic algorithm technique, J. Volc. and 
Geotherm. Res., 102, 2000.
Tiampo, K.F., Fernández, J., Charco, M. Jentzsch, G., and Rundle, 
J.B. New results at Mayon, Philippines, from a joint inversion of 
gravity and deformation measurements, PAGEOPH, 61, 2004.

Good luck, Kristy (ktiampo@uwo.ca)