Darren Lewis
CS426/BMI226/EE392K

Using LIL-GP for Genetic Programming

Downloading and installing the software

1. Go to your cs426 directory on Leland AFS:

   cd cs426

2. Copy the files to your space:

   cp -r /afs/ir/class/cs426/lilgp .

   This will create a directory called lilgp in your cs426 directory.
3. Change to your new lilgp directory:

   cd lilgp

   You should have one file: lilgp.tar
4. Uncompress with the following command:

   tar -xvf lilgp.tar

   You should now have a directory called lilgp1.1. Under this directory you only need to deal with the following subdirectories:
   
   1.1: This directory contains all the code and samples.
   htmlMan: This directory contains all the documentation.
   
   Don’t worry about the others.

Where to start

Running the samples

1. Compile and build with the command make.
2. Run with the command gp -f input.file.

Implementing a problem with LIL-GP

1. Create a tableau for the problem. If any questions arise about terminology, see Genetic Programming (Koza 1992) Chapters 6 and 7, because LIL-GP was modeled after the format of genetic programming described in this book. For those of you who are wondering, that's the fat textbook.
2. Write the code. You will need to provide five files:
   1. app.h - This file defines the global data structure used to pass information back and forth between files.
   2. app.c - Defines all the user callback functions – initialization, fitness evaluation, etc.
   3. appdef.h - This file contains two #defines: MAX_ARGS, and DATATYPE.
      MAX_ARGS: Maximum number of arguments your functions will take.
      DATATYPE: Type that all of your functions will return (ex: double, int, etc.).
   4. function.h - Prototypes for the functions that your genetic program will use.
   5. function.c - Implementation of the functions whose prototypes are in function.h.
3. Create a parameter file for the run. For the sample applications that are provided, the parameter file is named input.file. You can modify these files to suit your problems. For more information about the extensive array of parameters that can be set, please refer to the documentation.
4. Build and compile the code. Just type make from the directory where your files are. If there is a compile or link error, it will alert you at this step. Fix these errors before continuing.
5. Run the code. Run your code simply by typing gp -f input.file where input.file is the parameter file created in step 3.
6. Examine the output files. LIL-GP generates a number of interesting output files for each run. They are:
   • .sys – general info about the run
   • .gen – stats on tree size and depth
   • .prg – stats on fitness and hits
   • .bst – info about current best individual
   • .his – history of the .bst file
   • .stt – unreadable version of all stats
You will want the .bst and .his files.

Hints

• Once again, there is not a lot of code to write for this assignment. In fact, two of the sample applications provided with LIL-GP are symbolic regression of a simple function, and symbolic regression of a boolean-11 multiplexer. For problem 1, simply modify the provided files in the "regression" directory. Not very many modifications will need to be made to solve the homework problem. For problem 2, just modify the "multiplexer" code to work for even-3-parity. You will only need to add a few functions and change the fitness measure. For problem 3, simply modify your working code from problem 2. Follow the "lawnmower" example for defining function sets for automatically defined functions.
• Don't change any code in the "kernel" directory.
• We're serious about not having to write much code - just modify the existing code. There is absolutely no need to rewrite all the callback functions in app.c. If it ain't broke, don't fix it.
• Don't change any code in the "kernel" directory.
• If you get stuck, read the documentation. It is honestly helpful, well-indexed, and very clear.

Good luck!