README
1+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
2
3Symbolic regression (symbreg): A simple GP example with Open BEAGLE
4
5Copyright (C) 2001-2003
6by Christian Gagne <cgagne@gmail.com>
7and Marc Parizeau <parizeau@gel.ulaval.ca>
8
9+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
10
11
12Getting started
13===============
14
15 Example is compiled in binary 'symbreg'. Usage options is described by
16 executing it with command-line argument '-OBusage'. The detailed help can
17 also be obtained with argument '-OBhelp'.
18
19Objective
20=========
21
22 Find a function of one independent variable and one dependent variable, in
23 symbolic form, that fits a given sample of 20 $(x_i,y_i)$ data points,
24 where the target function is the quadratic polynomial $x^4 + x^3 + x^2 + x$.
25
26Terminal set
27============
28
29 X (the independent variable)
30 PI
31 Ephemeral constants randomly generated in $[-1,1]$
32
33Function set
34============
35
36 +
37 -
38 *
39 / (protected division)
40 SIN
41 COS
42 EXP
43 LOG (protected logarithm)
44
45Fitness cases
46=============
47
48 The given sample of 20 data points $(x_i,y_i)$, randomly chosen within
49 interval [-1,1].
50
51Fitness
52=======
53
54 $\frac{1.}{1.+RMSE}$ where RMSE is the Root Mean Square Error on the
55 fitness cases.
56
57Stopping criteria
58=================
59
60 When the evolution reaches the maximum number of generations.
61
62Reference
63=========
64
65 John R. Koza, "Genetic Programming: On the Programming of Computers by Means
66 of Natural Selection", MIT Press, 1992, pages 162-169.
67
68