readme.rpn

标准的GP源代码,由Andy Singleton维护

*************************************************************************
GPDATA Version 3 release Easter 1997 W.B.Langdon for the Reverse
Polish Notation (rpn) calculator problem as described in GP-96
$Revision: 1.4 $.

NB The source code use the term "calc" in some places rather than rpn.
   The version using stack primitives is denoted by the conditional
   compilation symbol KARP2

*************************************************************************
Instructions
*************************************************************************

To Make GPcalc2: (ie use stack primitives)
	Create a symbolic link from calc2.h to prob.h
	(eg ln -s calc2.h prob.h)

	If gcc is available, make with calc2.make 
	Eg make -f calc2.make

	If running on an alpha, make with calc2.make.alpha

To run GPcalc2:
	copy rpn.ini to gp.ini and gp.test.rpn to gp.test Then run GPcalc2

	GPcalc2 takes optionals command line arguments, these allow the
	user to override the parameter settings in gp.ini Eg

		GPcalc2 pPopSeed:102

	Evolved solutions can be loaded and their execution traced by
	creating a file gp.load containing the program.

To Make GPcalc: (ie use indexed memory primitives)
	Create a symbolic link from calc.h to prob.h
	(eg ln -s calc.h prob.h)

	If gcc is available, make with calc.make 
	Eg make -f calc.make

	If running on an alpha, make with calc.make.alpha

To run GPcalc:
	copy rpn.ini to gp.ini and gp.test.calc to gp.test

	GPcalc takes optionals command line arguments, these allow the
	user to override the parameter settings in gp.ini 
	It is neccessary to set pMaxExpr to atleast 350 eg

		GPcalc pMaxExpr:350

	Hand coded solutions can be loaded and their execution traced by
	creating a file gp.load containing the program.

BUGS:
	Sometimes with sparc-sun-solaris2.4 gcc version 2.6.3 fails
        with "Bus error" at end of creating initial population
        Ok with gcc version 2.6.3 on sparc-sun-sunos4.1.3 and 
        DEC C++ V5.3-004 on Digital UNIX (Alpha)

	A number of bugs have been fixed since the Reverse Polish
	Calculator problem was first solved. However the code in this
	kit reproduces the reported behaviour in GP-96. Symbols
	defined in  calc%.h (eg ORIGINAL_DEME) are used in some source
	files to disable some bug fixes. You may like to revise the
	sources, eg to use new symbols. 

	Behavour is not _identical_ to earlier version but nontheless
	solutions to the RPN problem are evolved with GPcalc2