Re: Particle Swarm Optimization

William James <w_a_x_man@[EMAIL PROTECTED]
> writes Re: Particle Swarm Optimization

> On Feb 2, 4:19 am, m...@[EMAIL PROTECTED]
 (Marcel Hendrix) wrote:

>> \ The target function to MAXIMIZE
>> : result ( F: x y -- z )
>>         FLOCALS| y x |
>>         1 +TO funcs
>>         x -2e F- FSQR  y -2e F- FSQR F+ FSQRT FNEGATE 3e   F/ FEXP   5e
F*
>>         x  2e F- FSQR  y  2e F- FSQR F+ FSQRT FNEGATE 2e   F/ FEXP  15e
F* F+
>>         x  2e F- FSQR  y -2e F- FSQR F+ FSQRT FNEGATE 2e   F/ FEXP  12e
F* F+
>>         x -2e F- FSQR  y  2e F- FSQR F+ FSQRT FNEGATE 2e   F/ FEXP  -4e
F* F+ ;

> // Free Pascal

> function result( x, y : real ): real;
>  function f( a, b, c, d : real ): real; inline;
>  begin
>    f := exp( -sqrt( sqr(x - a ) + sqr( y - b ) ) / c ) * d
>  end;
> begin
>  funcs += 1;
>  exit(
>    f( -2.0, -2.0, 3.0,  5.0 ) +
>    f(  2.0,  2.0, 2.0, 15.0 ) +
>    f(  2.0, -2.0, 2.0, 12.0 ) +
>    f( -2.0,  2.0, 2.0, -4.0 )
>  )
> end;

That hurts :-)

I am very pleased with your complaint about my use of unfactored 
Forth. Next time, I'll be motivated to do better.

\ The target function to MAXIMIZE
: peak ( F: d c y b x a -- r ) 
	      F- FSQR  
	-FROT F- FSQR F+ FSQRT FNEGATE  
	revF/ FEXP F* ; ( auto inlined )

: result ( F: x y -- z )
	FLOCALS| y x |
        1 +TO funcs
	 5e 3e  y -2e  x -2e  peak
        15e 2e  y  2e  x  2e  peak F+
	12e 2e  y -2e  x  2e  peak F+
	-4e 2e  y  2e  x -2e  peak F+ ;

-marcel