Re: Particle Swarm Optimization

John Doty <jpd@[EMAIL PROTECTED]
> writes Re: Particle Swarm
Optimization
[..]
> Cool. I've done similar but rather less sophisticated things in 
> Mathematica using a gnetlist back end I wrote to translate gEDA 
> schematics into Mathematica equations. Take a look at:
[..]
> I don't suppose it would be hard to modify the code (Guile) to feed a 
> different math package. My code is now part of the regular gEDA 
> distribution, all GPL. It's nice to be able to draw the circuit rather 
> than code the equations!

The netlist for a switched circuit does not tell the whole story.
In principle, an hierarchically structured netlist for each valid 
combination of the switches (including diodes) is needed. From these
netlists state-space matrices are build. Together with some Petri-net
massaged constraint equations, finally the network can be solved
with exponential matrix manipulations. Unfortunately, the constraints 
are numerical, not symbolical. One of my famous Forth toys can do all
of this (from an annotated netlist), but I didn't use it (yet) in this 
instance.

A normal linear circuit is a special case of a switched circuit, so I
can handle it too. The graphic output is currently done through
Matlab, but I guess gnuplot will be used in the future (when I have
time, and when gnuplot handles TeX labeling).

-marcel