[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Non-linear systems
Hi,
I remember that I have had the problem with a large number of polinomial
non-linear equations. In one year I tried with a lot of algorithms and
finally found a subroutine in CERN library was almost perfect with very
good error control and optimization. I tried to download it and related
routines but it turned out to be very big, probably because they tried
to make it work for the possibly largest class of equations.
The N=31 was really not a matter, I remember I had few hundred equations and
got the solution in couple minutes. In fact I ran the program on an IBM
mainframe. The algorithm was some gradient one as Anh Thanh said.
If you require a good precision or your equation system is stiff enough,
don't trust in any MAtlab, Maple
and Mathematica. By that time, I fed the Matlab solution into my program
for an interative solution. The program ran into an infinite diverging loop
(!!!). In fact, my equations were very bad and only CERN library could
handle. That is why I trust this library.
Cheers
Aiviet