[Clp] Performance optimization

[Hauke Pribnow]

Hello everybody!

I have a pretty small Simplex problem (three rows, about nine to twenty 
columns) where almost all columns will stay constant most of the time 
whith the exception of the last one which will change quite often.

The column bounds for the mostly constant colums are 0 <= value <= 1 and 
for the often changing one either 0 <= value <= infinity or -infinity <= 
value <= 0. The row bounds are set to value = 0 (or 0 <= value <= 0) for 
all the time.

There is only one objective coefficient: the one for the last row, and 
that is either -1 or 1. Same goes for the optimization direction... it 
is -1 or 1 as well. (When the objective coefficient is 1, the 
optimziation direction is -1 and vice-versa. Concerning the col bounds 
there's also such a rule: When the objective coefficient is 1, the col 
bounds for the last column is 0 <= value and in the other case value <= 0.)

This (changing) Simplex problem will have to be solved about 120 times 
in a second, which is the reason why I am looking for a way to optimize 
the performance of it. Since I think that a lot of time gets lost by 
using the deleteColumns() and addColumn() functions, I'm especially 
looking for a way to write data directly into the matrix.

In case you can help me to find a way to do that or even to optimize my 
problem further, I'd be happy about any type of reply. :)

Thanks in advance!


Hauke Pribnow