[Coin-ipopt] Dense problem

[fvadakku at purdue.edu]

Hi all,

I was planning to use IPOPT C++ version to solve a dense constrained 
optimization problem. I expect the final version of the problem to have the 
following dimensions:

# variables: a few hundred (200 - 400)
# constraints: approximately 100,000

The objective function and all constraints are nonlinear. All constraints are 
inequality constraints. Every constraint depends on every variable.

However, after reading some posts in the archives, I suspect that I may run 
into problems such as insufficient memory since IPOPT is designed to solve 
sparse problems.

Could someone tell me what kind of difficulties I should expect? Will the 
sparse solver MA27 give me trouble?

Approximately, how much memory would I need? The machine I use has 1GB RAM.

So far I have been using the MATLAB routine fmincon() in  medium-scale mode 
which is based on SQP. Does anyone know how the memory requirements of IPOPT 
comapres with those of fmincon()?

Thank you
Fijoy