During the numerical solution of current-voltage characteristics of a FINFET, i encountered a large matrix (but sparse) of 1500000*1500000 size. I need to invert the matrix. I used many techniques still couldn't calculate in my PC with 16 GB RAM (Matlab show "Memory Problem"). Can anyone help me with that?
For proper suggestion, i should add that i already used the sparsity property of the matrix and also some matrix decomposition techniques. Still the problem won't go.
Thanks Gregory.However problem is memory related.It can't invert such large sparse matrix at least on my PC.
Ah yes did not read your question fully (:
Try contact MATLAB they are very helpful. I cannot recall but I am sure there is a feature to allow the matrix size to be reduced by reordering. There are few different reordering algorithms. Have you tried?
Check under sparse reordering in MATLAB help. There are several reordering methods.
There is some chance the result will not be sparse. Do you really need that inverse, or do you instead need to solve linear equations Ax=b (where the matix is A)? If the latter, there are many indirect solvers that do not do matrix decomposition.
If you really think you need that inverse, it might be a useful exercise to consider ways to recast your problem so that you don't need it...
- Badges
- Science topic