Hello,

Our research group goal is to calculate circular dichroism (CD) for protiens. Our program is working fine for small protiens, but it is outputing garbage values, or stopping (because of memory issues), when we're inputing big protiens.

We're getting the above problem because we're using the equation as shown in below:

AB = A11-A12*(InvA22)*A21.

As the size of the protien increases, size of AB increases there by causing the problem.

What is the best way to deal this probelm?

Any advice would be greatly appreciated.

Looking forward for your reply!

Could you give us some hints about the average sizes of each involved matrix A* as well as there structure ?

Perhaps your matrices are sparse, i.e., with many zeros, in which case you could use Eigen's SparseMatrix...

You can also rewrite your expression as:

AB = A11;
AB.noalias() -= A12*(InvA22*A21);

to save one temporary of the same size than AB.