Eigen efficient type for dense symmetric matrix

Question

Does Eigen have efficient type for store dense, fixed-size, symmetric matrix? (hey, they are ubiquitous!)

I.e. for N=9, it should store only (1+9)*9/2==45 elements and it has appropriate operations. For instance there should be efficient addition of two symmetric matrices, which returns simmilar symmetric matrix.

If there is no such thing, which actions (looks like this) I should make to introduce such type to Eigen? Does it has concepts of "Views"? Can I write something like "matrix view" for my own type, which would make it Eigen-friednly?

P.S. Probably I can treat plain array as 1xN matrix using map, and do operations on it. But it is not the cleanest solution.

Answer 1

Yes, eigen3 has the concept of views. It doesn't do anything to the storage though. Just as an idea though, you might be able to share a larger block for two symmetric matrices of the same type:

Matrix<float,4,4> A1, A2; // assume A1 and A2 to be symmetric
Matrix<float,5,4> A;
A.topRightCorner<4,4>().triangularView<Upper>() = A1;
A.bottomLeftCorner<4,4>().triangularView<Lower>() = A2;

Its pretty cumbersome though, and I would only use it if your memory is really precious.