The rigid stiffness matrices

[pato]

Hi,

When I track your codes about building rigid stiffness matrices (in function FESolidSolver::RigidStiffness), I find that when both node j and node i are attached to some rigid bodies, you only couple the non-rigid degrees of node i to the rigid dofs of node j, but you do not couple the non-rigid degrees of node j to the rigid dofs of node i.

Then I check your implement of sparse matrix (class SkylineMatrix), there is a very strong assumption that the matrix must be symmetric, when you add some value in the matrix you only add to the upper triangular part, so the result is right.

However, I think maybe this is a very dangerous implement but I am not sure there is any other re-check in your implement.

I am a freshman in FEM, so maybe my question is naive
Thanks for your understanding.

Kun

[maas]

Hi Kun,

Sorry I didn't get to this sooner. Unless I misunderstood your question, it seems to me that all combinations of rigid versus non-rigid degrees of freedom are considered.

It is true though that the Skyline solver assumes a symmetric matrix and only grabs the upper-triangular part of the element stiffness matrix. However, since most analysis modes in FEBio create symmetric matrices, this is not a big problem. Only for biphasic analysis you may need a non-symmetric matrix in which case you cannot use Skyline.

I hope that answers your question. Again, sorry for the late reply. Let me know if you have any further questions.

Cheers,

Steve.

[pato]

Hi Steve,

That is my question. Because I think maybe some application will also use non-symmetric matrix. Thanks for your reply. I am clear now

Best
Kun

Hi Kun,

Sorry I didn't get to this sooner. Unless I misunderstood your question, it seems to me that all combinations of rigid versus non-rigid degrees of freedom are considered.

It is true though that the Skyline solver assumes a symmetric matrix and only grabs the upper-triangular part of the element stiffness matrix. However, since most analysis modes in FEBio create symmetric matrices, this is not a big problem. Only for biphasic analysis you may need a non-symmetric matrix in which case you cannot use Skyline.

I hope that answers your question. Again, sorry for the late reply. Let me know if you have any further questions.

Cheers,

Steve.