Extracted stiffness matrix not giving sensical modes

[jjalora]

I've managed to extract the stiffness matrix at the equilibrium point using the new (experimental) feature.

I did a modal analysis of the Diamond soft robot (see SoftRobot plugin) by solving the generalized eigenvalue problem for the extracted $M$ and $K$ (in this case, I took -$K$ of what I get from SOFA) evaluated at the equilibrium point.

I get the following modes:

Clearly this is wrong. I'm wondering if any of you (particularly, @hugtalbot or @alxbilger) could replicate these results? Do I need to preprocess the $K$ matrix in any way prior to doing the analysis?

Thanks for the help!

[jjalora]

Seeing that the first coupled modes look like translation modes, is it possible that the extracted stiffness matrix is the unconstrained version (i.e., not constrained to the base)? If so, would it be possible to extract the constrained version of the stiffness matrix?

[alxbilger]

I confirm that the stiffness matrix is not constrained. It should be made more explicit.
Currently, there is no option to add the constraints or not. But since you are in a Python environment, you can manipulate the matrix manually. If you know the index of a fixed DOF, you can set the rows and columns (3 * index, 3 * index + 1 and 3 * index + 2) of the matrix to zero, except the diagonal where it is 1.

Let us know if it leads to better results.

[jjalora]

@alxbilger - this seems to be the solution! Thanks for the help.

Now I get the following modal displacements after a modal analysis of the $M$ and $K$ matrices.

[alxbilger]

I am glad it worked