[larsblatny]

This PR does the following

• replaces the eigen SVD with a tailored SVD
• reducing the stencil for the particle grid interpolation depending on the spline
• makes cubic spline possible when use_sparse=true
• adds a test for MUSL and Sparse MPM (because these are affected by the above changes)
• improves N-R scheme for viscous models

This does not change any documentation for the user, hopefully they will only experience a faster execution.

[larsblatny]

@Yidong-ZHAO I made some changes to the sparse implementation to make it work for cubic spline. Would you mind having a quick look at g2p and p2g in case I messed something up?

[atpellet]

(Sorry, I misclicked and closed the PR, I opened it again. i hope it did cause any issue)

[Yidong-ZHAO]

@Yidong-ZHAO I made some changes to the sparse implementation to make it work for cubic spline. Would you mind having a quick look at g2p and p2g in case I messed something up?

Hi @larsblatny, I had a quick look and I don't see any problems. Basically you just need to change the base and stencil as you have done.

[larsblatny]

@Yidong-ZHAO thank you!

[larsblatny]

Hi everyone, I added another feature (the latest commit about N-R schemes) to this PR. Perhaps it should have been a separate PR as it is only partially related to the rest of this PR. If so I'm very sorry!