>>12789708I did, but it's a rabbit hole of optimizations. Just doing a FEM is literally babby-tier, but then comes the isoparametric concept, sparse calculations, bandwidth optimization, problem detection like bad Jacobians, hourglassing, locking and mitigations like an extra energy term, and if course abstracting to allowing arbitrary elements to be used, hybrid elements..
That's where I stopped. I think it's necessary for every engineer to program FEM yourself at least once to understand it.
