equation choice and pivoting #16
Description
In the formulation of the linear system in the final part of the algorithm, choosing the right subset of equations (matrix rows) to reduce to single linear equation and the order in which to reduce them (matrix pivoting) is not written in a fully general way. incorrect equation choice may lead to missing D.O.F.s (variables) in the solution (which appears to happen and results in high residuals), unnecessarily overdetermined systems (which definitely happens) with resulting soft-test failure (residuals acceptable). Poor pivoting may increase numerical error (especially for very complex systems?) or lead directly to NaNs if 0 on main diagonal (there is code to avoid this, but not necessarily fully general).