It’s another year on so another post, another update for FEST…
I’ve been running transient models and even for small problems, a second or so per solve leads to long (Minutes!) run times for many timestep problems. My home coded matrix solver is fine and very robust, but there are quicker options available and Eigen (Licenced under the free MIT licence) offers a range of linear algebra solvers. YULE-FEST (Yet-another Update for Linear-algebra using Eigen for FEST) implements a range of sparse and dense matrix solver options. The sparse matrix LU decomposition speeds up the transient solution to a small radial couple heat-groundwater flow problem by around two orders of magnitude. 200 timesteps takes around 20 seconds instead of one coffee break (10 mins) so thats a win for my night’s sleep (as I get less caffeine tolerant after lunch!).
Here’s a Yule-tide colour themed example…
