search

jackhamel16 / MagQuEST

GNU General Public License v2.0
0 stars 0 forks source link

22-Fix-jfnk-test #9

Open jackhamel16 opened 5 years ago

jackhamel16 commented 5 years ago

Fix JFNK solver test so that it is testing the solver algorithm

jackhamel16 commented 5 years ago

The test is being axed because writing it requires advanced unit test skills or rewriting the code to be more modular. I don't think this is a good use of time so I will do a test directly in the code against the analytical solution to just the precession term of the LLG equation.

jackhamel16 commented 5 years ago

Ran truncated LLG with dt=1e-15 and total_t=1e-10 and max error was ~1e-9 at max timestep. There is a slow linear accumulation of error throughout run. This shows that the JFNK solver is correct.