RBCsimulation / Blog: Recent posts

Date: June 15, 2014

RBC force computation is generally working. There are three major issues.

1. grad Pi.v for virtual displacement v selected to be each spherical harmonic function does not match exactly with finite difference approximation (compare [avw1; avw1_fd] etc.). The match improves as nlat is increased (for example, nlat = 21 is much better than nlat = 5, 7, 9). I think this is because of aliasing.
2. When verbose = 2, grad Pi is not tangent to the graph of Pi in the very last steps of steepest descent for computing eta (particularly when tolerance asked is lowered). I think is also because of aliasing.
3. When the results from mySphericalHarmonics.m are compared with shsgcm (using testSH.m), there appears to be a sign difference with the b coefficients.... read more