Closed oskooi closed 8 months ago
128 might be a little on the large side for e.g. m=0
?
Probably it should increase proportional to m
.
Would e.g. 16 + int(4*abs(m))
be sufficient here?
Would e.g. 16 + int(4*abs(m)) be sufficient here?
Verified that this produces nearly the same results in #2726 (comment) as N0 = 128
.
Closes #2764.
Rather than make the number of grid points in the $\phi$ quadrature a user parameter in
get_farfields
, it is simpler just to increase the hard-coded value forN0
set internally ingreencyl
. The impact of integrating over a larger number of grid points should be insignificant when compared to the accuracy improvements enabled by this modification as this integral converges exponentially fast.