You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, the computation of the far fields in cylindrical coordinates involves an integration in $\phi$ (azimuthal angle) which is based on repeatedly doubling a grid of $\phi$ points using a hard-coded initial value:
/* Perform phi integral. Since phi integrand is smooth, quadrature with equally spaced points
should converge exponentially fast with the number N of quadrature points. We
repeatedly double N until convergence to tol is achieved, re-using previous points. */
constint N0 = 4;
The value of 4 turns out to be too small in some cases which produces discretization artifacts in the output as demonstrated in #2726 (comment). As such, it would be useful to make the initial number of $\phi$ grid points a user parameter in the routines get_farfield and get_farfields.
The text was updated successfully, but these errors were encountered:
Currently, the computation of the far fields in cylindrical coordinates involves an integration in$\phi$ (azimuthal angle) which is based on repeatedly doubling a grid of $\phi$ points using a hard-coded initial value:
meep/src/near2far.cpp
Lines 288 to 291 in 04fa305
The value of 4 turns out to be too small in some cases which produces discretization artifacts in the output as demonstrated in #2726 (comment). As such, it would be useful to make the initial number of$\phi$ grid points a user parameter in the routines
get_farfield
andget_farfields
.The text was updated successfully, but these errors were encountered: