Hi Alex,
There seems to be an issue with the power normalisation for bsc_pointmatch_farfield. Around line 175, rw is calculated from tan(theta).^2, where theta is in the range ((1:ntheta)-0.5)/ntheta*pi which will sometimes be approximately pi/2. Even when it is not exactly pi/2, nearby values still cause a problem since rw is later used in exp(rw/2 + ...) which quickly becomes too large.
Could you take a look at this.
Cheers,
Isaac
https://github.com/ilent2/ott/blob/70202a19828a3c96fb82f6e3cac468a12c77811c/ott-toolbox/bsc_pointmatch_farfield.m#L175
Hi Alex, There seems to be an issue with the power normalisation for bsc_pointmatch_farfield. Around line 175, rw is calculated from tan(theta).^2, where theta is in the range ((1:ntheta)-0.5)/ntheta*pi which will sometimes be approximately pi/2. Even when it is not exactly pi/2, nearby values still cause a problem since rw is later used in exp(rw/2 + ...) which quickly becomes too large. Could you take a look at this. Cheers, Isaac