Closed aroffringa closed 1 year ago
Please be sure to test the GPU code -- I haven't done that.
The CPU code now produces the same Gaussians as Dp3 when Gaussians are circular, and Dp3 has been well tested to produce the correct result. I haven't tested whether the projection and non-circular Gaussians work correctly now.
Thanks, I will do some more tests and merge this.
Sagecal simulates Gaussian sources with an integrated flux density that is ~1.6 times too high. It appears that this is caused by a "half pi" factor that is in the Gaussian prediction unnecessary.