Closed smao-astro closed 3 years ago
Thanks for the comment. The 2*pi
factor shouldn't play a big role -- its effect can be absorbed in the frequency used to initialize the Fourier features. The results reported in the paper correspond to the code provided here, i.e. using cos(b * x), sin(b * x)
.
I see, thank you.
Hi,
Thank you very much for make the code available.
In the Fourier feature part, such as https://github.com/PredictiveIntelligenceLab/MultiscalePINNs/blob/325b62c929d55028de848d8884e4c59377e68754/Poisson1D/models_tf.py#L310-L316 and all the other part, seems that you are doing
cos(b * x), sin(b * x)
, however, in the initial paper and your paper, iscos(2 * pi * b * x), sin(2 * pi * b * x)
, I know it is only a difference of factor, but would you like to explain which one gives the result of your paper?Thanks.