Different bandwidth results by using scott/silverman method

[Gooda11]

Hello, I am currently using KDE. I used the silverman/scott methods to obtain the bandwidth in KDEpy. Then I used the silverman/scott methods in Scipy and Sklearn, but the bandwidths from KDEpy are different from the ones using Scipy and Sklearn, though the method is the same.
Anyone knows why?
Thanks in advance

[tommyod]

Hi!

How large are the differences? Do you have a working example (code) that illustrates it?

You'll have to look at the source code to investigate why the implementations might differ. I doubt that it's a bug, but it could be. If you find a bug, please do report it.