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
Hello,
I think there is an issue in the method generate() of NumpySeedSequence.
The state calculated when the words's size is 64 bits it's not the same as numpy.
So we don't have same results for a same seed.
I think you have to change return (result_type)((uint64_t)state1 << 32 | state2); to return (result_type)(state1 | (uint64_t)state2 << 32) Thanks to this we have same randoms values like numpy.
Thanks for you project :)
The text was updated successfully, but these errors were encountered:
Hey thanks for the kind words, I think the result might've altered because I was using little endian architecture and you're uing big endian machine, my bad I should've commented those out but anyways can you do further testing, and open a pull request?
Hello,
I think there is an issue in the method generate() of NumpySeedSequence.
The state calculated when the words's size is 64 bits it's not the same as numpy.
So we don't have same results for a same seed.
I think you have to change
return (result_type)((uint64_t)state1 << 32 | state2);
toreturn (result_type)(state1 | (uint64_t)state2 << 32)
Thanks to this we have same randoms values like numpy.Thanks for you project :)
The text was updated successfully, but these errors were encountered: