How to fix "FloatingPointError: divide by zero encountered in true_divide" ?

[chenghancc]

When I use DBGAN model for Link prediction, the following error occurs：

Traceback (most recent call last):
File "C:/Users/Monarch/Desktop/DBGAN-master/DBGAN_link/run.py", line 14, in
runner.erun()
File "C:\Users\Monarch\Desktop\DBGAN-master\DBGAN_link\link_prediction.py", line 43, in erun
DPP.sample_exact_k_dpp(size=0)
File "D:\Users\Monarch\Anaconda3\envs\tf1\lib\site-packages\dppy\finite_dpps.py", line 506, in sample_exact_k_dpp
random_state=rng)
File "D:\Users\Monarch\Anaconda3\envs\tf1\lib\site-packages\dppy\finite_dpps.py", line 492, in sample_exact_k_dpp
self.L_eig_vals = self.K_eig_vals / (1.0 - self.K_eig_vals)
FloatingPointError: divide by zero encountered in true_divide

Process finished with exit code 1

How to solve this problem ？

[SsGood]

Plz check the dppy version is 0.2.0 or not. If the version of dppy is correct while the problem still appear, you could also achieve the k-dpp sampling by another package, pydpp.

[chenghancc]

Plz check the dppy version is 0.2.0 or not. If the version of dppy is correct while the problem still appear, you could also achieve the k-dpp sampling by another package, pydpp.

Yes, the version of the dppy I use is 0.2.0. I will try pydpp. Thank you for your reply.

[chenghancc]

Plz check the dppy version is 0.2.0 or not. If the version of dppy is correct while the problem still appear, you could also achieve the k-dpp sampling by another package, pydpp.

Yes, the version of the dppy I use is 0.2.0. I will try pydpp. Thank you for your reply.

When I change dppy to pydpp the code works, thanks a lot！