T-PHATE algorithm

[apathanasiadis]

Hello, I recently came across the T-PHATE algorithm you published. Regarding its implementation, I was wondering how did you interfere with the PHATE algorithm. As far as I understand, you multiply the diffusion operator $P_D$ with the normalized autocorrelative transition probability matrix $P_T$. Can I somehow input this new operator $P=P_D P_T$ for the transformation of data? I know I can get the diffusion operator, diff_op, after the fitting, but I am not sure how to input $P$ as the new diffusion operator matrix for the transformation.
Thanks for any help in advance.