PyTorch matrix factorization with fixed item matrix

Question

I estimate ratings in a user-item matrix by decomposing the matrix into two matrices P and Q using PyTorch Matrrix Factorization. I got my loss function L(X-PQ). Let\'s say rows