chiamin
chiamin
When getting a block, can we also get the corresponding quantum numbers?
The parameter `keepdim` in `Svd_truncate` should have a default value (which can be the largest integer) so that users can use only `cutoff` to decide the truncated weight.
The function `UniTensor.combineBonds` changes the tensor so there should be an underscore in the function name.
I suggest not printing all the elements if the tensor has too many elements. I am not sure what is the best way to print it though.
I have a strange error. If I import cytnx, I got Segmentation fault when using matplotlib show(). This happened for either build-from-scratch or conda build. See a demo as follows...
When doing DMRG, from time to time the program is stopped and I got the following message: ``` [WARNING] Lanczos_Gnd -> Tridiag error: Lanczos continues automatically. # Cytnx warning occur...
I hope we can use Network but keep the final memory layout for the input tensors, so that we don't need to move memory in every call. This is useful...
We will need a simple way to create an MPO from the physical operators, similar to AutoMPO in ITensor.