-
```
That is we need to calculate
(jk,lm|LM)
```
Original issue reported on code.google.com by `kirill.s...@gmail.com` on 31 Mar 2008 at 6:33
-
Hi,
i am attempting to build a equivariant Variational Encoder-Decoder framework.
For this I am using R2Conv() and R3(Conv) layers in the encoder with trivial-representation input & output and …
-
Although there are already some ideas in #3, we should have more fine-grained discussion on the content of QuBase. Maybe we can use this issue to collect some ideas and refine the roadmap in #3.
The …
acroy updated
6 years ago
-
Imported from SourceForge on 2024-07-05 09:26:20
Created by **[zmth](https://sourceforge.net/u/zmth/)** on 2014-04-28 23:12:34
Original: https://sourceforge.net/p/maxima/bugs/2728
---
wxMaxima versi…
rtoy updated
3 months ago
-
In the current implementation, the process of computing the Clebsh-Gordan coefficients is implemented in one large function.
Therefore, I propose to split the function to improve code maintainability…
-
The [orthogonality relations](https://en.wikipedia.org/wiki/3-j_symbol#Orthogonality_relations) are a nice test, as they only pass if all properties are correct.
Also, I'm not sure that the Clebsh…
-
I'm confusing about learnable weights implemented in `FullyConnectedTensorProduct`.
For example
```
tp = FullyConnectedTensorProduct(irreps_in1="1x0e",irreps_in2="2x0e",irreps_out="1x0e")
w = list…
-
Remove the optional unused `prec` argument from wigner_3j and clebsch_gordan. It's confusing as it has no effect and the docstring statements about it affecting the speed of the calculation are wrong.…