-
I will admit i am not confident this is in fact a bug. Still it seems like odd behaviour and it took me a few hours to pin-point the NaN-source.
Minimal reproducing example:
```python
import jax…
-
It may be helpful to have a Gate node, or super node or at least some usage docs/pattern docs.
A `gate` node would filter one stream by another stream. This could be done by
```python
s1 = Stream…
-
Currently our loss functions are coded as straightforward functions working on torch inputs. Some loss functions have additional parameters that are set at initialization, for example,
`def entropy…
-
Dear Victor,
In "An Algorithm for Optimal Lambda Calculus Reduction" on page 5 Lamping writes "An obvious idea is to label the fans in the graph, with fans paired if and only if they have identical l…
-
We can improve the PME overlap by subtracting out the exceptions computed in reciprocal space.
The relevant OpenMM code that implements this [is located here](https://github.com/pandegroup/openmm/blo…
-
```
Hi,
I have tried to use nsum() for some cosine series, like: cos(nx)/(n^2)
however it doesn't work for x=pi*0.05. All methods produce wrong result with
fixed error.
The reason is wrong stopping …
-
Ufuncs support passing a dtype kwarg to pick the output value. This is particularly useful for casting the output to complex.
For example, the function `lambda x: np.real_if_close(np.cos(np.arccos…
-
In order to support Private deployments of the Data lake solution, allow for the configuration of a VPC deployed Elasticsearch cluster. I'm willing to collaborate and contribute on this change request…
-
# Feature request:
### Component:
- [x] Code
- [ ] Documentation
- [ ] Build system
- [ ] Continuous integration
### Desired feature:
We decided a long time ago to use `std::array` inst…
-
I've had a quick shot at running ABFEs with SOMD2. The overall experience was a massive step up from SOMD! The recent sire documentation was also super clear and comprehensive. However, I've had a cou…