-
The "pyrely.performance" subpackage should provide the API for performance functions (scalar output only in a first time). The following features need to be studied:
- Should the user provide a classi…
-
Hi,
Would it be possible to extent test.py to show how to use finite differences to check the backward implementation? Or is there any other resource on this topic than the test directory in the main…
-
**Please describe the problem you are trying to solve.**
If you use Runge-Kutta methods as time stepping schemes, you can relatively easy implement them by
defining the weak form for the stages of …
-
the method your propose NLST is Interesting.
I have some questions about the retrial normal.
In your paper, you said that We retrieve the surface normal from the SDF via finite differences and trans…
-
Previously, the first time domain source used in the forward simulation was pulled to construct the source for the adjoint simulation (calculating a scaling factor). With the change in #1515:
https…
-
Hi @avik-pal!
I'm heading towards multi-argument and non-array support in DI, and I'd like to start testing Lux layers. For this I would need two things:
- Suggestions for a test suite of layers s…
-
How gradient and jacobian is computed in PSOPT to supply for IPOPT. I have gone through the example code, garadient of jacobian computation are not computed.
-
For simple functions that can be written as SymPy expressions, we could automatically generate the RobOptim function (`impl_compute`, `impl_gradient`, and even `impl_hessian`) from the SymPy expressio…
-
The "core" of `x/optym` is the convention `def thing(fg: callable)`, where `fg` returns `(cost, grad)` based on the parameter vector `x`.
This is in a way restrictive, since gradient-less optimizer…
-
The other day, I tried implementing the scalar Stefan problem with Phaseflow, while debugging the coupled concentration equation. This was apparently the first time I'd ever tried a scalar problem wi…