-
DiffEqDiffTools supports much faster finite differencing routines than Calculus.
Ref: discussion on gitter.
-
## 🐛 Bug
Torch fails to build (even with `BUILD_TEST=0`) on `/opt/pytorch/torch/csrc/Dtype.cpp`; looks like another instance of warning converted to errors:
```bash
[0mFirst, rewinding head to…
-
Sometimes gradients are vectorized, and sometimes they are not. Consider this example where the gradient is vectorized (ie. array in, array of derivatives out).
```
def f(x):
return x**2
d…
-
Hamiltonian Monte Carlo (HMC), also known as Hybrid Monte Carlo, is an efficient Markov Chain Monte Carlo (MCMC) method that exploits gradient information to improve on the simpler MCMC methods. See t…
-
I am sorry for always asking stupid questions, but I really work hard on it before asking.
I encounter a JIT issue as below:
```
what(): ArrayFire Exception (Internal error:998):
In function st…
-
Similarly to #112, reverse mode autodiff support needs to be added for MCMC samplers that require higher order derivatives.
-
Hey, i have an issue. I need to inverse a Matrix with autograd numpy(i can't find a method for that) and at the end i have to do the gradient out of it. I know that an inverse Matrix is not differenti…
-
This issue is essentially a way of cross-referencing JuliaDiff/ReverseDiffSource.jl#28 from Lora. It will enable reverse mode autodiff via ReverseDiffSource in the case of models with more than one va…
-
Unlike `ReverseDiffSparse.jl`, `ReverseDiffSource.jl` doesn't unroll loops. It should be able to roughly match the speed of what we've coded manually.
For this issue, we'll fork `ReverseDiffSource.jl…
-
The coming 3.3 release of Eigen has more changes than usual
http://eigen.tuxfamily.org/index.php?title=3.3
And Stan Math has a few build errors with it.
```
./stan/math/prim/mat/fun/LDLT_factor.hpp:1…