-
On the documentation page [LV-Univ.md]( https://diffeqflux.sciml.ai/dev/examples/LV-Univ/), the introductory text states:
> Here's an example of doing this with both reverse-mode autodifferentiatio…
-
Already in the abstract we have a phrase like "With a bit of elementary mathematics," which is completely unhelpful. Adjoints from category theory, and much else besides, does show up here, and these …
-
This issue tracks the necessary changes after https://github.com/JuliaDiff/ChainRulesCore.jl/issues/167 is merged:
- ChainRulesCore:
- [x] Update `ChainRulesCore.@scalar_rule` to conjugate the pa…
-
#### Summary:
It appears that in `stan_lm()`, `singular.ok` does the opposite of what it is supposed to do. I think the logic is backwards somewhere.
#### Description:
In the example below, I del…
-
Here is a MWE:
```
using KernelFunctions, Flux, LinearAlgebra
k = transform(SqExponentialKernel(), 2.0)
ps = Flux.params(k)
X = rand(10, 1); x = vec(X)
A = rand(10, 10)
g = gradient(ps) do
t…
-
I've been working on adding a `TiledIndex` with a `IndexTiled
-
Hello,
The following instruction
`F[indj*d+l] += wjk * parts[indk*d+l];`
Make the code needs quadratic memory in the backward pass.
```
#include
#include
#include
#include
#includ…
-
Do we want a section on how to take derivatives with respect to tensors? It ought to be easy compared to [matrices](https://en.wikipedia.org/wiki/Matrix_calculus).
-
## Description
When adding ```CXXFLAGS += -march=native -mtune=native``` the mix tests for ```pow``` fail due to all tests resulting in zero gradients, but *only* when compiling with ```g++```. The t…
-
When using ReverseDiff to compute the gradients of an inplace function `f!`, we found that the output value of ReverseDiff depends on the input value of the tape (y*2 in the MWE). For the non-mutating…