-
## Description
Currently, `is_nan` is called in the `fvar` constructor. Remove. It's not done for primitives and `stan::math::var`s.
Please also remove the include to `is_nan.hpp`.
## Additiona…
-
## Description
There are two definitions of the incomplete beta function, both of which delegate to ```boost::math::ibeta```:
## Example
```prim/scal/fun/ibeta.hppp```:
```
inline double ibeta(…
-
**System information**
- Have I written custom code (as opposed to using a stock example script provided in TensorFlow): yes
- OS Platform and Distribution (e.g., Linux Ubuntu 16.04): colab
- Tenso…
-
Edit (@syclik): 4/24/2019. Reducing the issue to just reverse mode. We can create new issues for forward and mix later.
## Description
The goal here is to make a nice, generic testing framework bu…
-
#### Summary:
Create a test framework for functions based on legal arguments and values that will test all the derivative combinations.
#### Description:
There's an example already for unary …
-
This is a followup to #1907.
The code from #1907 does work now with the fix, but I am still getting the NotImplemented error I referred. The minimal example (below) is more complicated, but I hope …
-
# Issue
When the objective function includes a `pmap` (or, more generally, some sort of distributed call), use of auto-differentiation in `optimize` no longer works.
## Versions
Julia version…
-
Currently FMI 2.0 provides and interface to provide partial derivatives in the form of dirctional derivatives (e.g., Jacobian J times direction vector v, Jv)
For several use cases, it would be ben…
-
First off, thanks for developing JAX, it's awesome!
I recently noticed that when one computes higher order derivatives of expressions involving eigenvectors obtained from `jax.numpy.eigh` or `jax.s…
-
Hi Rob-
I noticed that the samplers use different auto differentiation algorithms, which will qualify the conclusions we can draw from the benchmarks. My naive understanding is that forward autodif…