Comments (3)
I also use a mixed C++ and Python approach:
https://reaktoro.org/tutorials/equilibrium/co2-solubility-nacl-brine.html
And I'll also eventually need to have python bindings (I also use pybind11) for at least dual
!
I'm currently considering not to generate template expression trees at compile time when performing mathematical/arithmetic operations on dual
numbers. This was originally conceived to avoid temporaries as much as possible, and also to allow certain optimizations at compile time. However, since dual
is entirely allocated on the stack, it may be that we don't need to be too much concerned about temporaries (since stack allocation and deallocation are fast operations). We can also use move semantics to re-use temporaries as much as possible.
If this gets in place, then we'll be able to perform auto differentiation in Python as well (but of course, this should be done very sparingly, since Python will add some overhead into the mathematical/arithmetic operations that do not exist in the C++ side).
from autodiff.
Definitely would be awesome to get this capability in Python. At the moment my solution is multicomplex algebra that I have been working on with some colleagues and preparing to release pretty soon. Might still be faster than AD, but likely AD is faster.
So far the AD solutions in Python are VERY slow in my experience, so even a reasonable amount of overhead would be pretty acceptable in my opinion.
from autodiff.
We'll definitely have this; also high priority for my research needs!
from autodiff.
Related Issues (20)
- [Feature request] Code generation HOT 2
- Combined gradients HOT 2
- cmake compile error: numbertraits.hpp:70:16 HOT 1
- Directional Cross Derivatives HOT 1
- Is non-const argument in `gradient` necessary? HOT 1
- Dual will not work correctly for complex numbers HOT 3
- How to get the fastest performance of the autodiff. HOT 2
- Having trouble adding autodiff and Eigen to a project HOT 1
- Gitter link on website HOT 2
- Unable to build Python bindings with current master HOT 7
- An unexpected result when getting derivative of simple square function HOT 6
- [feature] GPU support HOT 2
- Adding const quietly invalidates results
- add gamma function
- Stack overflow HOT 3
- Derivative wrt x returns nan, but function does not depend on x at all. HOT 3
- CUDA: thrust::reduce fails for autodiff::real HOT 6
- Reverse-mode Hessian crashing on memory error
- `dual` and `real`, what's the difference? HOT 2
- current autodiff main branch does not work with current eigen master branch HOT 3
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from autodiff.