Symdiff
GPU-powered symbolic differentiation in haskell
Install / Use
/learn @kathawala/SymdiffREADME
symdiff
GPU-powered symbolic differentiation in haskell
Status
Proof of concept generation of gradient expressions is in the test function in Diff.hs
##Build Steps Before building, you will need to have CUDA installed on your system, both the developer driver and SDK.
You will also need the git versions of the accelerate library and the accelerate-cuda library. These versions are not available on hackage, so you will need to clone the repo and build it yourself.
Assuming you have CUDA and nvcc in your PATH, then you can download and build the library as follows:
git clone https://github.com/AccelerateHS/accelerate.git
git clone https://github.com/AccelerateHS/accelerate-cuda.git
git clone https://github.com/kathawala/symdiff.git
cd symdiff
cabal sandbox init
cabal sandbox add-source ../accelerate
cabal sandbox add-source ../accelerate-cuda
cabal sandbox add-source ./cublas
cabal install
