AugurV2 generates compositional MCMC inference algorithms for a simple probabilistic modeling language.
AugurV2 additional depends on:
- GSL - GNU scientific library (tested with version 2.3)
- clang - C compiler
- Cuda - Cuda Toolkit (tested with version 7.5)
Option 1: Build the AugurV2 compiler from source (using cabal).
$cd compiler/augur
$cabal sandbox init
$cabal install --dependencies-only
$cabal build
Option 2: Build the AugurV2 compiler from source (using stack).
$cd compiler/augur
$stack init --solver
$stack build
Build the AugurV2 runtime.
$cd cbits
$make libcpu
If you have Cuda support with dynamic parallelism (architecture >= sm_35), you can also build the GPU library.
$make libgpu
Build the AugurV2 Python interface.
$cd pyaugur
$python setup.py install
- release optimizations
Apache-2.0