Haskell library for monadic composition of probabilistic functions and construction of stochastic processes.
Resulting computations can then be run to generate data.
Blog post has some along with examples.
- See where performance can be improved.
- Check if compiling with -O2 really gives a performance improvement.
- Upload charts with profiled tests and memory usage.
- Add more usage information to README.
- Implement learning for generative models.
- Extend StochProcess to allow for ARCH models.
- Separate StochProcess and Sample modules.
- See if building on top of mwc-random is better than carrying around StdGens.
- Reduce composed probability distributions that have closed-form solutions before sampling.