George Ho

Upstream PR: https://github.com/pymc-devs/pymc3/pull/4858 Switches from `dill` to `cloudpickle`. This has implications for `littlemcmc` and we should support this.

Upstream PR: https://github.com/pymc-devs/pymc3/pull/4500 This is a huge PR that I don't have the time to properly vet. Leaving this open until I can figure out what to do about it.

Upstream PRs: https://github.com/pymc-devs/pymc3/pull/4261 and https://github.com/pymc-devs/pymc3/pull/4211 Both seem relevant to LittleMCMC.

Upstream PR: https://github.com/pymc-devs/pymc3/pull/3990 This adds more details to the divergence warnings, which will be very helpful here.

Upstream PR: https://github.com/pymc-devs/pymc3/pull/4285 Fixes regression caused by https://github.com/pymc-devs/pymc3/pull/4211. This seems relevant.

Hello @jmccorriston! > However, in my mental model of the boundaries between Alphalens/Pyfolio/Zipline, [looking at different holding periods is] a job for Zipline. At the risk of inflaming disagreement (😇),...

Alphalens uses versioneer to pull release tags from git tags. Right now it's [set up to expect a git tag `v0.4.0`](https://github.com/quantopian/alphalens/blob/master/setup.cfg#L9), but the latest tag is just [`0.4.0`](https://github.com/quantopian/alphalens/tree/0.4.0). Could that...

Example: ```python # Create toy data dates = pd.date_range("20190101", "20190105") prices_dates = pd.date_range("20190101", "20190201") assets = list(range(10)) factor = pd.DataFrame(index=pd.MultiIndex.from_product([dates, assets]), data=np.random.randn(50)) prices = pd.DataFrame(index=prices_dates, columns=assets, data=np.abs(np.random.randn(32, 10))) factor.index.set_names(["date", "asset"],...

Edited!

Update: #46 merges in some more random variables (those that can be created via tfp's bijectors and transformed variables). There are still quite a few random variables outstanding, though.