Jeffrey Newman

fix copy and pickle so they work with freezing

4

This PR makes it so frozen Dict's can be copied and pickled. Addresses #136 ``` >>> from addict import Dict >>> a = Dict().copy() >>> a.foo.bar {} >>> b =...

allow namedtuples to be values

1

For optimization reasons (e.g. numba), sometimes it's preferred to store some branches of a Dict as namedtuples instead of like normal Dict branches.

Dependabot is warning of a security issue with numpy < 1.22. To fix it we need to also upgrade to numba 0.56. This PR is currently erroring because numba 0.56...

[Phase 7] Sharrow -- 1 Zone

1

This PR supercedes #542, and is written with black enabled to simplify code review (relatively)

Need OP_SKIMS in trip destination model / error in ARC example model

2

There is a line in the spec for the example ARC model marked with the comment: > THIS IS WRONG ***** RIGHT HERE **** I CHANGED THE BELOW TO MAKE...

The psutil library cannot measure USS values greater than 2^37 bytes on Windows. That's about 128GB. The limitation is hard-coded [here](https://github.com/giampaolo/psutil/blob/4f053312934cdca739c3fefe4ada8a4f2d096bdd/psutil/_psutil_windows.c#L523). This limitation probably doesn't matter in practice -- if...

Clean up documentation of the base MTC example

2

There is documentation for the inputs of the demo "MTC" model here: https://activitysim.github.io/activitysim/examples.html#inputs But it isn't formatted well and it is hard to read and follow. These docs could benefit...

I suggest we follow the "new" (from 18 months ago) [default branch naming scheme at GitHub](https://www.zdnet.com/article/github-to-replace-master-with-alternative-term-to-avoid-slavery-references/), and swap out the "master" name for "main". This change should have minimal impact...

One comment I will make: I feel as though the way the aggregation maps are currently defined isn't super user-friendly. For example, the major access trip mode in the MTC...

Bitwidth expectations

4

The current ActivitySim implementation generally does not explicitly define bitwidths for numerical values, but instead allows numpy to self-select the numerical precision. This generally results in double precision floats almost...