search

econ-ark / DARKolo

fusion projects combining of Dolo/DolARK and Econ-ARK
4 stars 8 forks source link

Aiyagari chimera #12

Open sbenthall opened 4 years ago

sbenthall commented 4 years ago

Aiyagari '94 model

CDC students are working on this already for getting into HARK. Dolark ?? …

(from https://github.com/econ-ark/HARK/issues/414 )

sbenthall commented 4 years ago

@llorracc Have your students gotten this model into HARK? That's where this DARKolo left off earlier (in March).

llorracc commented 4 years ago

There's a PR from Mridul right now for the version he has been testing for compatibility with nbreproduce.

The PR looks odd though -- zillions of files are changed, and it looks like the PR is to merge it into econ-ark:master rather than into REMARK:master. Also, it seems to have a strange provenance ...

I've just posted a query to Mridul about whether it is ready to merge.

On Thu, Jul 16, 2020 at 5:51 PM Sebastian Benthall notifications@github.com wrote:

@llorracc https://github.com/llorracc Have your students gotten this model into HARK? That's where this DARKolo left off earlier (in March).

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/econ-ark/DARKolo/issues/12#issuecomment-659694196, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAKCK75NZFISH3EKDUAOJCDR35Y5ZANCNFSM4LF2MTXA .

--

sbenthall commented 4 years ago

Ok, I'm now looking at the Aiyagari REMARK in order to figure out how to turn it into a DARKolo chimera.

This is the notebook: https://github.com/econ-ark/REMARK/blob/902b71d242113a1ca26c7e67fcfc4a4374e24aa7/REMARKs/AiyagariIdiosyncratic/Aiyagari1994QJE.ipynb

This is the associated .yaml file: https://github.com/econ-ark/REMARK/blob/902b71d242113a1ca26c7e67fcfc4a4374e24aa7/REMARKs/AiyagariIdiosyncratic/Aiyagari.yaml

There is a lot of inventive stuff going on this REMARK that I'm not yet familiar with. Most notably, this REMARK uses dolark, which I haven't gotten a change to work with yet. It looks exciting. This is the Model API that is used in the YAML file. http://www.econforge.org/dolark/model_def/

It would be straightforward to copy in the dolark notebook and YAML file into this repository as a chimera.

I'm a little confused about what the chimeras here are up to if that's all that's required for this. I thought that the purpose of this repository was to compare Dolo and HARK implementations.

I wonder if it's higher priority to build HARK and Dolo representations, or to work directly with dolark.

sbenthall commented 4 years ago

Is there a pure HARK implementation of this model? Or is that what's required for this chimera in addition to the dolark implementation in the REMARK?

llorracc commented 4 years ago

There is not a HARK implementation of it. That is indeed what would be required for a chimera.

The Aiyagari model is basically a simplified version of the Krusell-Smith model, in which there are no aggregate shocks. (Or, to be more chronologically accurate, Krusell-Smith expanded Aiyagari to incorporate aggregate shocks).

Since Matt has just finished a revision to Krusell-Smith, probably the most effective way for you to begin would be to do a deep dive into Krusell-Smith; that would serve the additional purposes of

  1. vetting Matt’s new code;
  2. Identifying places where it is hard to follow (for someone not deeply versed in all of this stuff
    • And you can make PR’s for clarity/exposition improvements to the notebook as well as the new HARK code
  3. Wrapping your mind around what the Market class does
    • This is a core part of what you will need to do to advance the ABM project.

On Mon, Jul 20, 2020 at 9:38 PM Sebastian Benthall notifications@github.com wrote:

Is there a pure HARK implementation of this model? Or is that what's required for this chimera in addition to the dolark implementation in the REMARK?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/econ-ark/DARKolo/issues/12#issuecomment-661540798, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAKCK76PSUDVVALLUCSS6T3R4TWQ5ANCNFSM4LF2MTXA .

--

sbenthall commented 4 years ago

Ok, I follow. I will work towards a pure HARK Aiyagari '94 implementation as a chimera. The Dolark implementation is very helpful here. I will also read Aiyagari '94.

I've reviewed the KS HARK PR here. My view is that where I'm having trouble following it, it's for structural/design reasons that are not unique to this model. These have been much discussed in our meetings; there are some issues for them in the HARK repository. It would be my pleasure to submit PRs addressing these issues.

I see what you mean about the centrality of the market class for the ABM model.