lecture-python.myst icon indicating copy to clipboard operation
lecture-python.myst copied to clipboard

Published 20 hours ago • verified publisher icon QuantEcon

MAINT: Upgrade to anaconda=2024.02 and cuda=12.3.1

Open mmcky opened this issue 1 year ago • 10 comments
trafficstars

This PR

  • [x] upgrades to anaconda=2024.02
  • [x] upgrades to CUDA=12.3.1

this PR is part of https://github.com/QuantEcon/meta/issues/130

requires: #387

  • [ ] this PR is waiting on interpolations to be updated for numba>=0.59

mmcky avatar Mar 07 '24 04:03 mmcky

  • [x] the broken link is to BK19 as a reference but that doesn't seem to be in our quant-con.bib file
You can read  about Dynamic Mode Decomposition here {cite}`DMD_book` and here [[BK19](https://python.quantecon.org/zreferences.html#id25)] (section 7.2).

@jstac or @thomassargent30 do you know what BK19 would refer to and I can add it as a proper reference using bibtex

mmcky avatar Mar 07 '24 04:03 mmcky

@mmcky , it refers to this one: https://www.databookuw.com/

jstac avatar Mar 07 '24 06:03 jstac

  • [x] this pr requires #387

mmcky avatar Mar 08 '24 00:03 mmcky

  • [x] this cannot be merged until interpolations is updated or there is a substitute for mlinterp that is suitable for the odu lecture (cc @kp992)
ImportError[0m: cannot import name 'generated_jit' from 'numba' (/opt/conda/envs/quantecon/lib/python3.11/site-packages/numba/__init__.py)

mmcky avatar Mar 12 '24 03:03 mmcky

I see, from numba 0.60, generated_jit will fail.

kp992 avatar Mar 12 '24 16:03 kp992

I see, from numba 0.60, generated_jit will fail.

@kp992 I think from numba>=0.59 it will fail -- so that is why we will need to fix interpolation as we use it for minterp and some splines support. From my discussion with Pablo, the generated_jit will need to be replaced with an alternative (overload?) https://github.com/EconForge/interpolation.py/issues/110

mmcky avatar Mar 13 '24 00:03 mmcky

From my discussion with Pablo, the generated_jit will need to be replaced with an alternative (overload?)

I have tried fixing that locally but that's throwing some error. I followed the examples of moving from generated_jit to overload, but didn't work. Let's pin numba < 0.59 until we find and do robust a fix for this issue.

kp992 avatar Mar 13 '24 02:03 kp992

  • [x] check status of https://github.com/EconForge/interpolation.py/pull/114

mmcky avatar Mar 21 '24 03:03 mmcky

🚀 Deployed on https://662db0ad52d90f4fc4907697--nostalgic-wright-5fa355.netlify.app

github-actions[bot] avatar Mar 22 '24 07:03 github-actions[bot]

  • [x] re-enable build cache
  • [x] open an issue to update pip install from main branch of interpolation repo to pypi once released

mmcky avatar Mar 22 '24 20:03 mmcky

  • [x] reverted to install interpolation from pypi which is currently 2.2.6

mmcky avatar Apr 26 '24 06:04 mmcky