@tpoisot

Cool! :) The main 'tasks' and 'bugs' are coordinated at savannah [1]; we have some loosely organised irc/matrix channels e.g. [2].

To test a real science paper (peer-reviewed), not too heavy computationally, I recommend [3] (a not-quite-final version ran fully from scratch on a pinephone).

[1] https://savannah.nongnu.org/support/?func=additem&group=reproduce

[2] irc: https://libera.chat ##maneage
matrix bridge: #maneage-community:matrix.org

[3] https://arxiv.org/abs/2112.14174 = https://zenodo.org/record/6794222

@ethanwhite

@khinsen

#Maneage directory policy allows neat shifting of output dirs. Intra-install (no recompilation) tracing shows exact results-reproducibility backwards :) to the step running calc_dens.py [1]. This uses Voronoi + ConvexHull from scipy.spatial. Re-running this step via our workflow [2] gives different results on the same N-body simulation.

Are scipy.spatial #Voronoi or #ConvexHull known to give non-reproducible results?

@civodul

[1] https://archive.softwareheritage.org/browse/revision/41507a706e16f04075c0c35853ddc226d3c72df5/?path=reproduce/analysis/python/calc_dens.py

[2] https://archive.softwareheritage.org/browse/revision/41507a706e16f04075c0c35853ddc226d3c72df5/?path=reproduce/analysis/make/calc_densities.mk

@civodul @khinsen

We use #CosmicVoids in [1][2], which in N-body sims are traced by low num-densities of particles => high noise. Full #Maneage controls + fixed seed rng's. We still have intramachine + (higher) intermachine randomness. Statistical upper limits to results OK. But still untraced sources of randomness.

Any clues for remaining randomness [2]?

#Reproducibility #ArXiv_2304_00591 #OpenScience

[1] Frozen record: https://zenodo.org/record/7792910

[2] Live git: https://codeberg.org/mpeper/lensing

#CosmicVoids should be detectable via their #GravitationalLensing effects.
Check this starting with a floppy-disk sized source snapshot [1].
#Reproducibility is via the #Maneage template of shell+make files.
Full research paper: [2][3][4][5]

[1] https://zenodo.org/record/7792910/files/lensing-e4f7af0-snapshot.tar.gz

[2] Preprint: https://arxiv.org/abs/2304.00591 #ArXiv_2304_00591

[3] Frozen record: https://zenodo.org/record/7792910

[4] Live git: https://codeberg.org/mpeper/lensing

[5] SWH: https://archive.softwareheritage.org/swh:1:rev:41507a706e16f04075c0c35853ddc226d3c72df5%3borigin=https://codeberg.org/mpeper/lensing

#Cosmology #OpenScience

@ogrisel

Seems like a pity for python not to benefit from GSL, especially when the explanation is to weaken the protection for freedom. Scipy certainly appears to be developing fast, seems to work correctly, has a wide variety of useful functions, and is well documented. Some reproducibility controls on python 3.10.6 in #Maneage are waiting for merging [2], though we're missing a python uninstall system [3].

@civodul @khinsen

[2] https://savannah.nongnu.org/task/index.php?16268

[3] https://savannah.nongnu.org/task/index.php?16270

@Eldeberen @sebsauvage

Dans le cadre de la #reproducibility des projets de recherche à la #maneage, voilà où nous sommes pour installer un .tar.gz d'un paquet python après vérification du checksum - en gros c'est 'bdist_wheel' + 'installer' :

python setup.py bdist_wheel
python -m installer dist/*.whl

Discussion et commits ici :
* https://savannah.nongnu.org/task/?16268

#Maneage talk for tomorrow morning 10:00 UTC+2 Fri 16 Sep 2022 at the Ten Years of Guix meeting [1], in principle live-streamed [2]:

https://cosmo.torun.pl/~boud/Roukema20220916TYG.pdf

[1] https://10years.guix.gnu.org/program

[2] https://10years.guix.gnu.org/live

#OpenScience
#Reproducibility
@civodul

On Friday, we'll have talks on using #Guix for research, but also on other people & tools sharing common goals: #Maneage, #Nix, #SoftwareHeritage, #OCaml #bootstrapping, #bioinformatics, and more.

Cosmology research paper reproducible from plain text source on a #Mobian #Pinephone using the #Maneage template/git tree:

* preprint: https://arXiv.org/abs/2112.14174

* full source + pdf: https://zenodo.org/record/5806028

* live git: https://codeberg.org/boud/gevcurvtest

* archived git: https://archive.softwareheritage.org/swh:1:rev:54398b720ddbac269ede30bf1e27fe27f07567f7

#Reproducibility on pinephone (or other): download sources, compile, non-root install, run simulations, analyse results, verify checksums, produce final pdf.

Please post bug reports at #codeberg !

@civodul @mobian @PINE64

Science #reproducibility challenge: does your project fully reproduce on a Unix-like phone?

A science project that branches off #Maneage commit 5a6e6e6 fully reproduced on my #Mobian #arm64 #pinephone: downloading source, configuring/compiling, running cosmology simulations, analysing+plotting results, producing a pdf (nearly) ready for peer review. :))

https://savannah.nongnu.org/bugs/?61026 - patches upstreamed to #Maneage.

PS: sms/phone/camera work fine too :P

#OpenScience

#TLDRBHTWL (too long, didn't read, but happy to watch and listen)?
Good overviews of #Maneage by Mohammad Akhlaghi :

20 minute presentation June 2020 - https://peertube.stream/videos/watch/f8b1b565-0e8d-4279-b7c5-f20e51e1d8da

80 minute longer presentation , including questions, March 2021 - https://peertube.stream/videos/watch/e7fdf27b-61b8-40cd-8306-482d79e4d6c5

#PeerTube #Reproducibility #OpenScience

Here's a Nature Astronomy News & Views description of why our #Reproducibility criteria for longevity are "rigorous" and "onerous", by #MichelleKuttel. There's a nice description of #Maneage, with a warning that Maneage does not seem to cater to Windows users (oh dear!), and that "bells and whistles" for the #TyrannyOfConvenience generation are lacking.

Privacy disclaimer: https://rdcu.be/cmYVx gives the subscription-only pdf but requires something like chromium with absolutely no protection!

Apparently #Reproducibility is very popular at #CiSE (Computing in Science and Engineering). The #IEEE host server for CiSE claims [1] that our #Maneage paper, formally published a few days ago (Akhlaghi+2021, CiSE, 23, 82 [2]), is currently CiSE's sixth most popular article!

[1] https://ieeexplore.ieee.org/xpl/topAccessedArticles.jsp?punumber=5992
(archives: https://web.archive.org/web/20210620130655if_/https://ieeexplore.ieee.org/xpl/topAccessedArticles.jsp?punumber=5992 https://archive.today/c0olT)

[2] https://zenodo.org/record/4913277
https://arxiv.org/abs/2006.03018
https://archive.softwareheritage.org/swh:1:rev:1cfb83e01b9ab6375727da5afd89bb3e744b9894%3Borigin=http://git.maneage.org/paper-concept.git

#OpenScience

@VickyRampin @remram44

Mohammad (Akhlaghi, the real main author, not just alphabetically favoured) accepted my changes based on your comments on ReproZip, with minor adjustments:

http://git.maneage.org/paper-concept.git/commit/?id=f88b104f6ed0f

https://upload.disroot.org/r/PaJyxAQD#MTpx0hT9hFx0qqJ6gF90NNgWen4FUPvO97O7kvwWuqs

You're acknowledged on p7; the extended ReproZip section is App B.T on p 24.

I'm not sure you'll be completely happy, but I did circulate the draft on my fediverse account a year ago... #Maneage (2020-06-05, 2020-09-24, 2020-11-27)

@VickyRampin

Our #Maneage paper in CiSE (https://oadoi.org/10.1109/MCSE.2021.3072860) on criteria for reproducible scientific papers states in section T of Appendix B what ReproZip is and why it doesn't satisfy our criteria (sorry!). In Appendix A.D, there's a paragraph on git #ScholarlyEphemera. See https://codeberg.org/maneage/paper-concept (currently the ephemera paragraph is in branch 'appA_archiving'). [Not yet updated: https://arxiv.org/abs/2006.03018 https://doi.org/10.5281/zenodo.4291207 ]

#OpenScience #reproducibility

@civodul

Que pensent les organisateurs d'une éventuelle présentation de #Maneage le mardi entre 11h30 et 12h00 ? [1] On déplacerait alors la table ronde un peu. L'article de base, ArXiv:2006.03018, vient d'être accepté, et deux articles scientifiques sous Maneage, ArXiv:2007.11779 et ArXiv:2010.03742, ont passé la prèmiere phase du peer review. ArXiv:2007.11779 a déjà eu presque deux mille téléchargements (mille vues) sur Zenodo : https://zenodo.org/record/4432080 .

[1] https://maneage.org

@civodul

Reproducibility without the legal right to develop a scientific project further wastes time and threatens copyright attacks against similar projects.

The 8 #Maneage criteria [1] are more favourable for community development:
* completeness (no non-POSIX dep., no root, network optional, plain text);
* modular design;
* minimal complexity;
* scalability;
* verifiable inputs and outputs;
* version control;
* linking analysis with narrative;
* free software.

[1] https://doi.org/10.5281/zenodo.3911395

#OpenScience #Cosmology #Reproducibility @civodul

Make your own #galaxies starting from a POSIX system, and check our results with our #Maneage pipeline:

"The role of the elaphrocentre in low surface brightness galaxy formation"

* #ArXiv preprint: https://arxiv.org/abs/2010.03742
* #Zenodo: https://zenodo.org/record/4062461
* live git #codeberg: https://codeberg.org/boud/elaphrocentre
* archived git #swh: https://archive.softwareheritage.org/swh:1:dir:c4770e81288f340083dd8aa9fe017103c4eaf476

Default resources for 128^3 particles, levelmax=12, 4 cores: 8 hours, 14.8 Gb RAM, 162 Gb disk.

@civodul

Merci ! :). "Most likely to succeed" is at least as useful as "posixally correct". This branch of #maneage (for a particular research paper) should become public reasonably soon ;).

@fribbledom This is for the #Maneage #reproducibility system for science research papers - https://maneage.org https://arxiv.org/abs/2006.03018 - in case you or anyone wants to know the relevance. :)