An interface to Anders Buch's Littlewood-Richardson Calculator ``lrcalc``

[mhansen]

The attached patch, and spkg implements an interface to Anders Buch's Littlewood-Richardson Calculator lrcalc

The "Littlewood-Richardson Calculator" is a C library for fast computation of Littlewood-Richardson (LR) coefficients and products of Schubert polynomials. It handles single LR coefficients, products of and coproducts of Schur functions, skew Schur functions, and fusion products. All of the above are achieved by counting LR (skew)-tableaux (also called Yamanouchi (skew)-tableaux) of appropriate shape and content by iterating through them. Additionally, lrcalc handles products of Schubert polynomials.

The web page of lrcalc is: ​http://math.rutgers.edu/~asbuch/lrcalc/

Here is a short extract of the doc:

EXAMPLES::

    sage: import sage.libs.lrcalc.lrcalc as lrcalc

Compute a single Littlewood-Richardson coefficient::

    sage: lrcalc.lrcoef([3,2,1],[2,1],[2,1])              #optional - lrcalc
    2

Compute a product of Schur functions; return the coefficients in the Schur expansion::

    sage: lrcalc.mult([2,1], [2,1])                       #optional - lrcalc
    {[3, 3]: 1, [4, 2]: 1, [3, 1, 1, 1]: 1, [4, 1, 1]: 1, [2, 2, 2]: 1, [3, 2, 1]: 2, [2, 2, 1, 1]: 1}

Add optional package lrcalc-1.1.6beta.spkg​

Apply: trac_10333-lrcalc-final.patch​

Making lrcalc into a standard package in the next Sage release is the topic of the followup ticket: #11563

[aschilling]

Together with Anders Buch, I went through this patch at Sage Days 30 and checked all functionalities. We added further doctests and also access to fusion coefficients (which was not in the previous version of this patch).

From the mathematical point of view, I give this a positive review. Could someone please do a quick technical review of the patch?

Thanks!

Anne

Apply: trac_10333-lrcalc-final.patch

[aschilling]

I fixed some more typos and omissions on the doctest.

Positive review on the mathematics and features.

Could someone complete the technical review?

Cheers,

Anne

[chapoton]

for the bot:

Apply: trac_10333-lrcalc-final.patch

[nthiery]

Hi,

I used the occasion to do something I had promissed Anders a while ago: namely to add a standard GNU build system (autoconf/automake) to the original lrcalc sources. With this, the SPKG becomes trivial. I'll upload shortly a preversion of it. If Anders agrees with the changes, and once he will have done an official release, I'll upload a final version of the spkg.

Mike: could you review now the three little files: SPKG.txt, spkg-install and spkg-check? Those are unlikely to change.

[nthiery]

I fixed a couple tiny typos in lrcalc.pyx, and uploaded an updated patch. For me this part is good to go.

[nthiery]

According to the vote on ​http://groups.google.com/group/sage-devel/browse_thread/thread/2e7114375f6f88a5/, there is a consensus on making lrcalc a standard package, after a one release probation period in order to follow the official rule. See: #11563

[aschilling]

Can we now set a positive review on this patch?

Anne

[ncohen]

(pending an answer, if no one knows whether this ticket should be reviewed...)

[aschilling]

Replying to ncohen:

(pending an answer, if no one knows whether this ticket should be reviewed...)

As far as I know, everything is fine with the ticket except that Nicolas is waiting for confirmation from Anders Buch about the spkg. Nicolas, is everything ok now on this front?

Anne

[ncohen]

Got it ! Then what about setting it to needs_info in the meantime sothat it does not appear in the list of tickets needing a review ? I am trying to deal with some of them, and it would be nice to sort out the tickets which actually need one, those which are to be forgotten until some external event takes place, and those that are totally abandonned :-)

Nathann

[nthiery]

Done.

[nthiery]

lrcalc 1.1.6 beta release (for Anders)

[nthiery]

After discussion with Anders, I have rolled a new (beta) release of lrcalc, based on:

• lrcalc-1.1.5.tar.gz
• schmult-1.1.5.tar.gz
• an autotools build system

which Anders will later polish into an official 1.1.6 release. The spkg is basically identical to the previous 1.1.5b, except for some updated files in the lrcalc sources (README/AUTHOR/TODO), and the header installation configuration. Anders gave his +1 on the changes. I have installed the spkg together with the patch on sage 4.7.2 and sage-5.0.prealpha0, and in both cases, all test pass.

The patch is unchanged and has a positive review.

Anne (and others!): could you try out the new spkg? If it works fine for you, I guess we can set a positive review. Feel free to untar the spkg, and check the files src/README, src/AUTHORS, src/TODO, src/NEWS.

Cheers,

Nicolas

[aschilling]

I tried out the new package and ran the tests. Everything seemed to work. Positive review.

Anne

[jdemeyer]

I get a warning building the documentation (this is without the optional spkg installed):

/mnt/usb1/scratch/jdemeyer/merger/sage-5.0.beta1/devel/sage/doc/en/reference/sage/libs/lrcalc/lrcalc.rst:7: WARNING: autodoc can't import/find module 'sage.libs.lrcalc.lrcalc', it reported error: "No module named lrcalc", please check your spelling and sys.path

[nthiery]

Replying to jdemeyer:

I get a warning building the documentation (this is without the optional spkg installed):

/mnt/usb1/scratch/jdemeyer/merger/sage-5.0.beta1/devel/sage/doc/en/reference/sage/libs/lrcalc/lrcalc.rst:7: WARNING: autodoc can't import/find module 'sage.libs.lrcalc.lrcalc', it reported error: "No module named lrcalc", please check your spelling and sys.path

Yikes. Is there an idiom to conditionally include a piece of documentation in the reference manual depending on the presence or not of a spkg? (analogue of the is_package_installed in module.list.py).

If not, may we just kill the beast, and make lrcalc right away a standard spkg (#11563)?

Cheers,

[jdemeyer]

Maybe this should be discussed on sage-devel? I don't have an opinion either way.

[nthiery]

Replying to jdemeyer:

Maybe this should be discussed on sage-devel? I don't have an opinion either way.

Ok, I just relaunched the discussion there:

​http://groups.google.com/group/sage-devel/msg/9a7b4987ae974b2b

[slabbe]

Yikes. Is there an idiom to conditionally include a piece of documentation in the reference manual depending on the presence or not of a spkg? (analogue of the is_package_installed in module.list.py).

If you don't include the documentation in the reference manual (by that, I mean no modification to the file doc/en/reference/libs.rst ), does the warning still shows up when building the documentation?

What is the status of the documentation for other optional packages?

[nthiery]

Salut Sébastien!

Replying to slabbe:

Yikes. Is there an idiom to conditionally include a piece of documentation in the reference manual depending on the presence or not of a spkg? (analogue of the is_package_installed in module.list.py).

If you don't include the documentation in the reference manual (by that, I mean no modification to the file doc/en/reference/libs.rst ), does the warning still shows up when building the documentation?

I guess not, so worst case that will be the way to go. But not having the documentation linked to does not feel good.

What is the status of the documentation for other optional packages?

I did not find that many, and as far as I could see, the documentation was not linked to ...

Cheers,

[nthiery]

Oh well, I just stripped the link to the lrcalc from libs.rst. Ugly, but if that's the price to pay to abide to the rules ...

Jeroen: I am allowing myself to put back the positive review to catch your attention. Feel free to unset it if you find it innappropriate.

[jdemeyer]

One more thing: the spkg should follow the format of ​http://sagemath.org/doc/developer/producing_spkgs.html#the-file-spkg-txt. In particular, the Changelog should be formatted differently.

[nthiery]

Final spkg

[nthiery]

Replying to jdemeyer:

One more thing: the spkg should follow the format of ​http://sagemath.org/doc/developer/producing_spkgs.html#the-file-spkg-txt. In particular, the Changelog should be formatted differently.

Oops. Thanks for spotting this. I updated the spkg accordingly. Do you mind reviewing the new SPKG.txt?

[jdemeyer]

Actually, the documentation still doesn't work:

/mnt/usb1/scratch/jdemeyer/merger/sage-5.0.beta2/devel/sage/doc/en/reference/sage/libs/lrcalc/lrcalc.rst:7: WARNING: autodoc can't import/find module 'sage.libs.lrcalc.lrcalc', it reported error: "No module named lrcalc", please check your spelling and sys.path
/mnt/usb1/scratch/jdemeyer/merger/sage-5.0.beta2/local/lib/python2.7/site-packages/sage/misc/edit_module.py:docstring of sage.misc.edit_module.edit_devel:14: WARNING: Inline interpreted text or phrase reference start-string without end-string.

[nthiery]

Replying to jdemeyer:

Actually, the documentation still doesn't work:

/mnt/usb1/scratch/jdemeyer/merger/sage-5.0.beta2/devel/sage/doc/en/reference/sage/libs/lrcalc/lrcalc.rst:7: WARNING: autodoc can't import/find module 'sage.libs.lrcalc.lrcalc', it reported error: "No module named lrcalc", please check your spelling and sys.path
/mnt/usb1/scratch/jdemeyer/merger/sage-5.0.beta2/local/lib/python2.7/site-packages/sage/misc/edit_module.py:docstring of sage.misc.edit_module.edit_devel:14: WARNING: Inline interpreted text or phrase reference start-string without end-string.

I don't understand. Why is this lrcalc.rst file generated, whereas no such file is generated for, e.g. the analoguous sage/numerical/backends/coin_backend.pyx?

[jhpalmieri]

When I click on trac_10333-lrcalc-final.patch​, it still has a line adding sage/libs/lrcalc/lrcalc to doc/en/reference/libs.rst. That's presumably the problem.

[aschilling]

Replying to jhpalmieri:

When I click on trac_10333-lrcalc-final.patch​, it still has a line adding sage/libs/lrcalc/lrcalc to doc/en/reference/libs.rst. That's presumably the problem.

Nicolas, could you please also put the updated patch on the sage-combinat server.

Thanks,

Anne

[nthiery]

Replying to jhpalmieri:

When I click on trac_10333-lrcalc-final.patch​, it still has a line adding sage/libs/lrcalc/lrcalc to doc/en/reference/libs.rst. That's presumably the problem.

What's the heck? How did I do this?

Oh well; fixed. Thanks for spotting it!

[nthiery]

Nicolas, could you please also put the updated patch on the sage-combinat server.

Done.

[nthiery]

Jeroen: I let you set back the positive review. Sorry for all the back and forth.

[aschilling]

Replying to nthiery:

Jeroen: I let you set back the positive review. Sorry for all the back and forth.

The line referring to doc/en/reference/libs.rst seems to be gone now, so I set this back to positive review. Hopefully it will work now!

Anne

[nthiery]

Yippee :-)

Thanks everyone!