An interface to Anders Buch’s Littlewood-Richardson Calculator lrcalc

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/.

The following describes the Sage interface to this library.

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])
2

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

sage: lrcalc.mult([2,1], [2,1])
{[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}

Same product, but include only partitions with at most 3 rows. This corresponds to computing in the representation ring of gl(3):

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

We can also compute the fusion product, here for sl(3) and level 2:

sage: lrcalc.mult([3,2,1], [3,2,1], 3,2)
{[4, 4, 4]: 1, [5, 4, 3]: 1}

Compute the expansion of a skew Schur function:

sage: lrcalc.skew([3,2,1],[2,1])
{[3]: 1, [1, 1, 1]: 1, [2, 1]: 2}

Compute the coproduct of a Schur function:

sage: lrcalc.coprod([3,2,1])
{([2, 1, 1], [1, 1]): 1, ([3, 1, 1], [1]): 1, ([2, 1], [3]): 1, ([2, 1, 1], [2]): 1, ([3, 2, 1], []): 1,
([3, 1], [2]): 1, ([3, 2], [1]): 1, ([2, 1], [2, 1]): 2, ([1, 1, 1], [2, 1]): 1, ([2, 2], [2]): 1, ([2, 2, 1], [1]): 1,
([2, 2], [1, 1]): 1, ([3, 1], [1, 1]): 1}

Multiply two Schubert polynomials:

sage: lrcalc.mult_schubert([4,2,1,3], [1,4,2,5,3])
{[5, 4, 1, 2, 3]: 1, [5, 3, 1, 4, 2]: 1, [4, 5, 1, 3, 2]: 1, [6, 2, 1, 4, 3, 5]: 1}

Same product, but include only permutations of 5 elements in the result. This corresponds to computing in the cohomology ring of Fl(5):

sage: lrcalc.mult_schubert([4,2,1,3], [1,4,2,5,3], 5)
{[5, 4, 1, 2, 3]: 1, [5, 3, 1, 4, 2]: 1, [4, 5, 1, 3, 2]: 1}

List all Littlewood-Richardson tableaux of skew shape \(\mu/\nu\); in this example \(\mu=[3,2,1]\) and \(\nu=[2,1]\). Specifying a third entry \(maxrows\) restricts the alphabet to \(\{1,2,\ldots,maxrows\}\):

sage: list(lrcalc.lrskew([3,2,1],[2,1]))
[[[None, None, 1], [None, 1], [1]], [[None, None, 1], [None, 1], [2]],
[[None, None, 1], [None, 2], [1]], [[None, None, 1], [None, 2], [3]]]

sage: list(lrcalc.lrskew([3,2,1],[2,1],maxrows=2))
[[[None, None, 1], [None, 1], [1]], [[None, None, 1], [None, 1], [2]], [[None, None, 1], [None, 2], [1]]]

Todo

use this library in the SymmetricFunctions code, to make it easy to apply it to linear combinations of Schur functions.

See also

Underlying algorithmic in lrcalc

Here is some additional information regarding the main low-level C-functions in \(lrcalc\). Given two partitions outer and inner with inner contained in outer, the function:

skewtab *st_new(vector *outer, vector *inner, vector *conts, int maxrows)

constructs and returns the (lexicographically) first LR skew tableau of shape outer / inner. Further restrictions can be imposed using conts and maxrows.

Namely, the integer maxrows is a bound on the integers that can be put in the tableau. The name is chosen because this will limit the partitions in the output of skew() or mult() to partitions with at most this number of rows.

The vector conts is the content of an empty tableau(!!). More precisely, this vector is added to the usual content of a tableau whenever the content is needed. This affects which tableaux are considered LR tableaux (see mult() below). conts may also be the NULL pointer, in which case nothing is added.

The other function:

int *st_next(skewtab *st)

computes in place the (lexicographically) next skew tableau with the same constraints, or returns 0 if st is the last one.

For a first example, see the skew() function code in the lrcalc source code. We want to compute a skew schur function, so create a skew LR tableau of the appropriate shape with st_new (with conts = NULL), then iterate through all the LR tableaux with st_next(). For each skew tableau, we use that st->conts is the content of the skew tableau, find this shape in the res hash table and add one to the value.

For a second example, see mult(vector *sh1, vector *sh2, maxrows). Here we call st_new() with the shape sh1 / (0) and use sh2 as the conts argument. The effect of using sh2 in this way is that st_next will iterate through semistandard tableaux \(T\) of shape sh1 such that the following tableau:

     111111
     22222    <--- minimal tableau of shape sh2
     333
*****
**T**
****
**

is a LR skew tableau, and st->conts contains the content of the combined tableaux.

More generally, st_new(outer, inner, conts, maxrows) and st_next can be used to compute the Schur expansion of the product S_{outer/inner} * S_conts, restricted to partitions with at most maxrows rows.

AUTHORS:

  • Mike Hansen (2010): core of the interface
  • Anne Schilling, Nicolas M. Thiéry, and Anders Buch (2011): fusion product, iterating through LR tableaux, finalization, documentation
sage.libs.lrcalc.lrcalc.coprod(part, all=0)

Compute the coproduct of a Schur function.

Return a linear combination of pairs of partitions representing the coproduct of the Schur function given by the partition part.

INPUT:

  • part – a partition.
  • all – an integer.

If all is non-zero then all terms are included in the result. If all is zero, then only pairs of partitions (part1, part2) for which the weight of part1 is greater than or equal to the weight of part2 are included; the rest of the coefficients are redundant because Littlewood-Richardson coefficients are symmetric.

EXAMPLES:

sage: from sage.libs.lrcalc.lrcalc import coprod
sage: sorted(coprod([2,1]).items())
[(([1, 1], [1]), 1), (([2], [1]), 1), (([2, 1], []), 1)]
sage.libs.lrcalc.lrcalc.lrcoef(outer, inner1, inner2)

Compute a single Littlewood-Richardson coefficient.

Return the coefficient of outer in the product of the Schur functions indexed by inner1 and inner2.

INPUT:

  • outer – a partition (weakly decreasing list of non-negative integers).
  • inner1 – a partition.
  • inner2 – a partition.

Note

This function converts its inputs into Partition()‘s. If you don’t need these checks and your inputs are valid, then you can use lrcoef_unsafe().

EXAMPLES:

sage: from sage.libs.lrcalc.lrcalc import lrcoef
sage: lrcoef([3,2,1], [2,1], [2,1])
2
sage: lrcoef([3,3], [2,1], [2,1])
1
sage: lrcoef([2,1,1,1,1], [2,1], [2,1])
0
sage.libs.lrcalc.lrcalc.lrcoef_unsafe(outer, inner1, inner2)

Compute a single Littlewood-Richardson coefficient.

Return the coefficient of outer in the product of the Schur functions indexed by inner1 and inner2.

INPUT:

  • outer – a partition (weakly decreasing list of non-negative integers).
  • inner1 – a partition.
  • inner2 – a partition.

Warning

This function does not do any check on its input. If you want to use a safer version, use lrcoef().

EXAMPLES:

sage: from sage.libs.lrcalc.lrcalc import lrcoef_unsafe
sage: lrcoef_unsafe([3,2,1], [2,1], [2,1])
2
sage: lrcoef_unsafe([3,3], [2,1], [2,1])
1
sage: lrcoef_unsafe([2,1,1,1,1], [2,1], [2,1])
0
sage.libs.lrcalc.lrcalc.lrskew(outer, inner, weight=None, maxrows=0)

Return the skew LR tableaux of shape outer / inner.

INPUT:

  • outer – a partition.
  • inner – a partition.
  • weight – a partition (optional).
  • maxrows – an integer (optional).

OUTPUT: a list of SkewTableau`x. This will change to an iterator over such skew tableaux once Cython will support the ``yield` statement. Specifying a third entry \(maxrows\) restricts the alphabet to \(\{1,2,\ldots,maxrows\}\). Specifying \(weight\) returns only those tableaux of given content/weight.

EXAMPLES:

sage: from sage.libs.lrcalc.lrcalc import lrskew
sage: for st in lrskew([3,2,1],[2]):
...       st.pp()
.  .  1
1  1
2
.  .  1
1  2
2
.  .  1
1  2
3

sage: for st in lrskew([3,2,1],[2], maxrows=2):
...       st.pp()
.  .  1
1  1
2
.  .  1
1  2
2

sage: lrskew([3,2,1],[2], weight=[3,1])
[[[None, None, 1], [1, 1], [2]]]
sage.libs.lrcalc.lrcalc.mult(part1, part2, maxrows=None, level=None)

Compute a product of two Schur functions.

Return the product of the Schur functions indexed by the partitions part1 and part2.

INPUT:

  • part1 – a partition.
  • part2 – a partition.
  • maxrows – an integer or None.
  • level – an integer or None.

If maxrows is specified, then only partitions with at most this number of rows is included in the result.

If both maxrows and level are specified, then the function calculates the fusion product for \(sl(\mathrm{maxrows})\) of the given level.

EXAMPLES:

sage: from sage.libs.lrcalc.lrcalc import mult
sage: mult([2],[])
{[2]: 1}
sage: sorted(mult([2],[2]).items())
[([2, 2], 1), ([3, 1], 1), ([4], 1)]
sage: sorted(mult([2,1],[2,1]).items())
[([2, 2, 1, 1], 1), ([2, 2, 2], 1), ([3, 1, 1, 1], 1), ([3, 2, 1], 2), ([3, 3], 1), ([4, 1, 1], 1), ([4, 2], 1)]
sage: sorted(mult([2,1],[2,1],maxrows=2).items())
[([3, 3], 1), ([4, 2], 1)]
sage: mult([2,1],[3,2,1],3)
{[3, 3, 3]: 1, [5, 2, 2]: 1, [5, 3, 1]: 1, [4, 4, 1]: 1, [4, 3, 2]: 2}
sage: mult([2,1],[2,1],3,3)
{[3, 3]: 1, [3, 2, 1]: 2, [2, 2, 2]: 1, [4, 1, 1]: 1}
sage: mult([2,1],[2,1],None,3)
Traceback (most recent call last):
...
ValueError: maxrows needs to be specified if you specify the level
sage.libs.lrcalc.lrcalc.mult_schubert(w1, w2, rank=0)

Compute a product of two Schubert polynomials.

Return a linear combination of permutations representing the product of the Schubert polynomials indexed by the permutations w1 and w2.

INPUT:

  • w1 – a permutation.
  • w2 – a permutation.
  • rank – an integer.

If rank is non-zero, then only permutations from the symmetric group \(S(\mathrm{rank})\) are included in the result.

EXAMPLES:

sage: from sage.libs.lrcalc.lrcalc import mult_schubert
sage: result = mult_schubert([3, 1, 5, 2, 4], [3, 5, 2, 1, 4])
sage: sorted(result.items())
[([5, 4, 6, 1, 2, 3], 1), ([5, 6, 3, 1, 2, 4], 1),
 ([5, 7, 2, 1, 3, 4, 6], 1), ([6, 3, 5, 1, 2, 4], 1),
 ([6, 4, 3, 1, 2, 5], 1), ([6, 5, 2, 1, 3, 4], 1),
 ([7, 3, 4, 1, 2, 5, 6], 1), ([7, 4, 2, 1, 3, 5, 6], 1)]
sage.libs.lrcalc.lrcalc.skew(outer, inner, maxrows=0)

Compute the Schur expansion of a skew Schur function.

Return a linear combination of partitions representing the Schur function of the skew Young diagram outer / inner, consisting of boxes in the partition outer that are not in inner.

INPUT:

  • outer – a partition.
  • inner – a partition.
  • maxrows – an integer or None.

If maxrows is specified, then only partitions with at most this number of rows are included in the result.

EXAMPLES:

sage: from sage.libs.lrcalc.lrcalc import skew
sage: sorted(skew([2,1],[1]).items())
[([1, 1], 1), ([2], 1)]
sage.libs.lrcalc.lrcalc.test_iterable_to_vector(it)

A wrapper function for the cdef function iterable_to_vector and vector_to_list, to test that they are working correctly.

EXAMPLES:

sage: from sage.libs.lrcalc.lrcalc import test_iterable_to_vector
sage: x = test_iterable_to_vector([3,2,1]); x
[3, 2, 1]
sage.libs.lrcalc.lrcalc.test_skewtab_to_SkewTableau(outer, inner)

A wrapper function for the cdef function skewtab_to_SkewTableau for testing purposes.

It constructs the first LR skew tableau of shape outer/inner as an lrcalc skewtab, and converts it to a SkewTableau.

EXAMPLES:

sage: from sage.libs.lrcalc.lrcalc import test_skewtab_to_SkewTableau
sage: test_skewtab_to_SkewTableau([3,2,1],[])
[[1, 1, 1], [2, 2], [3]]
sage: test_skewtab_to_SkewTableau([4,3,2,1],[1,1]).pp()
.  1  1  1
.  2  2
1  3
2

Previous topic

Rubinstein’s lcalc library

Next topic

Victor Shoup’s NTL C++ Library

This Page