Generating Series¶
This file makes a number of extensions to lazy power series by endowing them with some semantic content for how they’re to be interpreted.
This code is based on the work of Ralf Hemmecke and Martin Rubey’s AldorCombinat, which can be found at http://www.risc.unilinz.ac.at/people/hemmecke/aldor/combinat/index.html. In particular, the relevant section for this file can be found at http://www.risc.unilinz.ac.at/people/hemmecke/AldorCombinat/combinatse10.html. One notable difference is that we use powersum symmetric functions as the coefficients of our cycle index series.
REFERENCES:
[BLL]  (1, 2, 3, 4) F. Bergeron, G. Labelle, and P. Leroux. “Combinatorial species and treelike structures”. Encyclopedia of Mathematics and its Applications, vol. 67, Cambridge Univ. Press. 1998. 
[BLLIntro]  Francois Bergeron, Gilbert Labelle, and Pierre Leroux. “Introduction to the Theory of Species of Structures”, March 14, 2008. 

class
sage.combinat.species.generating_series.
CycleIndexSeries
(A, stream=None, order=None, aorder=None, aorder_changed=True, is_initialized=False, name=None)¶ Bases:
sage.combinat.species.series.LazyPowerSeries

arithmetic_product
(g, check_input=True)¶ Return the arithmetic product of
self
withg
.For species \(M\) and \(N\) such that \(M[\\varnothing] = N[\\varnothing] = \\varnothing\), their arithmetic product is the species \(M \\boxdot N\) of “\(M\)assemblies of cloned \(N\)structures”. This operation is defined and several examples are given in [MM].
The cycle index series for \(M \\boxdot N\) can be computed in terms of the component series \(Z_M\) and \(Z_N\), as implemented in this method.
INPUT:
g
– a cycle index series having the same parent asself
.check_input
– (default:True
) a Boolean which, when set toFalse
, will cause input checks to be skipped.
OUTPUT:
The arithmetic product of
self
withg
. This is a cycle index series defined in terms ofself
andg
such that ifself
andg
are the cycle index series of two species \(M\) and \(N\), their arithmetic product is the cycle index series of the species \(M \\boxdot N\).EXAMPLES:
For \(C\) the species of (oriented) cycles and \(L_{+}\) the species of nonempty linear orders, \(C \\boxdot L_{+}\) corresponds to the species of “regular octopuses”; a \((C \\boxdot L_{+})\)structure is a cycle of some length, each of whose elements is an ordered list of a length which is consistent for all the lists in the structure.
sage: C = species.CycleSpecies().cycle_index_series() sage: Lplus = species.LinearOrderSpecies(min=1).cycle_index_series() sage: RegularOctopuses = C.arithmetic_product(Lplus) sage: RegOctSpeciesSeq = RegularOctopuses.generating_series().counts(8) sage: RegOctSpeciesSeq [0, 1, 3, 8, 42, 144, 1440, 5760]
It is shown in [MM] that the exponential generating function for regular octopuses satisfies \((C \\boxdot L_{+}) (x) = \\sum_{n \geq 1} \\sigma (n) (n  1)! \\frac{x^{n}}{n!}\) (where \(\\sigma (n)\) is the sum of the divisors of \(n\)).
sage: RegOctDirectSeq = [0] + [sum(divisors(i))*factorial(i1) for i in range(1,8)] sage: RegOctDirectSeq == RegOctSpeciesSeq True
AUTHORS:
 Andrew GainerDewar (2013)
REFERENCES:
[MM] (1, 2) M. Maia and M. Mendez. “On the arithmetic product of combinatorial species”. Discrete Mathematics, vol. 308, issue 23, 2008, pp. 54075427. arXiv math/0503436v2.

coefficient_cycle_type
(t)¶ Returns the coefficient of a cycle type
t
inself
.EXAMPLES:
sage: from sage.combinat.species.generating_series import CycleIndexSeriesRing sage: p = SymmetricFunctions(QQ).power() sage: CIS = CycleIndexSeriesRing(QQ) sage: f = CIS([0, p([1]), 2*p([1,1]),3*p([2,1])]) sage: f.coefficient_cycle_type([1]) 1 sage: f.coefficient_cycle_type([1,1]) 2 sage: f.coefficient_cycle_type([2,1]) 3

compositional_inverse
()¶ Return the compositional inverse of
self
if possible.(Specifically, if
self
is of the form \(0 + p_{1} + \dots\).)The compositional inverse is the inverse with respect to plethystic substitution. This is the operation on cycle index series which corresponds to substitution, a.k.a. partitional composition, on the level of species. See Section 2.2 of [BLL] for a definition of this operation.
EXAMPLES:
sage: Eplus = species.SetSpecies(min=1).cycle_index_series() sage: Eplus(Eplus.compositional_inverse()).coefficients(8) [0, p[1], 0, 0, 0, 0, 0, 0]
ALGORITHM:
Let \(F\) be a species satisfying \(F = 0 + X + F_2 + F_3 + \dots\) for \(X\) the species of singletons. (Equivalently, \(\lvert F[\varnothing] \rvert = 0\) and \(\lvert F[\{1\}] \rvert = 1\).) Then there exists a (virtual) species \(G\) satisfying \(F \circ G = G \circ F = X\).
It follows that \((F  X) \circ G = F \circ G  X \circ G = X  G\). Rearranging, we obtain the recursive equation \(G = X  (F  X) \circ G\), which can be solved using iterative methods.
Warning
This algorithm is functional but can be very slow. Use with caution!
See also
The compositional inverse \(\Omega\) of the species \(E_{+}\) of nonempty sets can be handled much more efficiently using specialized methods. See
LogarithmCycleIndexSeries()
AUTHORS:
 Andrew GainerDewar

count
(t)¶ Return the number of structures corresponding to a certain cycle type
t
.EXAMPLES:
sage: from sage.combinat.species.generating_series import CycleIndexSeriesRing sage: p = SymmetricFunctions(QQ).power() sage: CIS = CycleIndexSeriesRing(QQ) sage: f = CIS([0, p([1]), 2*p([1,1]), 3*p([2,1])]) sage: f.count([1]) 1 sage: f.count([1,1]) 4 sage: f.count([2,1]) 6

derivative
(order=1)¶ Return the speciestheoretic nth derivative of
self
, where n isorder
.For a cycle index series \(F (p_{1}, p_{2}, p_{3}, \dots)\), its derivative is the cycle index series \(F' = D_{p_{1}} F\) (that is, the formal derivative of \(F\) with respect to the variable \(p_{1}\)).
If \(F\) is the cycle index series of a species \(S\) then \(F'\) is the cycle index series of an associated species \(S'\) of \(S\)structures with a “hole”.
EXAMPLES:
The species \(E\) of sets satisfies the relationship \(E' = E\):
sage: E = species.SetSpecies().cycle_index_series() sage: E.coefficients(8) == E.derivative().coefficients(8) True
The species \(C\) of cyclic orderings and the species \(L\) of linear orderings satisfy the relationship \(C' = L\):
sage: C = species.CycleSpecies().cycle_index_series() sage: L = species.LinearOrderSpecies().cycle_index_series() sage: L.coefficients(8) == C.derivative().coefficients(8) True

expand_as_sf
(n, alphabet='x')¶ Returns the expansion of a cycle index series as a symmetric function in
n
variables.Specifically, this returns a
LazyPowerSeries
whose ith term is obtained by callingexpand()
on the ith term ofself
.This relies on the (standard) interpretation of a cycle index series as a symmetric function in the power sum basis.
INPUT:
self
– a cycle index seriesn
– a positive integeralphabet
– a variable for the expansion (default: \(x\))
EXAMPLES:
sage: from sage.combinat.species.set_species import SetSpecies sage: SetSpecies().cycle_index_series().expand_as_sf(2).coefficients(4) [1, x0 + x1, x0^2 + x0*x1 + x1^2, x0^3 + x0^2*x1 + x0*x1^2 + x1^3]

exponential
()¶ Return the speciestheoretic exponential of
self
.For a cycle index \(Z_{F}\) of a species \(F\), its exponential is the cycle index series \(Z_{E} \\circ Z_{F}\), where \(Z_{E}\) is the
ExponentialCycleIndexSeries()
.The exponential \(Z_{E} \circ Z_{F}\) is then the cycle index series of the species \(E \\circ F\) of “sets of \(F\)structures”.
EXAMPLES:
Let \(BT\) be the species of binary trees, \(BF\) the species of binary forests, and \(E\) the species of sets. Then we have \(BF = E \circ BT\):
sage: BT = species.BinaryTreeSpecies().cycle_index_series() sage: BF = species.BinaryForestSpecies().cycle_index_series() sage: BT.exponential().isotype_generating_series().coefficients(8) == BF.isotype_generating_series().coefficients(8) True

functorial_composition
(g)¶ Returns the functorial composition of
self
andg
.If \(F\) and \(G\) are species, their functorial composition is the species \(F \Box G\) obtained by setting \((F \Box G) [A] = F[ G[A] ]\). In other words, an \((F \Box G)\)structure on a set \(A\) of labels is an \(F\)structure whose labels are the set of all \(G\)structures on \(A\).
It can be shown (as in section 2.2 of [BLL]) that there is a corresponding operation on cycle indices:
\[Z_{F} \Box Z_{G} = \sum_{n \geq 0} \frac{1}{n!} \sum_{\sigma \in \mathfrak{S}_{n}} \operatorname{fix} F[ (G[\sigma])_{1}, (G[\sigma])_{2}, \dots ] \, p_{1}^{\sigma_{1}} p_{2}^{\sigma_{2}} \dots.\]This method implements that operation on cycle index series.
EXAMPLES:
The species \(G\) of simple graphs can be expressed in terms of a functorial composition: \(G = \mathfrak{p} \Box \mathfrak{p}_{2}\), where \(\mathfrak{p}\) is the
SubsetSpecies
. This is how it is implemented inSimpleGraphSpecies()
:sage: S = species.SimpleGraphSpecies() sage: S.cycle_index_series().coefficients(5) [p[], p[1], p[1, 1] + p[2], 4/3*p[1, 1, 1] + 2*p[2, 1] + 2/3*p[3], 8/3*p[1, 1, 1, 1] + 4*p[2, 1, 1] + 2*p[2, 2] + 4/3*p[3, 1] + p[4]]

generating_series
()¶ EXAMPLES:
sage: P = species.PartitionSpecies() sage: cis = P.cycle_index_series() sage: f = cis.generating_series() sage: f.coefficients(5) [1, 1, 1, 5/6, 5/8]

integral
(*args)¶ Given a cycle index \(G\), it is not in general possible to recover a single cycle index \(F\) such that \(F' = G\) (even up to addition of a constant term).
More broadly, it may be the case that there are many nonisomorphic species \(S\) such that \(S' = T\) for a given species \(T\). For example, the species \(3 C_{3}\) of 3cycles from three distinct classes and the species \(X^{3}\) of 3sets are not isomorphic, but \((3 C_{3})' = (X^{3})' = 3 X^{2}\).
EXAMPLES:
sage: C3 = species.CycleSpecies(size=3).cycle_index_series() sage: X = species.SingletonSpecies().cycle_index_series() sage: (3*C3).derivative().coefficients(8) == (3*X^2).coefficients(8) True sage: (X^3).derivative().coefficients(8) == (3*X^2).coefficients(8) True
Warning
This method has no implementation and exists only to prevent you from doing something strange. Calling it raises a
NotImplementedError
!

isotype_generating_series
()¶ EXAMPLES:
sage: P = species.PermutationSpecies() sage: cis = P.cycle_index_series() sage: f = cis.isotype_generating_series() sage: f.coefficients(10) [1, 1, 2, 3, 5, 7, 11, 15, 22, 30]

logarithm
()¶ Return the combinatorial logarithm of
self
.For a cycle index \(Z_{F}\) of a species \(F\), its logarithm is the cycle index series \(Z_{\Omega} \circ Z_{F}\), where \(Z_{\Omega}\) is the
LogarithmCycleIndexSeries()
.The logarithm \(Z_{\Omega} \circ Z_{F}\) is then the cycle index series of the (virtual) species \(\Omega \circ F\) of “connected \(F\)structures”. In particular, if \(F = E^{+} \circ G\) for \(E^{+}\) the species of nonempty sets and \(G\) some other species, then \(\Omega \circ F = G\).
EXAMPLES:
Let \(G\) be the species of nonempty graphs and \(CG\) be the species of nonempty connected graphs. Then \(G = E^{+} \circ CG\), so \(CG = \Omega \circ G\):
sage: G = species.SimpleGraphSpecies().cycle_index_series()  1 sage: from sage.combinat.species.generating_series import LogarithmCycleIndexSeries sage: CG = LogarithmCycleIndexSeries().compose(G) sage: CG.isotype_generating_series().coefficients(8) [0, 1, 1, 2, 6, 21, 112, 853]

pointing
()¶ Return the speciestheoretic pointing of
self
.For a cycle index \(F\), its pointing is the cycle index series \(F^{\bullet} = p_{1} \cdot F'\).
If \(F\) is the cycle index series of a species \(S\) then \(F^{\bullet}\) is the cycle index series of an associated species \(S^{\bullet}\) of \(S\)structures with a marked “root”.
EXAMPLES:
The species \(E^{\bullet}\) of “pointed sets” satisfies \(E^{\bullet} = X \cdot E\):
sage: E = species.SetSpecies().cycle_index_series() sage: X = species.SingletonSpecies().cycle_index_series() sage: E.pointing().coefficients(8) == (X*E).coefficients(8) True

stretch
(k)¶ Return the stretch of the cycle index series
self
by a positive integer \(k\).If
\[f = \sum_{n=0}^{\infty} f_n(p_1, p_2, p_3, \ldots ),\]then the stretch \(g\) of \(f\) by \(k\) is
\[g = \sum_{n=0}^{\infty} f_n(p_k, p_{2k}, p_{3k}, \ldots ).\]EXAMPLES:
sage: from sage.combinat.species.generating_series import CycleIndexSeriesRing sage: p = SymmetricFunctions(QQ).power() sage: CIS = CycleIndexSeriesRing(QQ) sage: f = CIS([p([]), p([1]), p([2]), p.zero()]) sage: f.stretch(3).coefficients(10) [p[], 0, 0, p[3], 0, 0, p[6], 0, 0, 0]

weighted_composition
(y_species)¶ Returns the composition of this cycle index series with the cycle index series of y_species where y_species is a weighted species.
Note that this is basically the same algorithm as composition except we can not use the optimization that the powering of cycle index series commutes with ‘stretching’.
EXAMPLES:
sage: E = species.SetSpecies(); C = species.CycleSpecies() sage: E_cis = E.cycle_index_series() sage: E_cis.weighted_composition(C).coefficients(4) [p[], p[1], p[1, 1] + p[2], p[1, 1, 1] + p[2, 1] + p[3]] sage: E(C).cycle_index_series().coefficients(4) [p[], p[1], p[1, 1] + p[2], p[1, 1, 1] + p[2, 1] + p[3]]


sage.combinat.species.generating_series.
CycleIndexSeriesRing
(R)¶ Return the ring of cycle index series over
R
.This is the ring of formal power series \(\Lambda[x]\), where \(\Lambda\) is the ring of symmetric functions over
R
in the \(p\)basis. Its purpose is to house the cycle index series of species (in a somewhat nonstandard notation tailored to Sage): If \(F\) is a species, then the cycle index series of \(F\) is defined to be the formal power series\[\sum_{n \geq 0} \frac{1}{n!} (\sum_{\sigma \in S_n} \operatorname{fix} F[\sigma] \prod_{z \text{ is a cycle of } \sigma} p_{\text{length of } z}) x^n \in \Lambda_\QQ [x],\]where \(\operatorname{fix} F[\sigma]\) denotes the number of fixed points of the permutation \(F[\sigma]\) of \(F[n]\). We notice that this power series is “equigraded” (meaning that its \(x^n\)coefficient is homogeneous of degree \(n\)). A more standard convention in combinatorics would be to use \(x_i\) instead of \(p_i\), and drop the \(x\) (that is, evaluate the above power series at \(x = 1\)); but this would be more difficult to implement in Sage, as it would be an element of a power series ring in infinitely many variables.
Note that it is just a
LazyPowerSeriesRing
(whose base ring is \(\Lambda\)) whose elements have some extra methods.EXAMPLES:
sage: from sage.combinat.species.generating_series import CycleIndexSeriesRing sage: R = CycleIndexSeriesRing(QQ); R Cycle Index Series Ring over Symmetric Functions over Rational Field in the powersum basis sage: R([1]).coefficients(4) # This is not combinatorially ....: # meaningful. [1, 1, 1, 1]

class
sage.combinat.species.generating_series.
CycleIndexSeriesRing_class
(R)¶ Bases:
sage.combinat.species.series.LazyPowerSeriesRing
EXAMPLES:
sage: from sage.combinat.species.generating_series import CycleIndexSeriesRing sage: R = CycleIndexSeriesRing(QQ); R Cycle Index Series Ring over Symmetric Functions over Rational Field in the powersum basis sage: R == loads(dumps(R)) True

sage.combinat.species.generating_series.
ExponentialCycleIndexSeries
(R=Rational Field)¶ Return the cycle index series of the species \(E\) of sets.
This cycle index satisfies
\[Z_{E} = \sum_{n \geq 0} \sum_{\lambda \vdash n} \frac{p_{\lambda}}{z_{\lambda}}.\]EXAMPLES:
sage: from sage.combinat.species.generating_series import ExponentialCycleIndexSeries sage: ExponentialCycleIndexSeries().coefficients(5) [p[], p[1], 1/2*p[1, 1] + 1/2*p[2], 1/6*p[1, 1, 1] + 1/2*p[2, 1] + 1/3*p[3], 1/24*p[1, 1, 1, 1] + 1/4*p[2, 1, 1] + 1/8*p[2, 2] + 1/3*p[3, 1] + 1/4*p[4]]

class
sage.combinat.species.generating_series.
ExponentialGeneratingSeries
(A, stream=None, order=None, aorder=None, aorder_changed=True, is_initialized=False, name=None)¶ Bases:
sage.combinat.species.series.LazyPowerSeries

count
(n)¶ Return the number of structures of size
n
.EXAMPLES:
sage: from sage.combinat.species.generating_series import ExponentialGeneratingSeriesRing sage: R = ExponentialGeneratingSeriesRing(QQ) sage: f = R([1]) sage: [f.count(i) for i in range(7)] [1, 1, 2, 6, 24, 120, 720]

counts
(n)¶ Return the number of structures on a set for size
i
for eachi
inrange(n)
.EXAMPLES:
sage: from sage.combinat.species.generating_series import ExponentialGeneratingSeriesRing sage: R = ExponentialGeneratingSeriesRing(QQ) sage: f = R(range(20)) sage: f.counts(5) [0, 1, 4, 18, 96]

functorial_composition
(y)¶ Return the exponential generating series which is the functorial composition of
self
withy
.If \(f = \sum_{n=0}^{\infty} f_n \frac{x^n}{n!}\) and \(g = \sum_{n=0}^{\infty} g_n \frac{x^n}{n!}\), then functorial composition \(f \Box g\) is defined as
\[f \Box g = \sum_{n=0}^{\infty} f_{g_n} \frac{x^n}{n!}\]REFERENCES:
 Section 2.2 of [BLL].
EXAMPLES:
sage: G = species.SimpleGraphSpecies() sage: g = G.generating_series() sage: g.coefficients(10) [1, 1, 1, 4/3, 8/3, 128/15, 2048/45, 131072/315, 2097152/315, 536870912/2835]


sage.combinat.species.generating_series.
ExponentialGeneratingSeriesRing
(R)¶ Return the ring of exponential generating series over
R
.Note that it is just a
LazyPowerSeriesRing
whose elements have some extra methods.EXAMPLES:
sage: from sage.combinat.species.generating_series import ExponentialGeneratingSeriesRing sage: R = ExponentialGeneratingSeriesRing(QQ); R Lazy Power Series Ring over Rational Field sage: R([1]).coefficients(4) [1, 1, 1, 1] sage: R([1]).counts(4) [1, 1, 2, 6]

class
sage.combinat.species.generating_series.
ExponentialGeneratingSeriesRing_class
(R)¶ Bases:
sage.combinat.species.series.LazyPowerSeriesRing
EXAMPLES:
sage: from sage.combinat.species.generating_series import ExponentialGeneratingSeriesRing sage: R = ExponentialGeneratingSeriesRing(QQ) sage: R == loads(dumps(R)) True

sage.combinat.species.generating_series.
LogarithmCycleIndexSeries
(R=Rational Field)¶ Return the cycle index series of the virtual species \(\Omega\), the compositional inverse of the species \(E^{+}\) of nonempty sets.
The notion of virtual species is treated thoroughly in [BLL]. The specific algorithm used here to compute the cycle index of \(\Omega\) is found in [Labelle].
EXAMPLES:
The virtual species \(\Omega\) is ‘properly virtual’, in the sense that its cycle index has negative coefficients:
sage: from sage.combinat.species.generating_series import LogarithmCycleIndexSeries sage: LogarithmCycleIndexSeries().coefficients(4) [0, p[1], 1/2*p[1, 1]  1/2*p[2], 1/3*p[1, 1, 1]  1/3*p[3]]
Its defining property is that \(\Omega \circ E^{+} = E^{+} \circ \Omega = X\) (that is, that composition with \(E^{+}\) in both directions yields the multiplicative identity \(X\)):
sage: Eplus = sage.combinat.species.set_species.SetSpecies(min=1).cycle_index_series() sage: LogarithmCycleIndexSeries().compose(Eplus).coefficients(4) [0, p[1], 0, 0]
REFERENCES:
[Labelle] G. Labelle. “New combinatorial computational methods arising from pseudosingletons.” DMTCS Proceedings 1, 2008.

class
sage.combinat.species.generating_series.
OrdinaryGeneratingSeries
(A, stream=None, order=None, aorder=None, aorder_changed=True, is_initialized=False, name=None)¶ Bases:
sage.combinat.species.series.LazyPowerSeries

count
(n)¶ Return the number of structures on a set of size
n
.EXAMPLES:
sage: from sage.combinat.species.generating_series import OrdinaryGeneratingSeriesRing sage: R = OrdinaryGeneratingSeriesRing(QQ) sage: f = R(range(20)) sage: f.count(10) 10

counts
(n)¶ Return the number of structures on a set for size
i
for eachi
inrange(n)
.EXAMPLES:
sage: from sage.combinat.species.generating_series import OrdinaryGeneratingSeriesRing sage: R = OrdinaryGeneratingSeriesRing(QQ) sage: f = R(range(20)) sage: f.counts(10) [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]


sage.combinat.species.generating_series.
OrdinaryGeneratingSeriesRing
(R)¶ Return the ring of ordinary generating series over
R
.Note that it is just a
LazyPowerSeriesRing
whose elements have some extra methods.EXAMPLES:
sage: from sage.combinat.species.generating_series import OrdinaryGeneratingSeriesRing sage: R = OrdinaryGeneratingSeriesRing(QQ); R Lazy Power Series Ring over Rational Field sage: R([1]).coefficients(4) [1, 1, 1, 1] sage: R([1]).counts(4) [1, 1, 1, 1]

class
sage.combinat.species.generating_series.
OrdinaryGeneratingSeriesRing_class
(R)¶ Bases:
sage.combinat.species.series.LazyPowerSeriesRing
EXAMPLES:
sage: from sage.combinat.species.generating_series import OrdinaryGeneratingSeriesRing sage: R = OrdinaryGeneratingSeriesRing(QQ) sage: R == loads(dumps(R)) True

sage.combinat.species.generating_series.
factorial_gen
()¶ A generator for the factorials starting at 0.
EXAMPLES:
sage: from sage.combinat.species.generating_series import factorial_gen sage: g = factorial_gen() sage: [next(g) for i in range(5)] [1, 1, 2, 6, 24]