Overview of Hecke triangle groups and modular forms for Hecke triangle groups

AUTHORS:

  • Jonas Jermann (2013): initial version

Hecke triangle groups and elements:

  • Hecke triangle group: The Von Dyck group corresponding to the triangle group with angles (pi/2, pi/n, 0) for n=3, 4, 5, ..., generated by the conformal circle inversion S and by the translation T by lambda=2*cos(pi/n). I.e. the subgroup of orientation preserving elements of the triangle group generated by reflections along the boundaries of the above hyperbolic triangle. The group is arithmetic iff n=3, 4, 6, infinity.

    The group elements correspond to matrices over ZZ[lambda], namely the corresponding order in the number field defined by the minimal polynomial of lambda (which embedds into AlgebraicReal accordingly).

    An exact symbolic expression of the corresponding transfinite diameter d (which is used as a formal parameter for Fourier expansion of modular forms) can be obtained. For arithmetic groups the (correct) rational number is returned instead.

    Basic matrices like S, T, U, V(j) are available.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(12)
sage: G
Hecke triangle group for n = 12
sage: G.is_arithmetic()
False
sage: G.dvalue()
e^(2*euler_gamma - 4*pi/(sqrt(6) + sqrt(2)) + psi(19/24) + psi(17/24))
sage: AA(G.lam())
1.9318516525781...?

sage: G = HeckeTriangleGroup(6)
sage: G
Hecke triangle group for n = 6
sage: G.is_arithmetic()
True
sage: G.dvalue()
1/108
sage: AA(G.lam()) == AA(sqrt(3))
True
sage: G.gens()
(
[ 0 -1]  [  1 lam]
[ 1  0], [  0   1]
)
sage: G.U()^3
[ lam   -2]
[   2 -lam]
sage: G.U().parent()
Hecke triangle group for n = 6
sage: G.U().matrix().parent()
Full MatrixSpace of 2 by 2 dense matrices over Maximal Order in Number Field in lam with defining polynomial x^2 - 3 with lam = 1.732050807568878?

  • Decomposition into product of generators: It is possible to decompose any group element into products of generators the S and T. In particular this allows to check whether a given matrix indeed is a group element.

    It also allows one to calculate the automorphy factor of a modular form for the Hecke triangle group for arbitrary arguments.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(6)
sage: G.element_repr_method("basic")
sage: A = G.V(2)*G.V(3)^(-2)
sage: (L, sgn) = A.word_S_T()
sage: L
(S, T^(-2), S, T^(-1), S, T^(-1))
sage: sgn
-1
sage: sgn.parent()
Hecke triangle group for n = 6

sage: G(matrix([[-1, 1+G.lam()],[0, -1]]))
Traceback (most recent call last):
...
TypeError: The matrix is not an element of Hecke triangle group for n = 6, up to equivalence it identifies two nonequivalent points.
sage: G(matrix([[-1, G.lam()],[0, -1]]))
-T^(-1)

sage: G.element_repr_method("basic")

sage: from sage.modular.modform_hecketriangle.space import ModularForms
sage: MF = ModularForms(G, k=4, ep=1)
sage: z = AlgebraicField()(1+i/2)
sage: MF.aut_factor(A, z)
37.62113890008...? + 12.18405525839...?*I

  • Representation of elements: An element can be represented in several ways:

    • As a matrix over the base ring (default)
    • As a product of the generators S and T
    • As a product of basic blocks conjugated by some element

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(n=5)
sage: el = G.S()*G.T(3)*G.S()*G.T(-2)

sage: G.element_repr_method("default")
sage: el
[        -1      2*lam]
[     3*lam -6*lam - 7]

sage: G.element_repr_method("basic")
sage: el
S*T^3*S*T^(-2)

sage: G.element_repr_method("block")
sage: el
-(S*T^3) * (V(4)^2*V(1)^3) * (S*T^3)^(-1)

sage: G.element_repr_method("conj")
sage: el
[-V(4)^2*V(1)^3]

sage: G.element_repr_method("default")

  • Group action on the (extended) upper half plane: The group action of Hecke triangle groups on the (extended) upper half plane (by linear fractional transformations) is implemented. The implementation is not based on a specific upper half plane model but is defined formally (for arbitrary arguments) instead.

    It is possible to determine the group translate of an element in the classic (strict) fundamental domain for the group, together with the corresponding mapping group element.

    The corresponding action of the group on itself by conjugation is supported as well.

    The usual \(slash\)-operator for even integer weights is also available. It acts on rational functions (resp. polynomials). For modular forms an evaluation argument is required.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(n=7)
sage: G.element_repr_method("basic")

sage: G.S().acton(i + exp(-2))
-1/(e^(-2) + I)
sage: A = G.V(2)*G.V(3)^(-2)
sage: A
-S*T^(-2)*S*T^(-1)*S*T^(-1)
sage: A.acton(CC(i + exp(-2)))
0.344549645079... + 0.0163901095115...*I

sage: G.S().acton(A)
-T^(-2)*S*T^(-1)*S*T^(-1)*S

sage: z = AlgebraicField()(4 + 1/7*i)
sage: G.in_FD(z)
False
sage: (A, w) = G.get_FD(z)
sage: A
T^2*S*T^(-1)*S
sage: w
0.516937798396...? + 0.964078044600...?*I

sage: A.acton(w) == z
True
sage: G.in_FD(w)
True

sage: z = PolynomialRing(G.base_ring(), 'z').gen()
sage: rat = z^2 + 1/(z-G.lam())
sage: G.S().slash(rat)
(z^6 - lam*z^4 - z^3)/(-lam*z^4 - z^3)

sage: G.element_repr_method("default")

  • Basic properties of group elements: The trace, sign (based on the trace), discriminant and elliptic/parabolic/hyperbolic type are available.

    Group elements can be displayed/represented in several ways:

    • As matrices over the base ring.
    • As a word in (powers of) the generators S and T.
    • As a word in (powers of) basic block matrices V(j) (resp. U, S in the elliptic case) together with the conjugation matrix that maps the element to this form (also see below).

    For the case n=infinity the last method is not properly implemented.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(n=7)
sage: A = -G.V(2)*G.V(3)^(-2)

sage: print(A.string_repr("default"))
[               lam         -lam^2 + 1]
[       2*lam^2 - 1 -2*lam^2 - lam + 2]
sage: print(A.string_repr("basic"))
S*T^(-2)*S*T^(-1)*S*T^(-1)
sage: print(A.string_repr("block"))
-(-S*T^(-1)*S) * (V(3)) * (-S*T^(-1)*S)^(-1)
sage: print(A.string_repr("conj"))
[-V(3)]
sage: A.trace()
-2*lam^2 + 2
sage: A.sign()
[-1  0]
[ 0 -1]
sage: A.discriminant()
4*lam^2 + 4*lam - 4
sage: A.is_elliptic()
False
sage: A.is_hyperbolic()
True

  • Fixed points: Elliptic, parabolic or hyperbolic fixed points of group can be obtained. They are implemented as a (relative) quadratic extension (given by the square root of the discriminant) of the base ring. It is possible to query the correct embedding into a given field.

    Note that for hyperbolic (and parabolic) fixed points there is a 1-1 correspondence with primitive hyperbolic/parabolic group elements (at least if n < infinity). The group action on fixed points resp. on matrices is compatible with this correspondence.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(n=7)

sage: A = G.S()
sage: A.fixed_points()
(1/2*e, -1/2*e)
sage: A.fixed_points(embedded=True)
(I, -I)

sage: A = G.U()
sage: A.fixed_points()
(1/2*e + 1/2*lam, -1/2*e + 1/2*lam)
sage: A.fixed_points(embedded=True)
(0.9009688679024...? + 0.4338837391175...?*I, 0.9009688679024...? - 0.4338837391175...?*I)

sage: A = -G.V(2)*G.V(3)^(-2)
sage: A.fixed_points()
((-3/7*lam^2 + 2/7*lam + 11/14)*e - 1/7*lam^2 + 3/7*lam + 3/7, (3/7*lam^2 - 2/7*lam - 11/14)*e - 1/7*lam^2 + 3/7*lam + 3/7)
sage: A.fixed_points(embedded=True)
(0.3707208390178...?, 1.103231619181...?)

sage: el = A.fixed_points()[0]
sage: F = A.root_extension_field()
sage: F == el.parent()
True
sage: A.root_extension_embedding(CC)
Relative number field morphism:
  From: Number Field in e with defining polynomial x^2 - 4*lam^2 - 4*lam + 4 over its base field
  To:   Complex Field with 53 bits of precision
  Defn: e |--> 4.02438434522465
        lam |--> 1.80193773580484

sage: G.V(2).acton(A).fixed_points()[0] == G.V(2).acton(el)
True

  • Lambda-continued fractions: For parabolic or hyperbolic elements (resp. their corresponding fixed point) the (negative) lambda-continued fraction expansion is eventually periodic. The lambda-CF (i.e. the preperiod and period) is calculated exactly.

    In particular this allows to determine primitive and reduced generators of group elements and the corresponding primitive power of the element.

    The case n=infinity is not properly implemented.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(n=7)
sage: G.element_repr_method("block")

sage: G.V(6).continued_fraction()
((1,), (1, 1, 1, 1, 2))
sage: (-G.V(2)).continued_fraction()
((1,), (2,))

sage: A = -(G.V(2)*G.V(3)^(-2))^2
sage: A.is_primitive()
False
sage: A.primitive_power()
2
sage: A.is_reduced()
False
sage: A.continued_fraction()
((1, 1, 1, 1), (1, 2))

sage: B = A.primitive_part()
sage: B
(-S*T^(-1)*S) * (V(3)) * (-S*T^(-1)*S)^(-1)
sage: B.is_primitive()
True
sage: B.is_reduced()
False
sage: B.continued_fraction()
((1, 1, 1, 1), (1, 2))
sage: A == A.sign() * B^A.primitive_power()
True

sage: B = A.reduce()
sage: B
(T*S*T) * (V(3)) * (T*S*T)^(-1)
sage: B.is_primitive()
True
sage: B.is_reduced()
True
sage: B.continued_fraction()
((), (1, 2))

sage: G.element_repr_method("default")

  • Reduced and simple elements, Hecke-symmetric elements: For primitive conjugacy classes of hyperbolic elements the cycle of reduced elements can be obtain as well as all simple elements. It is also possible to determine whether a class is Hecke-symmetric.

    The case n=infinity is not properly implemented.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(n=5)

sage: el = G.V(1)^2*G.V(2)*G.V(4)
sage: R = el.reduced_elements()
sage: [v.continued_fraction() for v in R]
[((), (2, 1, 1, 4)), ((), (1, 1, 4, 2)), ((), (1, 4, 2, 1)), ((), (4, 2, 1, 1))]

sage: el = G.V(1)^2*G.V(2)*G.V(4)
sage: R = el.simple_elements()
sage: [v.is_simple() for v in R]
[True, True, True, True]
sage: (fp1, fp2) = R[2].fixed_points(embedded=True)
sage: fp2 < 0 < fp1
True

sage: el = G.V(2)
sage: el.is_hecke_symmetric()
False
sage: (el.simple_fixed_point_set(), el.inverse().simple_fixed_point_set())
({1/2*e, (-1/2*lam + 1/2)*e}, {-1/2*e, (1/2*lam - 1/2)*e})
sage: el = G.V(2)*G.V(3)
sage: el.is_hecke_symmetric()
True
sage: el.simple_fixed_point_set() == el.inverse().simple_fixed_point_set()
True

  • Rational period functions: For each primitive (hyperbolic) conjugacy classes and each even weight k we can associate a corresponding rational period function. I.e. a rational function q of weight k which satisfies: q | S == 0 and q + q|U + ... + q|U^(n-1) == 0, where S, U are the corresponding group elements and | is the usual \(slash-operator\) of weight k.

    The set of all rational period function is expected to be generated by such functions.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(n=5)
sage: S = G.S()
sage: U = G.U()

sage: def is_rpf(f, k=None):
....:     if not f + S.slash(f, k=k) == 0:
....:         return False
....:     if not sum([(U^m).slash(f, k=k) for m in range(G.n())]) == 0:
....:         return False
....:     return True

sage: z = PolynomialRing(G.base_ring(), 'z').gen()
sage: [is_rpf(1 - z^(-k), k=k) for k in range(-6, 6, 2)]  # long time
[True, True, True, True, True, True]
sage: [is_rpf(1/z, k=k) for k in range(-6, 6, 2)]
[False, False, False, False, True, False]

sage: el = G.V(2)
sage: el.is_hecke_symmetric()
False
sage: rpf = el.rational_period_function(-4)
sage: is_rpf(rpf)
True
sage: rpf
-lam*z^4 + lam
sage: rpf = el.rational_period_function(-2)
sage: is_rpf(rpf)
True
sage: rpf
(lam + 1)*z^2 - lam - 1
sage: el.rational_period_function(0) == 0
True
sage: rpf = el.rational_period_function(2)
sage: is_rpf(rpf)
True
sage: rpf
((lam + 1)*z^2 - lam - 1)/(lam*z^4 + (-lam - 2)*z^2 + lam)

sage: el = G.V(2)*G.V(3)
sage: el.is_hecke_symmetric()
True
sage: el.rational_period_function(-4) == 0
True
sage: rpf = el.rational_period_function(-2)
sage: rpf
(8*lam + 4)*z^2 - 8*lam - 4
sage: rpf = el.rational_period_function(2)
sage: is_rpf(rpf)
True
sage: rpf.denominator()
(144*lam + 89)*z^8 + (-618*lam - 382)*z^6 + (951*lam + 588)*z^4 + (-618*lam - 382)*z^2 + 144*lam + 89
sage: el.rational_period_function(4) == 0
True

sage: G = HeckeTriangleGroup(n=4)
sage: G.rational_period_functions(k=4, D=12)
[(z^4 - 1)/z^4]
sage: G.rational_period_functions(k=2, D=14)
[(z^2 - 1)/z^2, 1/z, (24*z^6 - 120*z^4 + 120*z^2 - 24)/(9*z^8 - 80*z^6 + 146*z^4 - 80*z^2 + 9), (24*z^6 - 120*z^4 + 120*z^2 - 24)/(9*z^8 - 80*z^6 + 146*z^4 - 80*z^2 + 9)]

  • Block decomposition of elements: For each group element a very specific conjugacy representative can be obtained. For hyperbolic and parabolic elements the representative is a product V(j)-matrices. They all have non-negative trace and the number of factors is called the block length of the element (which is implemented).

    Note: For this decomposition special care is given to the sign (of the trace) of the matrices.

    The case n=infinity for everything above is not properly implemented.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(n=7)
sage: G.element_repr_method("block")

sage: A = -G.V(2)*G.V(6)^3*G.V(3)
sage: A
-(T*S*T) * (V(6)^3*V(3)*V(2)) * (T*S*T)^(-1)
sage: A.sign()
-1
sage: (L, R, sgn) = A.block_decomposition()
sage: L
((-S*T^(-1)*S) * (V(6)^3) * (-S*T^(-1)*S)^(-1), (T*S*T*S*T) * (V(3)) * (T*S*T*S*T)^(-1), (T*S*T) * (V(2)) * (T*S*T)^(-1))
sage: prod(L).sign()
1
sage: A == sgn * (R.acton(prod(L)))
True
sage: t = A.block_length()
sage: t
5
sage: AA(A.discriminant()) >= AA(t^2 * G.lam() - 4)
True

  • Class number and class representatives: The block length provides a lower bound for the discriminant. This allows to enlist all (representatives of) matrices of (or up to) a given discriminant.

    Using the 1-1 correspondence with hyperbolic fixed points (and certain hyperbolic binary quadratic forms) this makes it possible to calculate the corresponding class number (number of conjugacy classes for a given discriminant).

    It also allows to list all occurring discriminants up to some bound. Or to enlist all reduced/simple elements resp. their corresponding hyperbolic fixed points for the given discriminant.

    Warning: The currently used algorithm is very slow!

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.hecke_triangle_groups import HeckeTriangleGroup
sage: G = HeckeTriangleGroup(n=4)
sage: G.element_repr_method("basic")
sage: G.is_discriminant(68)
True
sage: G.class_number(14)
2
sage: G.list_discriminants(D=68)
[4, 12, 14, 28, 32, 46, 60, 68]
sage: G.list_discriminants(D=0, hyperbolic=False, primitive=False)
[-4, -2, 0]
sage: G.class_number(68)
4
sage: G.class_representatives(68)
[S*T^(-2)*S*T^(-1)*S*T, -S*T^(-1)*S*T^2*S*T, S*T^(-5)*S*T^(-1)*S, T*S*T^5]
sage: R = G.reduced_elements(68)
sage: all(v.is_reduced() for v in R)  # long time
True
sage: R = G.simple_elements(68)
sage: all(v.is_simple() for v in R)  # long time
True
sage: G.element_repr_method("default")

sage: G = HeckeTriangleGroup(n=5)
sage: G.element_repr_method("basic")
sage: G.list_discriminants(9*G.lam() + 5)
[4*lam, 7*lam + 6, 9*lam + 5]
sage: G.list_discriminants(D=0, hyperbolic=False, primitive=False)
[-4, -lam - 2, lam - 3, 0]
sage: G.class_number(9*G.lam() + 5)
2
sage: G.class_representatives(9*G.lam() + 5)
[S*T^(-2)*S*T^(-1)*S, T*S*T^2]
sage: R = G.reduced_elements(9*G.lam() + 5)
sage: all(v.is_reduced() for v in R)  # long time
True
sage: R = G.simple_elements(7*G.lam() + 6)
sage: for v in R: print(v.string_repr("default"))
[lam + 2     lam]
[    lam       1]
[      1     lam]
[    lam lam + 2]
sage: G.element_repr_method("default")

Modular forms ring and spaces for Hecke triangle groups:

  • Analytic type: The analytic type of forms, including the behavior at infinity:

    • Meromorphic (and meromorphic at infinity)
    • Weakly holomorphic (holomorphic and meromorphic at infinity)
    • Holomorphic (and holomorphic at infinity)
    • Cuspidal (holomorphic and zero at infinity)

    Additionally the type specifies whether the form is modular or only quasi modular.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.analytic_type import AnalyticType
sage: AnalyticType()(["quasi", "cusp"])
quasi cuspidal

  • Modular form (for Hecke triangle groups): A function of some analytic type which transforms like a modular form for the given group, weight k and multiplier epsilon:

    • f(z+lambda) = f(lambda)
    • f(-1/z) = epsilon * (z/i)^k * f(z)

    The multiplier is either 1 or -1. The weight is a rational number of the form 4*(n*l+l')/(n-2) + (1-epsilon)*n/(n-2). If n is odd, then the multiplier is unique and given by (-1)^(k*(n-2)/2). The space of modular forms for a given group, weight and multiplier forms a module over the base ring. It is finite dimensional if the analytic type is holomorphic.

    Modular forms can be constructed in several ways:

    • Using some already available construction function for modular forms (those function are available for all spaces/rings and in general do not return elements of the same parent)
    • Specifying the form as a rational function in the basic generators (see below)
    • For weakly holomorphic modular forms it is possible to exactly determine the form by specifying (sufficiently many) initial coefficients of its Fourier expansion.
    • There is even hope (no garantuee) to determine a (exact) form from the initial numerical coefficients (see below).
    • By specifying the coefficients with respect to a basis of the space (if the corresponding space supports coordinate vectors)
    • Arithmetic combination of forms or differential operators applied to forms

    The implementation is based on the implementation of the graded ring (see below). All calculations are exact (no precision argument is required). The analytic type of forms is checked during construction. The analytic type of parent spaces after arithmetic/differential operations with elements is changed (extended/reduced) accordingly.

    In particular it is possible to multiply arbitrary modular forms (and end up with an element of a modular forms space). If two forms of different weight/multiplier are added then an element of the corresponding modular forms ring is returned instead.

    Elements of modular forms spaces are represented by their Fourier expansion.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.space import CuspForms, ModularForms, MeromorphicModularForms
sage: MeromorphicModularForms(n=4, k=8, ep=1)
MeromorphicModularForms(n=4, k=8, ep=1) over Integer Ring
sage: CF = CuspForms(n=7, k=12, ep=1)
sage: CF
CuspForms(n=7, k=12, ep=1) over Integer Ring

sage: MF = ModularForms(k=12, ep=1)
sage: (x,y,z,d) = MF.pol_ring().gens()

Using existing functions:
sage: CF.Delta()
q + 17/(56*d)*q^2 + 88887/(2458624*d^2)*q^3 + 941331/(481890304*d^3)*q^4 + O(q^5)

Using rational function in the basic generators:
sage: MF(x^3)
1 + 720*q + 179280*q^2 + 16954560*q^3 + 396974160*q^4 + O(q^5)

Using Fourier expansions:
sage: qexp = CF.Delta().q_expansion(prec=2)
sage: qexp
q + O(q^2)
sage: qexp.parent()
Power Series Ring in q over Fraction Field of Univariate Polynomial Ring in d over Integer Ring
sage: MF(qexp)
q - 24*q^2 + 252*q^3 - 1472*q^4 + O(q^5)

Using coordinate vectors:
sage: MF([0,1]) == MF.f_inf()
True

Using arithmetic expressions:
sage: d = CF.get_d()
sage: CF.f_rho()^7 / (d*CF.f_rho()^7 - d*CF.f_i()^2) == CF.j_inv()
True
sage: MF.E4().serre_derivative() == -1/3 * MF.E6()
True

  • Hauptmodul: The j-function for Hecke triangle groups is given by the unique Riemann map from the hyperbolic triangle with vertices at rho, i and infinity to the upper half plane, normalized such that its Fourier coefficients are real and such that the first nontrivial Fourier coefficient is 1. The function extends to a completely invariant weakly holomorphic function from the upper half plane to the complex numbers. Another used normalization (in capital letters) is J(i)=1. The coefficients of j are rational numbers up to a power of d=1/j(i) which is only rational in the arithmetic cases n=3, 4, 6, infinity.

    All Fourier coefficients of modular forms are based on the coefficients of j. The coefficients of j are calculated by inverting the Fourier series of its inverse (the series inversion is also by far the most expensive operation of all).

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.graded_ring import WeakModularFormsRing
sage: from sage.modular.modform_hecketriangle.space import WeakModularForms
sage: WeakModularForms(n=3, k=0, ep=1).j_inv()
q^-1 + 744 + 196884*q + 21493760*q^2 + 864299970*q^3 + 20245856256*q^4 + O(q^5)
sage: WeakModularFormsRing(n=7).j_inv()
f_rho^7/(f_rho^7*d - f_i^2*d)
sage: WeakModularFormsRing(n=7, red_hom=True).j_inv()
q^-1 + 151/(392*d) + 165229/(2458624*d^2)*q + 107365/(15059072*d^3)*q^2 + 25493858865/(48358655787008*d^4)*q^3 + 2771867459/(92561489592320*d^5)*q^4 + O(q^5)

  • Basic generators: There exist unique modular forms f_rho, f_i and f_inf such that each has a simple zero at rho=exp(pi/n), i and infinity resp. and no other zeros. The forms are normalized such that their first Fourier coefficient is 1. They have the weight and multiplier (4/(n-2), 1), (2*n/(n-2), -1), (4*n/(n-2), 1) resp. and can be defined in terms of the Hauptmodul j.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.graded_ring import ModularFormsRing
sage: ModularFormsRing(n=5, red_hom=True).f_rho()
1 + 7/(100*d)*q + 21/(160000*d^2)*q^2 + 1043/(192000000*d^3)*q^3 + 45479/(1228800000000*d^4)*q^4 + O(q^5)
sage: ModularFormsRing(n=5, red_hom=True).f_i()
1 - 13/(40*d)*q - 351/(64000*d^2)*q^2 - 13819/(76800000*d^3)*q^3 - 1163669/(491520000000*d^4)*q^4 + O(q^5)
sage: ModularFormsRing(n=5, red_hom=True).f_inf()
q - 9/(200*d)*q^2 + 279/(640000*d^2)*q^3 + 961/(192000000*d^3)*q^4 + O(q^5)
sage: ModularFormsRing(n=5).f_inf()
f_rho^5*d - f_i^2*d

  • Eisenstein series and Delta: The Eisenstein series of weight 2, 4 and 6 exist for all n and are all implemented . Note that except for n=3 the series E4 and E6 do not coincide with f_rho and f_i.

    Similarly there always exists a (generalization of) Delta. Except for n=3 it also does not coincide with f_inf.

    In general Eisenstein series of all even weights exist for all n. In the non-arithmetic cases they are however very hard to determine (it’s an open problem(?) and consequently not yet implemented, except for trivial one-dimensional cases).

    The Eisenstein series in the arithmetic cases n = 3, 4, 6 are fully implemented though. Note that this requires a lot more work/effort for k != 2, 4, 6 resp. for multidimensional spaces.

    The case n=infinity is a special case (since there are two cusps) and is not implemented yet.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.graded_ring import ModularFormsRing
sage: from sage.modular.modform_hecketriangle.space import ModularForms
sage: ModularFormsRing(n=5).E4()
f_rho^3
sage: ModularFormsRing(n=5).E6()
f_rho^2*f_i
sage: ModularFormsRing(n=5).Delta()
f_rho^9*d - f_rho^4*f_i^2*d
sage: ModularFormsRing(n=5).Delta() == ModularFormsRing(n=5).f_inf()*ModularFormsRing(n=5).f_rho()^4
True

The basic generators in some arithmetic cases:
sage: ModularForms(n=3, k=6).E6()
1 - 504*q - 16632*q^2 - 122976*q^3 - 532728*q^4 + O(q^5)
sage: ModularForms(n=4, k=6).E6()
1 - 56*q - 2296*q^2 - 13664*q^3 - 73976*q^4 + O(q^5)
sage: ModularForms(n=infinity, k=4).E4()
1 + 16*q + 112*q^2 + 448*q^3 + 1136*q^4 + O(q^5)

General Eisenstein series in some arithmetic cases:
sage: ModularFormsRing(n=4).EisensteinSeries(k=8)
(-25*f_rho^4 - 9*f_i^2)/(-34)
sage: ModularForms(n=3, k=12).EisensteinSeries()
1 + 65520/691*q + 134250480/691*q^2 + 11606736960/691*q^3 + 274945048560/691*q^4 + O(q^5)
sage: ModularForms(n=6, k=12).EisensteinSeries()
1 + 6552/50443*q + 13425048/50443*q^2 + 1165450104/50443*q^3 + 27494504856/50443*q^4 + O(q^5)
sage: ModularForms(n=4, k=22, ep=-1).EisensteinSeries()
1 - 184/53057489*q - 386252984/53057489*q^2 - 1924704989536/53057489*q^3 - 810031218278584/53057489*q^4 + O(q^5)

  • Generator for ``k=0``, ``ep=-1``: If n is even then the space of weakly holomorphic modular forms of weight 0 and multiplier -1 is not empty and generated by one element, denoted by g_inv.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.space import WeakModularForms
sage: WeakModularForms(n=4, k=0, ep=-1).g_inv()
q^-1 - 24 - 3820*q - 100352*q^2 - 1217598*q^3 - 10797056*q^4 + O(q^5)
sage: WeakModularFormsRing(n=8).g_inv()
f_rho^4*f_i/(f_rho^8*d - f_i^2*d)

  • Quasi modular form (for Hecke triangle groups): E2 no longer transforms like a modular form but like a quasi modular form. More generally quasi modular forms are given in terms of modular forms and powers of E2. E.g. a holomorphic quasi modular form is a sum of holomorphic modular forms multiplied with a power of E2 such that the weights and multipliers match up. The space of quasi modular forms for a given group, weight and multiplier forms a module over the base ring. It is finite dimensional if the analytic type is holomorphic.

    The implementation and construction are analogous to modular forms (see above). In particular construction of quasi weakly holomorphic forms by their initial Laurent coefficients is supported as well!

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.graded_ring import ModularFormsRing
sage: from sage.modular.modform_hecketriangle.space import QuasiCuspForms, QuasiModularForms, QuasiWeakModularForms
sage: QuasiCuspForms(n=7, k=12, ep=1)
QuasiCuspForms(n=7, k=12, ep=1) over Integer Ring
sage: QuasiModularForms(n=4, k=8, ep=-1)
QuasiModularForms(n=4, k=8, ep=-1) over Integer Ring

sage: QuasiModularForms(n=4, k=2, ep=-1).E2()
1 - 8*q - 40*q^2 - 32*q^3 - 104*q^4 + O(q^5)

A quasi weak form can be constructed by using its initial Laurent expansion:
sage: QF = QuasiWeakModularForms(n=8, k=10/3, ep=-1)
sage: qexp = (QF.quasi_part_gens(min_exp=-1)[4]).q_expansion(prec=5)
sage: qexp
q^-1 - 19/(64*d) - 7497/(262144*d^2)*q + 15889/(8388608*d^3)*q^2 + 543834047/(1649267441664*d^4)*q^3 + 711869853/(43980465111040*d^5)*q^4 + O(q^5)
sage: qexp.parent()
Laurent Series Ring in q over Fraction Field of Univariate Polynomial Ring in d over Integer Ring
sage: QF(qexp).as_ring_element()
f_rho^3*f_i*E2^2/(f_rho^8*d - f_i^2*d)
sage: QF(qexp).reduced_parent()
QuasiWeakModularForms(n=8, k=10/3, ep=-1) over Integer Ring

Derivatives of (quasi weak) modular forms are again quasi (weak) modular forms:
sage: CF.f_inf().derivative() == CF.f_inf()*CF.E2()
True

  • Ring of (quasi) modular forms: The ring of (quasi) modular forms for a given analytic type and Hecke triangle group. In fact it is a graded algebra over the base ring where the grading is over 1/(n-2)*Z x Z/(2Z) corresponding to the weight and multiplier. A ring element is thus a finite linear combination of (quasi) modular forms of (possibly) varying weights and multipliers.

    Each ring element is represented as a rational function in the generators f_rho, f_i and E2. The representations and arithmetic operations are exact (no precision argument is required).

    Elements of the ring are represented by the rational function in the generators.

    If the parameter red_hom is set to True (default: False) then operations with homogeneous elements try to return an element of the corresponding vector space (if the element is homogeneous) instead of the forms ring. It is also easier to use the forms ring with red_hom=True to construct known forms (since then it is not required to specify the weight and multiplier).

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.graded_ring import QuasiModularFormsRing, ModularFormsRing
sage: QuasiModularFormsRing(n=5, red_hom=True)
QuasiModularFormsRing(n=5) over Integer Ring
sage: ModularFormsRing()
ModularFormsRing(n=3) over Integer Ring
sage: (x,y,z,d) = ModularFormsRing().pol_ring().gens()

sage: ModularFormsRing()(x+y)
f_rho + f_i

sage: QuasiModularFormsRing(n=5, red_hom=True)(x^5-y^2).reduce()
1/d*q - 9/(200*d^2)*q^2 + 279/(640000*d^3)*q^3 + 961/(192000000*d^4)*q^4 + O(q^5)

  • Construction of modular forms spaces and rings: There are functorial constructions behind all forms spaces and rings which assure that arithmetic operations between those spaces and rings work and fit into the coercion framework. In particular ring elements are interpreted as constant modular forms in this context and base extensions are done if necessary.

  • Fourier expansion of (quasi) modular forms (for Hecke triangle groups): Each (quasi) modular form (in fact each ring element) possesses a Fourier expansion of the form sum_{n>=n_0} a_n q^n, where n_0 is an integer, q=exp(2*pi*i*z/lambda) and the coefficients a_n are rational numbers (or more generally an extension of rational numbers) up to a power of d, where d is the (possibly) transcendental parameter described above. I.e. the coefficient ring is given by Frac(R)(d).

    The coefficients are calculated exactly in terms of the (formal) parameter d. The expansion is calculated exactly up to the specified precision. It is also possible to get a Fourier expansion where d is evaluated to its numerical approximation.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.graded_ring import ModularFormsRing, QuasiModularFormsRing
sage: ModularFormsRing(n=4).j_inv().q_expansion(prec=3)
q^-1 + 13/(32*d) + 1093/(16384*d^2)*q + 47/(8192*d^3)*q^2 + O(q^3)
sage: QuasiModularFormsRing(n=5).E2().q_expansion(prec=3)
1 - 9/(200*d)*q - 369/(320000*d^2)*q^2 + O(q^3)
sage: QuasiModularFormsRing(n=5).E2().q_expansion_fixed_d(prec=3)
1.000000000000... - 6.380956565426...*q - 23.18584547617...*q^2 + O(q^3)

  • Evaluation of forms: (Quasi) modular forms (and also ring elements) can be viewed as functions from the upper half plane and can be numerically evaluated by using the Fourier expansion.

    The evaluation uses the (quasi) modularity properties (if possible) for a faster and more precise evaluation. The precision of the result depends both on the numerical precision and on the default precision used for the Fourier expansion.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.graded_ring import ModularFormsRing
sage: f_i = ModularFormsRing(n=4).f_i()
sage: f_i(i)
0
sage: f_i(infinity)
1
sage: f_i(1/7 + 0.01*i)
32189.02016723... + 21226.62951394...*I

  • L-functions of forms: Using the (pari based) function Dokchitser L-functions of non-constant holomorphic modular forms are supported for all values of n.

    Note: For non-arithmetic groups this involves an irrational conductor. The conductor for the arithmetic groups n = 3, 4, 6, infinity is 1, 2, 3, 4 respectively.

    EXAMPLES:

    sage: from sage.modular.modform.eis_series import eisenstein_series_lseries
sage: from sage.modular.modform_hecketriangle.space import ModularForms
sage: f = ModularForms(n=3, k=4).E4()/240
sage: L = f.lseries()
sage: L.conductor
1
sage: L.check_functional_equation() < 2^(-50)
True
sage: L(1)
-0.0304484570583...
sage: abs(L(1) - eisenstein_series_lseries(4)(1)) < 2^(-53)
True
sage: L.taylor_series(1, 3)
-0.0304484570583... - 0.0504570844798...*z - 0.0350657360354...*z^2 + O(z^3)
sage: coeffs = f.q_expansion_vector(min_exp=0, max_exp=20, fix_d=True)
sage: abs(L(10) - sum([coeffs[k] * ZZ(k)^(-10) for k in range(1,len(coeffs))]).n(53)) < 10^(-7)
True

sage: L = ModularForms(n=6, k=6, ep=-1).E6().lseries(num_prec=200)
sage: L.conductor
3
sage: L.check_functional_equation() < 2^(-180)
True
sage: L.eps
-1
sage: abs(L(3)) < 2^(-180)
True

sage: L = ModularForms(n=17, k=12).Delta().lseries()
sage: L.conductor
3.86494445880...
sage: L.check_functional_equation() < 2^(-50)
True
sage: L.taylor_series(6, 3)
2.15697985314... - 1.17385918996...*z + 0.605865993050...*z^2 + O(z^3)

sage: L = ModularForms(n=infinity, k=2, ep=-1).f_i().lseries()
sage: L.conductor
4
sage: L.check_functional_equation() < 2^(-50)
True
sage: L.taylor_series(1, 3)
0.000000000000... + 5.76543616701...*z + 9.92776715593...*z^2 + O(z^3)

  • (Serre) derivatives: Derivatives and Serre derivatives of forms can be calculated. The analytic type is extended accordingly.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.graded_ring import ModularFormsRing
sage: from sage.modular.modform_hecketriangle.space import QuasiModularForms
sage: f_inf = ModularFormsRing(n=4, red_hom=True).f_inf()
sage: f_inf.derivative()/f_inf == QuasiModularForms(n=4, k=2, ep=-1).E2()
True
sage: ModularFormsRing().E4().serre_derivative() == -1/3 * ModularFormsRing().E6()
True

  • Basis for weakly holomorphic modular forms and Faber polynomials: (Natural) generators of weakly holomorphic modular forms can be obtained using the corresponding generalized Faber polynomials.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.space import WeakModularForms, CuspForms
sage: MF = WeakModularForms(n=5, k=62/3, ep=-1)
sage: MF.disp_prec(MF._l1+2)

sage: MF.F_basis(2)
q^2 - 41/(200*d)*q^3 + O(q^4)
sage: MF.F_basis(1)
q - 13071/(640000*d^2)*q^3 + O(q^4)
sage: MF.F_basis(-0)
1 - 277043/(192000000*d^3)*q^3 + O(q^4)
sage: MF.F_basis(-2)
q^-2 - 162727620113/(40960000000000000*d^5)*q^3 + O(q^4)

  • Basis for quasi weakly holomorphic modular forms: (Natural) generators of quasi weakly holomorphic modular forms can also be obtained. In most cases it is even possible to find a basis consisting of elements with only one non-trivial Laurent coefficient (up to some coefficient).

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.space import QuasiWeakModularForms
sage: QF = QuasiWeakModularForms(n=8, k=10/3, ep=-1)
sage: QF.default_prec(1)
sage: QF.quasi_part_gens(min_exp=-1)
[q^-1 + O(q),
 1 + O(q),
 q^-1 - 9/(128*d) + O(q),
 1 + O(q),
 q^-1 - 19/(64*d) + O(q),
 q^-1 + 1/(64*d) + O(q)]
sage: QF.default_prec(QF.required_laurent_prec(min_exp=-1))
sage: QF.q_basis(min_exp=-1)    # long time
[q^-1 + O(q^5),
 1 + O(q^5),
 q + O(q^5),
 q^2 + O(q^5),
 q^3 + O(q^5),
 q^4 + O(q^5)]

  • Dimension and basis for holomorphic or cuspidal (quasi) modular forms: For finite dimensional spaces the dimension and a basis can be obtained.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.space import QuasiModularForms
sage: MF = QuasiModularForms(n=5, k=6, ep=-1)
sage: MF.dimension()
3
sage: MF.default_prec(2)
sage: MF.gens()
[1 - 37/(200*d)*q + O(q^2),
 1 + 33/(200*d)*q + O(q^2),
 1 - 27/(200*d)*q + O(q^2)]

  • Coordinate vectors for (quasi) holomorphic modular forms and (quasi) cusp forms: For (quasi) holomorphic modular forms and (quasi) cusp forms it is possible to determine the coordinate vectors of elements with respect to the basis.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.space import ModularForms
sage: ModularForms(n=7, k=12, ep=1).dimension()
3
sage: ModularForms(n=7, k=12, ep=1).Delta().coordinate_vector()
(0, 1, 17/(56*d))

sage: from sage.modular.modform_hecketriangle.space import QuasiCuspForms
sage: MF = QuasiCuspForms(n=7, k=20, ep=1)
sage: MF.dimension()
13
sage: el = MF(MF.Delta()*MF.E2()^4 + MF.Delta()*MF.E2()*MF.E6())
sage: el.coordinate_vector()    # long time
(0, 0, 0, 1, 29/(196*d), 0, 0, 0, 0, 1, 17/(56*d), 0, 0)

  • Subspaces: It is possible to construct subspaces of (quasi) holomorphic modular forms or (quasi) cusp forms spaces with respect to a specified basis of the corresponding ambient space. The subspaces also support coordinate vectors with respect to its basis.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.space import ModularForms
sage: MF = ModularForms(n=7, k=12, ep=1)
sage: subspace = MF.subspace([MF.E4()^3, MF.Delta()])
sage: subspace
Subspace of dimension 2 of ModularForms(n=7, k=12, ep=1) over Integer Ring
sage: el = subspace(MF.E6()^2)
sage: el.coordinate_vector()
(1, -61/(196*d))
sage: el.ambient_coordinate_vector()
(1, -61/(196*d), -51187/(614656*d^2))

sage: from sage.modular.modform_hecketriangle.space import QuasiCuspForms
sage: MF = QuasiCuspForms(n=7, k=20, ep=1)
sage: subspace = MF.subspace([MF.Delta()*MF.E2()^2*MF.E4(), MF.Delta()*MF.E2()^4])    # long time
sage: subspace    # long time
Subspace of dimension 2 of QuasiCuspForms(n=7, k=20, ep=1) over Integer Ring
sage: el = subspace(MF.Delta()*MF.E2()^4)    # long time
sage: el.coordinate_vector()    # long time
(0, 1)
sage: el.ambient_coordinate_vector()    # long time
(0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 17/(56*d), 0, 0)

  • Theta subgroup: The Hecke triangle group corresponding to n=infinity is also completely supported. In particular the (special) behavior around the cusp -1 is considered and can be specified.

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.graded_ring import QuasiMeromorphicModularFormsRing
sage: MR = QuasiMeromorphicModularFormsRing(n=infinity, red_hom=True)
sage: MR
QuasiMeromorphicModularFormsRing(n=+Infinity) over Integer Ring
sage: j_inv = MR.j_inv().full_reduce()
sage: f_i = MR.f_i().full_reduce()
sage: E4 = MR.E4().full_reduce()
sage: E2 = MR.E2().full_reduce()

sage: j_inv
q^-1 + 24 + 276*q + 2048*q^2 + 11202*q^3 + 49152*q^4 + O(q^5)
sage: MR.f_rho() == MR(1)
True
sage: E4
1 + 16*q + 112*q^2 + 448*q^3 + 1136*q^4 + O(q^5)
sage: f_i
1 - 24*q + 24*q^2 - 96*q^3 + 24*q^4 + O(q^5)
sage: E2
1 - 8*q - 8*q^2 - 32*q^3 - 40*q^4 + O(q^5)
sage: E4.derivative() == E4 * (E2 - f_i)
True
sage: f_i.serre_derivative() == -1/2 * E4
True
sage: MF = f_i.serre_derivative().parent()
sage: MF
ModularForms(n=+Infinity, k=4, ep=1) over Integer Ring
sage: MF.dimension()
2
sage: MF.gens()
[1 + 240*q^2 + 2160*q^4 + O(q^5), q - 8*q^2 + 28*q^3 - 64*q^4 + O(q^5)]
sage: E4(i)
1.941017189...
sage: E4.order_at(-1)
1

sage: MF = (E2/E4).reduced_parent()
sage: MF.quasi_part_gens(order_1=-1)
[1 - 40*q + 552*q^2 - 4896*q^3 + 33320*q^4 + O(q^5),
 1 - 24*q + 264*q^2 - 2016*q^3 + 12264*q^4 + O(q^5)]
sage: prec = MF.required_laurent_prec(order_1=-1)
sage: qexp = (E2/E4).q_expansion(prec=prec)
sage: qexp
1 - 3/(8*d)*q + O(q^2)
sage: MF.construct_quasi_form(qexp, order_1=-1) == E2/E4
True
sage: MF.disp_prec(6)
sage: MF.q_basis(m=-1, order_1=-1, min_exp=-1)
q^-1 - 203528/7*q^5 + O(q^6)

    Elements with respect to the full group are automatically coerced to elements of the Theta subgroup if necessary:

    sage: el = QuasiMeromorphicModularFormsRing(n=3).Delta().full_reduce() + E2
sage: el
(E4*f_i^4 - 2*E4^2*f_i^2 + E4^3 + 4096*E2)/4096
sage: el.parent()
QuasiModularFormsRing(n=+Infinity) over Integer Ring

  • Determine exact coefficients from numerical ones: There is some experimental support for replacing numerical coefficients with corresponding exact coefficients. There is however NO guarantee that the procedure will work (and most probably there are cases where it won’t).

    EXAMPLES:

    sage: from sage.modular.modform_hecketriangle.space import WeakModularForms, QuasiCuspForms
sage: WF = WeakModularForms(n=14)
sage: qexp = WF.J_inv().q_expansion_fixed_d(d_num_prec=1000)
sage: qexp.parent()
Laurent Series Ring in q over Real Field with 1000 bits of precision
sage: qexp_int = WF.rationalize_series(qexp)
doctest:...: UserWarning: Using an experimental rationalization of coefficients, please check the result for correctness!
sage: qexp_int.parent()
Laurent Series Ring in q over Fraction Field of Univariate Polynomial Ring in d over Integer Ring
sage: qexp_int == WF.J_inv().q_expansion()
True
sage: WF(qexp_int) == WF.J_inv()
True

sage: QF = QuasiCuspForms(n=8, k=22/3, ep=-1)
sage: el = QF(QF.f_inf()*QF.E2())
sage: qexp = el.q_expansion_fixed_d(d_num_prec=1000)
sage: qexp_int = QF.rationalize_series(qexp)
sage: qexp_int == el.q_expansion()
True
sage: QF(qexp_int) == el
True

Future ideas:

  • Complete support for the case n=infinity (e.g. lambda-CF)
  • Properly implemented lambda-CF
  • Binary quadratic forms for Hecke triangle groups
  • Cycle integrals
  • Maybe: Proper spaces (with coordinates) for (quasi) weakly holomorphic forms with bounds on the initial Fourier exponent
  • Support for general triangle groups (hard)
  • Support for “congruence” subgroups (hard)