passagemath-flint 10.6.1rc10__cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl

sage/rings/polynomial/hilbert.pyx

@@ -0,0 +1,602 @@
+# sage_setup: distribution = sagemath-flint
+r"""
+Compute Hilbert series of monomial ideals
+
+This implementation was provided at :issue:`26243` and is supposed to be a way
+out when Singular fails with an int overflow, which will regularly be the case
+in any example with more than 34 variables.
+"""
+#*****************************************************************************
+#
+#    Tools to compute Hilbert Poincaré series of monomial ideals
+#
+#    Copyright (C) 2018 Simon A. King <simon.king@uni-jena.de>
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 2 of the License, or
+# (at your option) any later version.
+#                  http://www.gnu.org/licenses/
+#
+#*****************************************************************************
+
+import sage.interfaces.abc
+
+from sage.rings.polynomial.polydict cimport ETuple
+from sage.rings.polynomial.polynomial_integer_dense_flint cimport Polynomial_integer_dense_flint
+
+from cysignals.memory cimport sig_malloc
+from cpython.list cimport PyList_GET_ITEM
+
+# A boilerplate class to implement a binary tree
+# with some additional information
+
+from sage.libs.flint.fmpz_poly cimport *
+from sage.libs.flint.types cimport fmpz_poly_t
+
+cdef class Node:
+    """
+    A node of a binary tree
+
+    It has slots for data that allow to recursively compute
+    the first Hilbert series of a monomial ideal.
+    """
+    cdef Node Back, Left, Right
+    cdef list Id
+    cdef fmpz_poly_t LMult
+    cdef fmpz_poly_t RMult
+    cdef fmpz_poly_t LeftFHS
+
+    def __cinit__(self):
+        fmpz_poly_init(self.LMult)
+        fmpz_poly_init(self.RMult)
+        fmpz_poly_init(self.LeftFHS)
+
+    def __dealloc__(self):
+        fmpz_poly_clear(self.LMult)
+        fmpz_poly_clear(self.RMult)
+        fmpz_poly_clear(self.LeftFHS)
+
+cdef inline size_t median(list v) noexcept:
+    """
+    Specialized version of :func:`from sage.stats.basic_stats.median`.
+    """
+    cdef list values = sorted(v)
+    cdef size_t nvals = len(values)
+    if nvals % 2 == 1:
+        return values[(nvals+1)//2 - 1]
+    else:
+        lower = values[(nvals+1)//2 - 1]
+        upper = values[nvals // 2]
+        return (lower + upper) // 2
+
+###
+#   cdef functions related with lists of monomials
+###
+
+cdef inline bint indivisible_in_list(ETuple m, list L, size_t i) noexcept:
+    """
+    Return if ``m`` divisible by any monomial in ``L[:i]``.
+    """
+    cdef size_t j
+    for j in range(i):
+        if (<ETuple>PyList_GET_ITEM(L,j)).divides(m):
+            return False
+    return True
+
+cdef inline list interred(list L):
+    """
+    Return interreduction of a list of monomials.
+
+    .. NOTE::
+
+        The given list will be sorted in-place.
+
+    INPUT:
+
+    - ``L`` -- list of :class:`~sage.rings.polynomial.polydict.ETuple`
+
+    OUTPUT:
+
+    The interreduced list, where we interprete each ETuple as
+    a monomial in a multivariate ring.
+    """
+    # First, we sort L ascendingly by total unweighted degree.
+    # Afterwards, no monomial in L is divisible by a monomial
+    # that appears later in L.
+    if not L:
+        return []
+    L.sort(key=ETuple._unweighted_degree)
+    cdef size_t i
+    cdef ETuple m
+    cdef list result = [<ETuple> PyList_GET_ITEM(L, 0)]
+    for i in range(1, len(L)):
+        m = <ETuple> PyList_GET_ITEM(L, i)
+        if indivisible_in_list(m, L, i):
+            result.append(m)
+    return result
+
+cdef list quotient(list L, ETuple m):
+    """
+    Return the quotient of the ideal represented by ``L`` and the
+    monomial represented by ``m``.
+    """
+    cdef list result = L[:len(L)] # creates a copy
+    cdef size_t i
+    for i in range(len(L)):
+        result.append((<ETuple>PyList_GET_ITEM(L,i)).divide_by_gcd(m))
+    return interred(result)
+
+cdef list quotient_by_var(list L, size_t index):
+    """
+    Return the quotient of the ideal represented by ``L`` and the
+    variable number ``index``.
+    """
+    cdef list result = L[:len(L)] # creates a copy
+    cdef size_t i
+    for i in range(len(L)):
+        if (<ETuple> PyList_GET_ITEM(L, i)).get_exp(index):
+            result.append((<ETuple> PyList_GET_ITEM(L, i)).divide_by_var(index))
+    return interred(result)
+
+cdef ETuple sum_from_list(list L, size_t s, size_t l):
+    """
+    Compute the vector sum of the ETuples in ``L[s:s+l]`` in a balanced way.
+    """
+    if l == 1:
+        return <ETuple>PyList_GET_ITEM(L,s)
+    if l == 2:
+        return (<ETuple>PyList_GET_ITEM(L,s)).eadd(<ETuple>PyList_GET_ITEM(L,s+1))
+    cdef size_t l2 = l // 2
+    cdef ETuple m1, m2
+    m1 = sum_from_list(L, s, l2)
+    m2 = sum_from_list(L, s+l2, l-l2)
+    return m1.eadd(m2)
+
+cdef bint HilbertBaseCase(Polynomial_integer_dense_flint fhs, Node D, tuple w) noexcept:
+    """
+    Try to compute the first Hilbert series of ``D.Id``, or return
+    ``NotImplemented``.
+
+    The third parameter ``w`` is a tuple of integers, the degree weights
+    to be used.
+
+    ``fhs`` is supposed to be zero. In some base cases, ``fhs`` will be
+    changed to the value of the Hilbert series and ``True`` is returned.
+    Otherwiese, ``False`` is returned.
+    """
+    cdef size_t i, j, exp
+    # First, the easiest cases:
+    if not D.Id: # The zero ideal
+        fmpz_poly_set_coeff_si(fhs._poly, 0, 1) # = PR(1)
+        return True
+    cdef ETuple m = <ETuple>PyList_GET_ITEM(D.Id, len(D.Id)-1)
+    if m._nonzero == 0: # The one ideal
+        return True
+
+    # Second, another reasy case: D.Id is generated by variables.
+    # D.Id is sorted ascendingly. Hence, if the last generator is a single
+    # variable, then ALL are.
+    cdef fmpz_poly_t poly_tmp
+    if m._nonzero == 1 and m._data[1] == 1:
+        fmpz_poly_init(poly_tmp)
+        fmpz_poly_set_coeff_si(poly_tmp, 0, 1)
+        fmpz_poly_set_coeff_si(fhs._poly, 0, 1) # = PR(1)
+        if w is None:
+            for i in range(len(D.Id)):
+                exp = (<ETuple>PyList_GET_ITEM(D.Id, i)).unweighted_degree()
+                fmpz_poly_set_coeff_si(poly_tmp, exp, -1)
+                fmpz_poly_mul(fhs._poly, fhs._poly, poly_tmp)
+                fmpz_poly_set_coeff_si(poly_tmp, exp, 0)
+        else:
+            for i in range(len(D.Id)):
+                exp = (<ETuple>PyList_GET_ITEM(D.Id, i)).weighted_degree(w)
+                fmpz_poly_set_coeff_si(poly_tmp, exp, -1)
+                fmpz_poly_mul(fhs._poly, fhs._poly, poly_tmp)
+                fmpz_poly_set_coeff_si(poly_tmp, exp, 0)
+        fmpz_poly_clear(poly_tmp)
+        return True # PR.prod([(1-t**degree(m,w)) for m in D.Id])
+
+    # Thirdly, we test for proper powers of single variables.
+    cdef bint easy = True
+    for i in range(len(D.Id)):
+        m = <ETuple>PyList_GET_ITEM(D.Id, i) # will be used later when we are breaking
+        if m._nonzero > 1: # i.e., the generator contains more than a single var
+            easy = False
+            break
+    if easy:
+        # The ideal is generated by some powers of single variables, i.e., it splits.
+        fmpz_poly_init(poly_tmp)
+        fmpz_poly_set_coeff_si(poly_tmp, 0, 1)
+        fmpz_poly_set_coeff_si(fhs._poly, 0, 1) # = PR(1)
+        if w is None:
+            for i in range(len(D.Id)):
+                exp = (<ETuple>PyList_GET_ITEM(D.Id, i)).unweighted_degree()
+                fmpz_poly_set_coeff_si(poly_tmp, exp, -1)
+                fmpz_poly_mul(fhs._poly, fhs._poly, poly_tmp)
+                fmpz_poly_set_coeff_si(poly_tmp, exp, 0)
+        else:
+            for i in range(len(D.Id)):
+                exp = (<ETuple>PyList_GET_ITEM(D.Id, i)).weighted_degree(w)
+                fmpz_poly_set_coeff_si(poly_tmp, exp, -1)
+                fmpz_poly_mul(fhs._poly, fhs._poly, poly_tmp)
+                fmpz_poly_set_coeff_si(poly_tmp, exp, 0)
+        fmpz_poly_clear(poly_tmp)
+        return True # PR.prod([(1-t**degree(m,w)) for m in D.Id])
+
+    easy = True
+    cdef ETuple m2
+    for j in range(i+1, len(D.Id)):
+        if (<ETuple>PyList_GET_ITEM(D.Id,j))._nonzero > 1:
+            # i.e., another generator contains more than a single var
+            easy = False
+            break
+    cdef fmpz_poly_t FirstSummand, SecondSummand
+    if easy:
+        # The ideal only has a single non-simple power, in position i.
+        # We have preserved that monomial, it is called m.
+        fmpz_poly_init(poly_tmp)
+        fmpz_poly_init(FirstSummand)
+        fmpz_poly_init(SecondSummand)
+        fmpz_poly_set_coeff_si(poly_tmp, 0, 1)
+        fmpz_poly_set_coeff_si(FirstSummand, 0, 1)
+        if w is None:
+            fmpz_poly_set_coeff_si(SecondSummand, m.unweighted_degree(), -1)
+            # Since the ideal is interreduced and all other monomials are
+            # simple powers, we have the following formula:
+            #~ return prod([(1-t**degree(m2,w)) for m2 in D.Id if m2 is not m])
+            #~   - t**degree(m,w)*prod((1-t**quotient_degree(m2,m,w)) for m2 in D.Id if m is not m2)
+            for j in range(len(D.Id)):
+                if i != j:
+                    m2 = <ETuple>PyList_GET_ITEM(D.Id,j)
+                    exp = m2.unweighted_degree()
+                    fmpz_poly_set_coeff_si(poly_tmp, exp, -1)
+                    fmpz_poly_mul(FirstSummand, FirstSummand, poly_tmp)
+                    fmpz_poly_set_coeff_si(poly_tmp, exp, 0)
+                    exp = m2.unweighted_quotient_degree(m)
+                    fmpz_poly_set_coeff_si(poly_tmp, exp, -1)
+                    fmpz_poly_mul(SecondSummand, SecondSummand, poly_tmp)
+                    fmpz_poly_set_coeff_si(poly_tmp, exp, 0)
+        else:
+            fmpz_poly_set_coeff_si(SecondSummand, m.weighted_degree(w), -1)
+            # Since the ideal is interreduced and all other monomials are
+            # simple powers, we have the following formula:
+            #~ return prod([(1-t**degree(m2,w)) for m2 in D.Id if m2 is not m])
+            #~   - t**degree(m,w)*prod((1-t**quotient_degree(m2,m,w)) for m2 in D.Id if m is not m2)
+            for j in range(len(D.Id)):
+                if i != j:
+                    m2 = <ETuple>PyList_GET_ITEM(D.Id,j)
+                    exp = m2.weighted_degree(w)
+                    fmpz_poly_set_coeff_si(poly_tmp, exp, -1)
+                    fmpz_poly_mul(FirstSummand, FirstSummand, poly_tmp)
+                    fmpz_poly_set_coeff_si(poly_tmp, exp, 0)
+                    exp = m2.weighted_quotient_degree(m,w)
+                    fmpz_poly_set_coeff_si(poly_tmp, exp, -1)
+                    fmpz_poly_mul(SecondSummand, SecondSummand, poly_tmp)
+                    fmpz_poly_set_coeff_si(poly_tmp, exp, 0)
+        fmpz_poly_clear(poly_tmp)
+        fmpz_poly_add(fhs._poly, fhs._poly, FirstSummand)
+        fmpz_poly_add(fhs._poly, fhs._poly, SecondSummand)
+        fmpz_poly_clear(FirstSummand)
+        fmpz_poly_clear(SecondSummand)
+        return True
+    # We are in a truly difficult case and give up for now...
+    return False
+
+cdef make_children(Node D, tuple w):
+    """
+    Create child nodes in ``D`` that allow to compute the first Hilbert
+    series of ``D.Id``.
+
+    Basically, the first Hilbert series of ``D.Id`` will be
+    ``D.LMult`` times the first Hilbert series of ``D.Left.Id``,
+    possibly plus ``D.RMult`` times the first Hilbert series of ``D.Right.Id``
+    if ``D.Right`` is not ``None``.
+    """
+    cdef size_t j, m
+    cdef int i
+    # Determine the variable that appears most often in the monomials.
+    # If "most often" means "only once", then instead we choose a variable that is
+    # guaranteed to appear in a composed monomial.
+    # We will raise it to a reasonably high power that still guarantees that
+    # many monomials will be divisible by it.
+    cdef ETuple all_exponents = sum_from_list(D.Id, 0, len(D.Id))
+    m = 0
+    cdef list max_exponents = []
+    for i in range(0, 2*all_exponents._nonzero, 2):
+        j = all_exponents._data[i+1]
+        if j > m:
+            max_exponents = [all_exponents._data[i]]
+            m = j
+        elif j == m:
+            max_exponents.append(all_exponents._data[i])
+    cdef size_t e # will be the exponent, if the monomial used for cutting is power of a variable
+    cdef ETuple cut
+    cdef list Id2
+    cdef size_t cut_deg
+    # Cases:
+    # - m==1, which means that all variables occur at most once.
+    #   => we cut by a variable that appears in a decomposable generator
+    # - max_exponents = [j]
+    #   => cut = var(j)**e, where e is the median of all j-exponents
+    # - max_exponents = [j1,...,jk]
+    #   => cut = prod([var(j1),...,var(jk)]) or something of that type.
+    if m == 1:
+        # D.Id is sorted, which means that the last generator is decomposable
+        all_exponents = <ETuple> PyList_GET_ITEM(D.Id, len(D.Id)-1)
+        j = all_exponents._data[2*all_exponents._nonzero-2]
+        cut = all_exponents._new()
+        cut._nonzero = 1
+        cut._data = <int*> sig_malloc(sizeof(int)*2)
+        cut._data[0] = j
+        cut._data[1] = 1
+        if w is None:
+            cut_deg = cut.unweighted_degree()
+        else:
+            cut_deg = cut.weighted_degree(w)
+        # var(j) *only* appears in D.Id[-1]. Hence, D.Id+var(j) will be a split case,
+        # with var(j) and D.Id[:-1]. So, we do the splitting right now.
+        # Only the last generator contains var(j). Hence, D.Id/var(j) is obtained
+        # from D.Id by adding the quotient of its last generator divided by var(j),
+        # of course followed by interreduction.
+        #~ D.LMult = 1 - t**degree(cut,w)
+        fmpz_poly_set_coeff_si(D.LMult, 0, 1)
+        fmpz_poly_set_coeff_si(D.LMult, cut_deg, -1)
+        D.Left  = Node.__new__(Node)
+        D.Left.Id = D.Id[:len(D.Id)-1]
+        D.Left.Back = D
+        Id2 = D.Id[:len(D.Id)-1]
+        Id2.append(all_exponents.divide_by_var(j))
+        D.Right = Node.__new__(Node)
+        D.Right.Id = interred(Id2)
+        D.Right.Back = D
+        #~ D.RMult = 1 - D.LMult
+        fmpz_poly_set_coeff_si(D.RMult, cut_deg, 1)
+    else:
+        j = max_exponents[0]
+        e = median([(<ETuple> mon).get_exp(j) for mon in D.Id if (<ETuple> mon).get_exp(j)])
+        cut = all_exponents._new()
+        cut._nonzero = 1
+        cut._data = <int*> sig_malloc(sizeof(int)*2)
+        cut._data[0] = j
+        cut._data[1] = e
+        if w is None:
+            cut_deg = cut.unweighted_degree()
+        else:
+            cut_deg = cut.weighted_degree(w)
+        try:
+            i = D.Id.index(cut)
+        except ValueError:
+            i = -1
+        if i >= 0:
+            # var(j)**e is a generator. Hence, e is the maximal exponent of
+            # var(j) in D.Id, by D.Id being interreduced. But it also is the
+            # truncated median, hence, there cannot be smaller exponents
+            # (for otherwise the median would be strictly smaller than the
+            # maximum). Conclusion: var(j) only appears in the generator var(j)**e
+            #  -- we have a split case.
+            Id2 = D.Id[:len(D.Id)]
+            Id2.pop(i)
+            #~ D.LMult = 1-t**degree(cut,w)
+            fmpz_poly_set_coeff_si(D.LMult, 0, 1)
+            fmpz_poly_set_coeff_si(D.LMult, cut_deg, -1)
+            D.Left  = Node.__new__(Node)
+            D.Left.Id = Id2
+            D.Left.Back = D
+            D.Right = None
+        else:
+            cut = all_exponents._new()
+            cut._nonzero = 1
+            cut._data = <int*> sig_malloc(sizeof(int)*2)
+            cut._data[0] = j
+            cut._data[1] = e
+            if e > 1:
+                # D.LMult = 1
+                fmpz_poly_set_coeff_si(D.LMult, 0, 1)
+                Id2 = D.Id[:len(D.Id)]
+                Id2.append(cut)
+                D.Left  = Node.__new__(Node)
+                D.Left.Id = interred(Id2)
+                D.Left.Back = D
+                D.Right = Node.__new__(Node)
+                D.Right.Id = quotient(D.Id,cut)
+                D.Right.Back = D
+            else:
+                # m > 1, therefore var(j) cannot be a generator (D.Id is interreduced).
+                # D.Id+var(j) will be a split case. So, we do the splitting right now.
+                #~ D.LMult = 1-t**(1 if w is None else w[j])
+                fmpz_poly_set_coeff_si(D.LMult, 0, 1)
+                fmpz_poly_set_coeff_si(D.LMult, (1 if w is None else w[j]), -1)
+                D.Left  = Node.__new__(Node)
+                D.Left.Id = [(<ETuple> mon) for mon in D.Id if (<ETuple> mon).get_exp(j) == 0]
+                D.Left.Back = D
+                D.Right = Node.__new__(Node)
+                D.Right.Id = quotient_by_var(D.Id, j)
+                D.Right.Back = D
+            #~ D.RMult = t**(e if w is None else e*w[j])
+            fmpz_poly_set_coeff_si(D.RMult, (e if w is None else e * w[j]), 1)
+     #else:
+     #    It may be a good idea to form the product of some of the most
+     #    frequent variables. But this isn't implemented yet. TODO?
+
+
+def first_hilbert_series(I, grading=None, return_grading=False):
+    """
+    Return the first Hilbert series of the given monomial ideal.
+
+    INPUT:
+
+    - ``I`` -- a monomial ideal (possibly defined in singular)
+    - ``grading`` -- (optional) a list or tuple of integers used as
+      degree weights
+    - ``return_grading`` -- boolean (default: ``False``); whether to return the
+      grading
+
+    OUTPUT:
+
+    A univariate polynomial, namely the first Hilbert function of ``I``, and
+    if ``return_grading==True`` also the grading used to compute the series.
+
+    EXAMPLES::
+
+        sage: from sage.rings.polynomial.hilbert import first_hilbert_series
+
+        sage: # needs sage.libs.singular
+        sage: R = singular.ring(0,'(x,y,z)','dp')
+        sage: I = singular.ideal(['x^2','y^2','z^2'])
+        sage: first_hilbert_series(I)
+        -t^6 + 3*t^4 - 3*t^2 + 1
+        sage: first_hilbert_series(I, return_grading=True)
+        (-t^6 + 3*t^4 - 3*t^2 + 1, (1, 1, 1))
+        sage: first_hilbert_series(I, grading=(1,2,3))
+        -t^12 + t^10 + t^8 - t^4 - t^2 + 1
+
+    TESTS:
+
+    We test against some corner cases::
+
+        sage: R.<x,y,z> = PolynomialRing(QQ)
+        sage: I = 0*R
+        sage: first_hilbert_series(I)
+        1
+        sage: first_hilbert_series(singular(I))                                         # needs sage.libs.singular
+        1
+        sage: I = 1*R
+        sage: first_hilbert_series(I)
+        0
+        sage: first_hilbert_series(singular(I))                                         # needs sage.libs.singular
+        0
+    """
+    from sage.rings.integer_ring import ZZ
+    from sage.rings.polynomial.polynomial_ring_constructor import PolynomialRing
+    PR = PolynomialRing(ZZ, 't')
+    cdef Node AN
+    # The "active node". If a recursive computation is needed, it will be equipped
+    # with a 'Left' and a 'Right' child node, and some 'Multipliers'. Later, the first Hilbert
+    # series of the left child node will be stored in 'LeftFHS', and together with
+    # the first Hilbert series of the right child node and the multiplier yields
+    # the first Hilbert series of 'Id'.
+    cdef tuple w
+    cdef Polynomial_integer_dense_flint fhs = Polynomial_integer_dense_flint.__new__(Polynomial_integer_dense_flint)
+    fhs._parent = PR
+    fhs._is_gen = 0
+    if isinstance(I, sage.interfaces.abc.SingularElement):
+        S = I._check_valid()
+        # First, we need to deal with quotient rings, which also covers the case
+        # of graded commutative rings that arise as cohomology rings in odd characteristic.
+        # We replace everything by a commutative version of the quotient ring.
+        br = S('basering')
+        if S.eval('isQuotientRing(basering)')=='1':
+            L = S('ringlist(basering)')
+            R = S('ring(list(%s[1..3],ideal(0)))' % L.name())
+            R.set_ring()
+            I = S('fetch(%s,%s)+ideal(%s)' % (br.name(), I.name(), br.name()))
+
+        I = [ETuple([int(x) for x in S.eval('string(leadexp({}[{}]))'.format(I.name(), i)).split(',')])
+              for i in range(1,int(S.eval('size({})'.format(I.name())))+1)]
+        br.set_ring()
+        if grading is None:
+            w = tuple(int(S.eval('deg(var({}))'.format(i))) for i in range(1,int(S.eval('nvars(basering)'))+1))
+        else:
+            w = tuple(grading)
+    else:
+        try:
+            I = [bla.exponents()[0] for bla in I if bla]
+        except TypeError:
+            I = [bla.exponents()[0] for bla in I.gens() if bla]
+        if grading is not None:
+            w = tuple(grading)
+        else:
+            w = None
+
+    AN = Node.__new__(Node)
+    AN.Id = interred(I)
+    AN.Back = None
+
+    # Invariant of this function:
+    # At each point, got_result will be false, or got_result is true and fhs
+    # is the first Hilbert series of AN.
+    cdef bint got_result = HilbertBaseCase(fhs, AN, w)
+    while True:
+        if not got_result:
+            make_children(AN, w)
+            AN = AN.Left
+            got_result = HilbertBaseCase(fhs, AN, w)
+        else:
+            if AN.Back is None: # We are back on top, i.e., fhs is the First Hilber Series of I
+                if return_grading:
+                    return fhs, w
+                else:
+                    return fhs
+            if AN is AN.Back.Left: # We store fhs and proceed to the sibling
+                # ... unless there is no sibling
+                if AN.Back.Right is None:
+                    AN = AN.Back
+                    #~ fhs *= AN.LMult
+                    fmpz_poly_mul(fhs._poly, fhs._poly, AN.LMult)
+                    got_result = True
+                else:
+                    fmpz_poly_set(AN.Back.LeftFHS, fhs._poly)
+                    fmpz_poly_set_si(fhs._poly, 0)
+                    AN = AN.Back.Right
+                    AN.Back.Left = None
+                    got_result = HilbertBaseCase(fhs, AN, w)
+            else: # FHS of the left sibling is stored, of the right sibling is known.
+                AN = AN.Back
+                AN.Right = None
+                #~ fhs = AN.LMult*AN.LeftFHS + AN.RMult*fhs
+                fmpz_poly_mul(AN.LMult, AN.LMult, AN.LeftFHS)
+                fmpz_poly_mul(AN.RMult, AN.RMult, fhs._poly)
+                fmpz_poly_add(fhs._poly, AN.LMult, AN.RMult)
+                got_result = True
+
+
+def hilbert_poincare_series(I, grading=None):
+    r"""
+    Return the Hilbert Poincaré series of the given monomial ideal.
+
+    INPUT:
+
+    - ``I`` -- a monomial ideal (possibly defined in Singular)
+    - ``grading`` -- (optional) a tuple of degree weights
+
+    EXAMPLES::
+
+        sage: # needs sage.libs.singular
+        sage: from sage.rings.polynomial.hilbert import hilbert_poincare_series
+        sage: R = PolynomialRing(QQ,'x',9)
+        sage: I = [m.lm()
+        ....:      for m in ((matrix(R, 3, R.gens())^2).list() * R).groebner_basis()] * R
+        sage: hilbert_poincare_series(I)
+        (t^7 - 3*t^6 + 2*t^5 + 2*t^4 - 2*t^3 + 6*t^2 + 5*t + 1)/(t^4 - 4*t^3 + 6*t^2 - 4*t + 1)
+        sage: hilbert_poincare_series((R * R.gens())^2, grading=range(1,10))
+        t^9 + t^8 + t^7 + t^6 + t^5 + t^4 + t^3 + t^2 + t + 1
+
+    The following example is taken from :issue:`20145`::
+
+        sage: # needs sage.libs.singular
+        sage: n=4; m=11; P = PolynomialRing(QQ, n*m, "x"); x = P.gens(); M = Matrix(n, x)
+        sage: from sage.rings.polynomial.hilbert import first_hilbert_series
+        sage: I = P.ideal(M.minors(2))
+        sage: J = P * [m.lm() for m in I.groebner_basis()]
+        sage: hilbert_poincare_series(J).numerator()
+        120*t^3 + 135*t^2 + 30*t + 1
+        sage: hilbert_poincare_series(J).denominator().factor()
+        (t - 1)^14
+    """
+    cdef Polynomial_integer_dense_flint HP
+    HP, grading = first_hilbert_series(I, grading=grading, return_grading=True)
+    PR = HP._parent
+
+    # If grading was None, but the ideal lives in Singular, then grading is now
+    # the degree vector of Singular's basering.
+    # Otherwise, it may still be None.
+    if grading is None:
+        if HP.leading_coefficient() >= 0:
+            return HP / ((1-PR.gen())**I.ring().ngens())
+        return (-HP) / (-((1-PR.gen())**I.ring().ngens()))
+    else:
+        t = PR.gen()
+        if HP.leading_coefficient() >= 0:
+            return HP / PR.prod([(1-t**d) for d in grading])
+        return (-HP) / (-PR.prod([(1-t**d) for d in grading]))

sage/rings/polynomial/polynomial_complex_arb.pxd

@@ -0,0 +1,7 @@
+# sage_setup: distribution = sagemath-flint
+from sage.libs.flint.acb_poly cimport *
+from sage.rings.polynomial.polynomial_element cimport Polynomial
+
+cdef class Polynomial_complex_arb(Polynomial):
+    cdef acb_poly_struct[1] _poly # https://github.com/cython/cython/issues/1984
+    cdef Polynomial_complex_arb _new(self)