julian/Calabi-Yau-Period-Geometry
1
0
Fork 0
Code Issues 4 Pull Requests Actions Packages Projects Releases Wiki Activity

Initial setup for period computation (#13)
CI / lint (push) Successful in 14s
Details
CI / test (push) Successful in 1m32s
Details

Browse Source 
Introduces:
 - class for Picard--Fuchs operators and their ideals
 - class for periods (their complex linear combinations in the Frobenius bases)

Implements:
 - method to obtain operator from period
 - method to get power series solution (at given indicials) to PF ideal

---------

Co-authored-by: Julian Piribauer <julian.piribauer@gmail.com>
Reviewed-on: #13
This commit was merged in pull request #13.
This commit is contained in:
julian
2026-08-17 19:51:31 +02:00
co-authored by Julian Piribauer
parent 59dbb8f8bf
commit 51a5612fe1
5 changed files with 798 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+83 -6
View File
@@ -12,12 +12,21 @@ elliptic_curve_D = ToricPolytopeProjectiveSpace([1, 2, 3], model_name="elliptic_
CY3_quintic = ToricPolytopeProjectiveSpace([1, 1, 1, 1, 1], model_name="quintic") CY3_quintic = ToricPolytopeProjectiveSpace([1, 1, 1, 1, 1], model_name="quintic")
CY3_bicubic = ToricPolytopeCICY([[3, 3]], model_name="bi-cubic") CY3_bicubic = ToricPolytopeCICY([[3, 3]], model_name="bi-cubic")
CICY3_two_parameter_manual_nef = ToricPolytope([[1,0,0,0,0,0],[0,1,0,0,0,0],[0,0,1,0,0,0], CICY3_two_parameter_manual_nef = ToricPolytope(
[0,0,0,1,0,0],[0,0,0,0,1,0],[0,0,0,0,0,1], [
[-1,-1,0,0,0,0],[0,0,-1,-1,-1,-1]], [1, 0, 0, 0, 0, 0],
nef_partition=[[0,1,2,3], [4,5,6,7]]) [0, 1, 0, 0, 0, 0],
[0, 0, 1, 0, 0, 0],
[0, 0, 0, 1, 0, 0],
[0, 0, 0, 0, 1, 0],
[0, 0, 0, 0, 0, 1],
[-1, -1, 0, 0, 0, 0],
[0, 0, -1, -1, -1, -1],
],
nef_partition=[[0, 1, 2, 3], [4, 5, 6, 7]],
)
CICY5_two_parameter = ToricPolytopeCICY([[6, 1],[0, 2]]) CICY5_two_parameter = ToricPolytopeCICY([[6, 1], [0, 2]])
``` ```
The discriminant factors and topological data, e.g. for the quintic, can then be computed with the methods below. The discriminant factors and topological data, e.g. for the quintic, can then be computed with the methods below.
@@ -71,4 +80,72 @@ DEBUG:__main__:
The result is saved in the folder `data/topdata` as a JSON file &mdash; giving a model name helps keeping The result is saved in the folder `data/topdata` as a JSON file &mdash; giving a model name helps keeping
track of these outputs. track of these outputs.
Note that for Calabi&ndash;Yau dimensions larger than four, the additional Note that for Calabi&ndash;Yau dimensions larger than four, the additional
## period_computation
`period_computation.sage` provides classes for working with Picard&ndash;Fuchs operators and their
period solutions: `PFOperator`, `PFIdeal` and `Period`, together with Ansatz variants of the first and
last (`PFOperatorAnsatz`, `PeriodAnsatz`) used to search for unknown operators or periods of a given
z- and theta-degree.
A `PFOperator` is parsed from a string in the variables `z0, ..., z<n-1>` and `theta0, ..., theta<n-1>`,
the logarithmic derivatives theta_i = z_i d/dz_i. For example, the quintic's Picard&ndash;Fuchs operator:
```python
L = PFOperator(
"theta0^4 - 3125*z0*theta0^4 - 6250*z0*theta0^3 - 4375*z0*theta0^2 - 1250*z0*theta0 - 120*z0",
no_variables=1,
)
L.simplify().operator_string
```
```term
'-5*(5*theta0 + 4)*(5*theta0 + 3)*(5*theta0 + 2)*(5*theta0 + 1)*z0 + theta0^4'
```
An operator (or a `PFIdeal` of several) can be solved for its power series solution at given indicial
exponents and order.
```python
ideal = PFIdeal([L])
period = ideal.find_power_series_solution(indicials=[0], order=3)[0]
period.period_string
```
```term
'168168000*z0^3 + 113400*z0^2 + 120*z0 + 1'
```
The reverse direction is supported too: given a `Period`, `find_annihilating_operators` searches for
`PFOperator`s of a given z- and theta-degree that annihilate it, by solving an Ansatz of unknown
coefficients via linear algebra. Both directions extend to several moduli, e.g. for the two-parameter
model P_{2,2,2,1,1}[8]:
```python
M1 = PFOperator(
"theta1*(-2*theta0 + 2*theta1 - 1) + 2*(theta0 - 2*theta1 - 1)*(theta0 - 2*theta1)*z1",
no_variables=2,
)
M2 = PFOperator(
"theta0^2*(2*(theta0 - 2*theta1)*z1 - theta1) - 16*(2*theta0 + 1)*(4*theta0 + 1)*(4*theta0 + 3)*z0*z1",
no_variables=2,
)
ideal = PFIdeal([M1, M2])
period = ideal.find_power_series_solution(indicials=[0, 1 / 2], order=6)[0]
recovered = period.find_annihilating_operators(z_degree=1, theta_degree=2)
period.period_string
recovered[0].simplify().operator_string
```
```term
'-1/45045*(60886425600*z0^3*z1^3 - 2767564800*z0^2*z1^4 + 100638720*z0*z1^5 - 14192640*z1^6 + 830269440*z0^2*z1^3 - 30750720*z0*z1^4 + 4193280*z1^5 - 242161920*z0^2*z1^2 + 9884160*z0*z1^3 - 1281280*z1^4 - 3459456*z0*z1^2 + 411840*z1^3 + 1441440*z0*z1 - 144144*z1^2 + 60060*z1 - 45045)*sqrt(z1)'
'-2*(theta0 - 2*theta1)*(theta0 - 2*theta1 - 1)*z1 + (2*theta0 - 2*theta1 + 1)*theta1'
```
`recovered[0]` is, up to scale, `M1` &mdash; recovered purely from `M1`, `M2`'s shared power series
solution.
pyproject.toml
+3
View File
@@ -15,3 +15,6 @@ ignore = [
# test_smoke.py star-imports sage.all and loads a .sage file, so ruff can't see where its names come from. # test_smoke.py star-imports sage.all and loads a .sage file, so ruff can't see where its names come from.
"tests/test_topdata_and_disc.py" = ["F403", "F405"] "tests/test_topdata_and_disc.py" = ["F403", "F405"]
# Same star-import + load() pattern as above.
"tests/test_period_computation.py" = ["F403", "F405"]
sage/period_computation.sage
+503
View File
@@ -0,0 +1,503 @@
import copy
import logging
from sage.all import sage_eval, PolynomialRing, QQ, SR, log, matrix, prod, var
load("sage/util.py")
# Logger
logger = logging.getLogger(__name__)
logging.basicConfig(level=logging.DEBUG)
class PFOperator:
"""
A class representing a Picard-Fuchs operator in a given number of variables (moduli).
Independent of the given order, the variables are assumed to be left of
the derivatives.
The variables extra_locals are needed for symbolic coefficients used for
operator Ansätze.
"""
def _operator_from_string(self, operator_string: str, extra_locals: dict = None):
z_names = ['z%d' % i for i in range(self.no_variables)]
theta_names = ['theta%d' % i for i in range(self.no_variables)]
base_ring = SR if extra_locals else QQ
self.ring = PolynomialRing(base_ring, z_names + theta_names)
self.z_gens = self.ring.gens()[:self.no_variables]
self.theta_gens = self.ring.gens()[self.no_variables:]
parse_locals = dict(self.ring.gens_dict())
if extra_locals:
parse_locals.update(extra_locals)
try:
operator = self.ring(sage_eval(operator_string, locals=parse_locals))
return operator
except Exception as e:
raise ValueError("Invalid operator string: %s" % e)
def __init__(self, operator_string: str, no_variables: int = 1, extra_locals: dict = None):
self.no_variables = no_variables
self.operator_string = operator_string
self.operator = self._operator_from_string(operator_string, extra_locals=extra_locals)
logger.info("Initialised PFOperator: %s", self.operator)
def simplify(self) -> "PFOperator":
"""
Groups the operator's terms by z-monomial and factorises the theta-polynomial
multiplying each z-monomial, e.g. turning a computed quintic operator into the
well-known theta0^4 - 5*z0*(5*theta0 + 1)*(5*theta0 + 2)*(5*theta0 + 3)*(5*theta0 + 4).
"""
z_monomial_theta_parts = {}
for coeff, monomial in self.operator:
exponents = monomial.exponents()[0]
z_exponents = exponents[:self.no_variables]
theta_exponents = exponents[self.no_variables:]
theta_monomial = SR(1)
for gen, exp in zip(self.theta_gens, theta_exponents):
theta_monomial *= SR(gen) ** exp
z_monomial_theta_parts[z_exponents] = z_monomial_theta_parts.get(z_exponents, SR(0)) + SR(coeff) * theta_monomial
simplified_expr = SR(0)
for z_exponents, theta_part in z_monomial_theta_parts.items():
z_monomial = SR(1)
for gen, exp in zip(self.z_gens, z_exponents):
z_monomial *= SR(gen) ** exp
simplified_expr += theta_part.factor() * z_monomial
simplified = copy.copy(self)
simplified.operator_string = str(simplified_expr)
logger.debug("Simplified PFOperator to: %s", simplified.operator_string)
return simplified
class PFOperatorAnsatz(PFOperator):
"""
A class for Picard-Fuchs operator Ansätze, characterised by number of variables and their z- and theta-multi-degrees.
"""
def __init__(self, no_variables: int, theta_degree: int, z_degree: int):
self.no_variables = no_variables
self.theta_degree = theta_degree
self.z_degree = z_degree
z_indices = _multi_indices(self.no_variables, z_degree)
theta_indices = _multi_indices(self.no_variables, theta_degree)
# Every (z_index, theta_index) pair allowed by the degree bounds gets its own
# fresh unknown, to be solved for once the Ansatz is applied to a period.
operator_terms = [
(var("b_" + "_".join(str(x) for x in z_index + theta_index)), (z_index, theta_index))
for z_index in z_indices
for theta_index in theta_indices
]
self.unknowns = [coeff for coeff, _ in operator_terms]
def _monomial_factors(index, name):
return [f"{name}{i}^{exp}" for i, exp in enumerate(index) if exp > 0]
operator_string = " + ".join(
"*".join([str(coeff), *_monomial_factors(z_index, "z"), *_monomial_factors(theta_index, "theta")])
for coeff, (z_index, theta_index) in operator_terms
)
extra_locals = {str(coeff): coeff for coeff in self.unknowns}
super().__init__(operator_string=operator_string, no_variables=no_variables, extra_locals=extra_locals)
class PFIdeal:
"""
A class representing a Picard-Fuchs ideal, which is a collection of PFOperators.
"""
def __init__(self, operators: list):
if not all(op.no_variables == operators[0].no_variables for op in operators):
raise ValueError("All operators must have the same number of variables.")
self.no_variables = operators[0].no_variables
self.operators = operators
logger.info("Initialised PFIdeal with %d operator(s).", len(self.operators))
def add_operator(self, pf_operator: PFOperator):
self.operators.append(pf_operator)
logger.info("Added PFOperator to PFIdeal: %s", pf_operator.operator_string)
def remove_operator(self, pf_operator: PFOperator):
self.operators.remove(pf_operator)
logger.info("Removed PFOperator from PFIdeal: %s", pf_operator.operator_string)
def find_power_series_solution(self, indicials: list, order: int) -> list:
"""
Finds power series solutions (no logs) to this PFIdeal at given indicial exponents/order.
A term c*z^p*theta^q sends a_k*z^k (true exponent k+indicials) to
c*(k+indicials)^q*a_k*z^(k+p): it shifts index k up by p (p>=0), never down or sideways.
E.g. z0*theta0 sends a_k*z0^k to (k+rho0)*a_k*z0^(k+1).
This is the multivariate Frobenius method: canonical series solutions of a regular
holonomic D-ideal via its indicial ideal. See M. Saito, B. Sturmfels, N. Takayama,
"Gröbner Deformations of Hypergeometric Differential Equations", Algorithms and
Computation in Mathematics vol. 6, Springer, 2000, chs. 2-3.
"""
if len(indicials) != self.no_variables:
raise ValueError(
"Indicials must have length no_variables=%d, got %d." % (self.no_variables, len(indicials))
)
indicials = [QQ(rho) for rho in indicials]
operator_terms = [
[
(monomial.exponents()[0][:self.no_variables], monomial.exponents()[0][self.no_variables:], coeff)
for coeff, monomial in pf_operator.operator
]
for pf_operator in self.operators
]
def eigenvalue(k, q):
# theta_i^q_i acts on z_i^(k_i + indicials[i]) as multiplication by
# (k_i + indicials[i])^q_i; theta^q's combined eigenvalue is the product over i.
value = QQ(1)
for i in range(self.no_variables):
if q[i]:
value *= (k[i] + indicials[i]) ** q[i]
return value
# Process multi-indices in order of increasing total degree: since every operator
# monomial has p >= 0 (componentwise), the z^m coefficient of L(y) only ever
# depends on a_k for k <= m, so by this point every k < m has already been solved.
z_indices = sorted(_multi_indices(self.no_variables, order), key=sum)
# solved[k] holds a_k written as a vector of coefficients over the `dimension`
# independent solutions found so far (a basis of the solution space up to k).
dimension = 0
solved = {}
for m in z_indices:
# For each operator, "coefficient of z^m in L(y) = 0" splits into a diagonal
# part (the p=0, theta-only monomials, whose unknown is a_m itself) plus a
# known part contributed by already-solved a_k with k = m - p, p > 0.
diagonals = []
known_contributions = []
for terms in operator_terms:
diagonal = QQ(0)
contribution = [QQ(0)] * dimension
for p, q, coeff in terms:
k = tuple(m[i] - p[i] for i in range(self.no_variables))
if any(ki < 0 for ki in k):
continue # this monomial would need a_k for a negative multi-index k: no such term
ev = eigenvalue(k, q)
if ev == 0:
continue # theta^q kills z_i^(k_i + indicials[i]) here, so this monomial contributes nothing
if k == m:
diagonal += coeff * ev # p = 0: coefficient multiplying the still-unknown a_m
else:
k_vector = solved[k] # p > 0: a_k is already known, add its contribution
for i in range(dimension):
contribution[i] += coeff * ev * k_vector[i]
diagonals.append(diagonal)
known_contributions.append(contribution)
# An operator with a nonzero diagonal lets us solve a_m = -(known part)/diagonal
# directly; this is exactly the indicial equation being nonzero at m + indicials.
active = next((r for r, d in enumerate(diagonals) if d != 0), None)
if active is not None:
value = [-known_contributions[active][i] / diagonals[active] for i in range(dimension)]
# Every operator's equation at m must independently be satisfied by this
# same a_m; disagreement means the ideal is inconsistent with these indicials.
for r, d in enumerate(diagonals):
if any(d * value[i] + known_contributions[r][i] != 0 for i in range(dimension)):
raise ValueError(
"Inconsistent Picard-Fuchs ideal or indicial exponents %s at multidegree %s."
% (indicials, m)
)
solved[m] = value
else:
# Resonance: every operator's indicial part vanishes at m + indicials, so a_m
# cannot be pinned down by this equation. If the already-known lower-degree
# data still forces a nonzero constraint here, satisfying it would require a
# log(z)-term solution, which this method (deliberately) does not compute.
if any(x != 0 for contribution in known_contributions for x in contribution):
raise ValueError(
"Resonance at multidegree %s for indicials %s would require a logarithmic "
"solution, which find_power_series_solution does not compute." % (m, indicials)
)
# Otherwise a_m is genuinely free: it starts a new independent solution, so
# extend every previously solved coefficient with a 0 in this new direction.
dimension += 1
for v in solved.values():
v.append(QQ(0))
solved[m] = [QQ(0)] * (dimension - 1) + [QQ(1)]
log_index = tuple([0] * self.no_variables)
solutions = [
Period(
no_variables=self.no_variables,
coefficients={log_index: {m: solved[m][i] for m in z_indices}},
order=order,
indicials=indicials,
)
for i in range(dimension)
]
logger.info(
"Found %d power series solution(s) for indicials %s at order %d.", len(solutions), indicials, order
)
return solutions
class Period:
"""
A class representing a period as a formal power series in z-variables and their logs.
The coefficients are stored in a dictionary of dictionaries, where the first key is the multi-index
of the logarithmic part and the second key is the multi-index of the z-variables.
So, for example, the coefficient of
log(z0)^2 * log(z1) * z0^3 * z1^2 would be stored as coefficients[(2, 1)][(3, 2)]
"""
def _initialise_ring(self):
z_names = ['z%d' % i for i in range(self.no_variables)]
log_names = ['L%d' % i for i in range(self.no_variables)]
ring = PolynomialRing(QQ, z_names + log_names)
z_gens = ring.gens()[:self.no_variables]
log_gens = ring.gens()[self.no_variables:]
return ring, z_gens, log_gens # type: (PolynomialRing, list, list)
def _period_from_string(self, period_string: str) -> dict:
ring, z_gens, log_gens = self._initialise_ring()
log_of_z = dict(zip(z_gens, log_gens))
def log(zi):
try:
return log_of_z[zi]
except (KeyError, TypeError):
raise ValueError(
"log(...) may only be applied to one of the z-variables z0, ..., z%d"
% (self.no_variables - 1)
)
parse_locals = dict(ring.gens_dict())
parse_locals['log'] = log
try:
expression = ring(sage_eval(period_string, locals=parse_locals))
except Exception as e:
raise ValueError("Invalid period string: %s" % e)
coefficients = {}
for coeff, monomial in expression:
exponents = monomial.exponents()[0]
z_index = tuple(exponents[:self.no_variables])
log_index = tuple(exponents[self.no_variables:])
coefficients.setdefault(log_index, {})[z_index] = coeff
return coefficients
def _period_to_string(self) -> str:
ring, z_gens, log_gens = self._initialise_ring()
expression = ring(0)
for log_index, z_dict in self.coefficients.items():
for z_index, coeff in z_dict.items():
monomial = coeff
for i in range(self.no_variables):
monomial *= (log_gens[i] ** log_index[i]) * (z_gens[i] ** z_index[i])
expression += monomial
log_substitutions = {log_gens[i]: log(SR(z_gens[i])) for i in range(self.no_variables)}
result = SR(expression).subs(log_substitutions)
# A nonzero indicial ρ_i means the coefficients above are for z_i^k, but the
# actual solution is z_i^(ρ_i + k); make that explicit in the printed form.
indicial_prefactor = prod(
(SR(z_gens[i]) ** self.indicials[i] for i in range(self.no_variables) if self.indicials[i] != 0),
SR(1),
)
if indicial_prefactor != 1:
result *= indicial_prefactor
return str(result)
def _max_z_degree(self) -> int:
# Highest total z-degree (sum of the z-multi-index) among all coefficients.
if not self.coefficients:
return 0
return max(
sum(z_index)
for z_dict in self.coefficients.values()
for z_index in z_dict
)
def _truncate_coefficients(self, order: int) -> dict:
# Drop every (log_index, z_index) entry whose total z-degree exceeds order,
# removing it from the dictionary rather than merely zeroing it out.
truncated = {}
for log_index, z_dict in self.coefficients.items():
kept = {z_index: coeff for z_index, coeff in z_dict.items() if sum(z_index) <= order}
if kept:
truncated[log_index] = kept
return truncated
def _apply_theta(self, coefficients: dict, index: int) -> dict:
"""
Applies the logarithmic derivative theta_i = z_i * d/dz_i once to a coefficients dict of the same
shape as self.coefficients. It uses the product rule
theta_i(z^a log(z)^k) = a_i * z^a log(z)^k + k_i * z^a log(z)^(k - e_i).
z_index entries are offsets from self.indicials: a stored z_index of a really means
z^(a + self.indicials[index]), so theta_i's eigenvalue is a_i + self.indicials[index]
rather than the bare a_i (self.indicials is all-zero unless explicitly given, in which
case this reduces to the ordinary power-series rule).
"""
result = {}
for log_index, z_dict in coefficients.items():
for z_index, coeff in z_dict.items():
a_i = z_index[index] + self.indicials[index]
if a_i != 0:
inner = result.setdefault(log_index, {})
inner[z_index] = inner.get(z_index, 0) + a_i * coeff
k_i = log_index[index]
if k_i != 0:
lowered_log_index = log_index[:index] + (k_i - 1,) + log_index[index + 1:]
inner = result.setdefault(lowered_log_index, {})
inner[z_index] = inner.get(z_index, 0) + k_i * coeff
return result
def apply_operator(self, pf_operator: PFOperator) -> "Period":
"""
Applies a PFOperator to this period and returns the result as a new Period,
truncated to self.order. Each monomial of the operator is normalised as
coeff * z^p * theta^q (see PFOperator's docstring), so theta^q is applied
to the period first and the result is then multiplied by coeff * z^p.
"""
if pf_operator.no_variables != self.no_variables:
raise ValueError(
"Variable count mismatch: period has %d variable(s), operator has %d."
% (self.no_variables, pf_operator.no_variables)
)
result_coefficients = {}
for monomial_coeff, monomial in pf_operator.operator:
exponents = monomial.exponents()[0]
z_exponents = exponents[:self.no_variables]
theta_exponents = exponents[self.no_variables:]
term = self.coefficients
for i, power in enumerate(theta_exponents):
for _ in range(power):
term = self._apply_theta(term, i)
for log_index, z_dict in term.items():
inner = result_coefficients.setdefault(log_index, {})
for z_index, coeff in z_dict.items():
shifted_z_index = tuple(z_index[i] + z_exponents[i] for i in range(self.no_variables))
inner[shifted_z_index] = inner.get(shifted_z_index, 0) + monomial_coeff * coeff
result_coefficients = {
log_index: {z_index: c for z_index, c in z_dict.items() if c != 0}
for log_index, z_dict in result_coefficients.items()
}
result_coefficients = {log_index: z_dict for log_index, z_dict in result_coefficients.items() if z_dict}
return Period(
no_variables=self.no_variables,
coefficients=result_coefficients,
order=self.order,
indicials=self.indicials,
)
def find_annihilating_operators(self, z_degree: int, theta_degree: int) -> list:
"""
Finds operators annihilating this period among PFOperator Ansätze of the given
z- and theta-degree.
"""
ansatz = PFOperatorAnsatz(self.no_variables, theta_degree, z_degree)
unknowns = ansatz.unknowns
equations = [
coeff
for z_dict in self.apply_operator(ansatz).coefficients.values()
for coeff in z_dict.values()
]
coefficient_rows = [[SR(equation).coefficient(b) for b in unknowns] for equation in equations]
kernel_basis = matrix(QQ, coefficient_rows, ncols=len(unknowns)).right_kernel().basis()
operators = []
for basis_vector in kernel_basis:
solution = {unknowns[j]: basis_vector[j] for j in range(len(unknowns))}
solved_operator = ansatz.operator.map_coefficients(lambda c: SR(c).subs(solution))
operators.append(PFOperator(str(solved_operator), no_variables=self.no_variables))
return operators
def __init__(
self,
no_variables: int = 1,
coefficients: dict = None,
period_string: str = None,
order: int = None,
indicials: list = None,
):
self.no_variables = no_variables
# indicials[i] is the (rational) Frobenius exponent ρ_i of z_i: a stored
# z-index of a really represents z^(a + indicials[i]).
self.indicials = list(indicials) if indicials is not None else [0] * no_variables
# Use coefficients or period_string to initialize the period
if period_string is not None:
if coefficients is not None:
raise ValueError("Provide either coefficients or period_string, not both.")
self.coefficients = self._period_from_string(period_string)
self.period_string = period_string
logger.debug("Using string for initialisation.")
elif coefficients is None:
self.coefficients = {}
else:
self.coefficients = coefficients
self.period_string = self._period_to_string()
logger.debug("Using coefficients for initialisation.")
if order is None:
self.order = self._max_z_degree()
logger.debug("Order not provided, using maximum z-degree: %d", self.order)
else:
self.coefficients = self._truncate_coefficients(order)
self.order = order
self.period_string = self._period_to_string()
logger.debug("Truncated coefficients to order %d.", self.order)
logger.info("Initialised Period in %d variables at order %d.", self.no_variables, self.order)
class PeriodAnsatz(Period):
"""
A class for period Ansätze, characterised by number of variables and their z- and log-multi-degrees.
"""
def __init__(self, no_variables: int, z_degree: int, log_degree: int, indicials: list = None):
self.no_variables = no_variables
self.z_degree = z_degree
self.log_degree = log_degree
log_indices = _multi_indices(self.no_variables, log_degree)
z_indices = _multi_indices(self.no_variables, z_degree)
# Every (log_index, z_index) pair allowed by the degree bounds gets its own
# fresh unknown, to be solved for once a PFOperator is applied to the Ansatz.
coefficients = {
log_index: {
z_index: var("a_" + "_".join(str(x) for x in log_index + z_index))
for z_index in z_indices
}
for log_index in log_indices
}
super().__init__(no_variables=no_variables, coefficients=coefficients, order=z_degree, indicials=indicials)
self.expansion_coefficients = self.coefficients
logger.info(
"Initialised PeriodAnsatz with %d unknown coefficient(s).",
sum(len(z_dict) for z_dict in self.expansion_coefficients.values()),
)
sage/util.py
+15
View File
@@ -0,0 +1,15 @@
"""
Utility functions for the Calabi-Yau period geometry project.
"""
def _multi_indices(no_variables, total_degree: int) -> list:
# All no_variables-tuples of non-negative integers whose entries sum to at most total_degree.
def helper(remaining_vars, remaining_degree):
if remaining_vars == 0:
yield ()
return
for first in range(remaining_degree + 1):
for rest in helper(remaining_vars - 1, remaining_degree - first):
yield (first,) + rest
return list(helper(no_variables, total_degree))
tests/test_period_computation.py
+194
View File
@@ -0,0 +1,194 @@
import os
from sage.all import * # noqa: F401
load(os.path.join(os.path.dirname(__file__), "..", "sage", "period_computation.sage"))
def test_apply_operator_univariate():
# theta^2 - z applied to log(z) should give -z*log(z), since theta(log z) = 1
# and theta^2(log z) = theta(1) = 0.
period = Period(no_variables=1, coefficients={(1,): {(0,): 1}}, order=5)
op = PFOperator("theta0^2 - z0", no_variables=1)
result = period.apply_operator(op)
assert result.coefficients == {(1,): {(1,): -1}}
def test_apply_operator_mixes_variables():
# z0*theta1 applied to z0*log(z1): theta1 strips log(z1) down to a bare 1
# (leaving z0 untouched), then multiplying by z0 gives z0^2.
period = Period(no_variables=2, coefficients={(0, 1): {(1, 0): 1}}, order=5)
op = PFOperator("z0*theta1", no_variables=2)
result = period.apply_operator(op)
assert result.coefficients == {(0, 0): {(2, 0): 1}}
def test_apply_operator_truncates_to_order():
# Multiplying by z0^2 pushes some terms above the period's order, so they
# should be dropped rather than kept with a nonzero coefficient.
period = Period(no_variables=1, coefficients={(0,): {(0,): 1, (1,): 1, (2,): 1}}, order=2)
op = PFOperator("z0^2", no_variables=1)
result = period.apply_operator(op)
assert result.coefficients == {(0,): {(2,): 1}}
assert result.order == 2
def test_find_annihilating_operators_recovers_theta_squared():
# theta0^2 annihilates log(z0): theta0(log z0) = 1, theta0^2(log z0) = theta0(1) = 0.
# Among degree-(0, 2) Ansatze this should be the only solution, up to scaling.
period = Period(no_variables=1, coefficients={(1,): {(0,): 1}}, order=5)
ops = period.find_annihilating_operators(z_degree=0, theta_degree=2)
# A one-dimensional solution space means exactly one basis operator.
assert len(ops) == 1
assert period.apply_operator(ops[0]).coefficients == {}
assert str(ops[0].operator) == "theta0^2"
def test_find_annihilating_operators_recovers_geometric_series_operator():
# sum_{k=0}^{4} z0^k is annihilated (up to truncation order) by
# (1 - z0)*theta0 - z0, i.e. theta0 - z0*theta0 - z0.
period = Period(
no_variables=1,
coefficients={(0,): {(0,): 1, (1,): 1, (2,): 1, (3,): 1, (4,): 1}},
order=4,
)
ops = period.find_annihilating_operators(z_degree=1, theta_degree=1)
assert len(ops) == 1
assert period.apply_operator(ops[0]).coefficients == {}
def test_find_annihilating_operators_returns_empty_list_when_no_solution_exists():
# No degree-0 (constant) operator other than the zero operator can annihilate a
# nonzero constant period, so the linear system's only solution is trivial.
period = Period(no_variables=1, coefficients={(0,): {(0,): 1}}, order=0)
ops = period.find_annihilating_operators(z_degree=0, theta_degree=0)
assert ops == []
def test_simplify_factorises_theta_polynomial_per_z_monomial():
# theta0^4 - 5*z0*(5*theta0+1)*(5*theta0+2)*(5*theta0+3)*(5*theta0+4), expanded, is the
# quintic's Picard-Fuchs operator. simplify() should recover the factorised form: the
# z0^0 part (theta0^4) has no theta-factor to pull out, while the z0^1 part factorises
# into the four linear pieces.
expanded = "theta0^4 - 3125*z0*theta0^4 - 6250*z0*theta0^3 - 4375*z0*theta0^2 - 1250*z0*theta0 - 120*z0"
op = PFOperator(expanded, no_variables=1)
simplified = op.simplify()
# The underlying (expanded) operator is unchanged - only the display string differs.
assert simplified.operator == op.operator
assert simplified.operator_string == "-5*(5*theta0 + 4)*(5*theta0 + 3)*(5*theta0 + 2)*(5*theta0 + 1)*z0 + theta0^4"
def test_find_power_series_solution_recovers_quintic_period():
# The quintic's Picard-Fuchs operator has indicial equation theta0^4 = 0 at z0 = 0
# (a quadruple root at 0), so its holomorphic (non-logarithmic) power series solution
# is found at indicial 0. Up to normalisation this is the classic quintic period
# 1 + 120*z0 + 113400*z0^2 + ...
op = PFOperator(
"theta0^4 - 5*z0*(5*theta0 + 1)*(5*theta0 + 2)*(5*theta0 + 3)*(5*theta0 + 4)",
no_variables=1,
)
ideal = PFIdeal([op])
solutions = ideal.find_power_series_solution(indicials=[0], order=3)
assert len(solutions) == 1
assert solutions[0].coefficients == {(0,): {(0,): 1, (1,): 120, (2,): 113400, (3,): 168168000}}
def test_find_power_series_solution_handles_rational_indicial():
# theta0 - 1/2 kills z0^(1/2 + k) only when k = 0, since its eigenvalue is 1/2 + k;
# so at indicial 1/2 there is exactly one solution (a constant multiple of sqrt(z0)),
# while at indicial 0 no power series solution exists at all.
op = PFOperator("theta0 - 1/2", no_variables=1)
ideal = PFIdeal([op])
solutions = ideal.find_power_series_solution(indicials=[1/2], order=3)
assert len(solutions) == 1
assert solutions[0].coefficients == {(0,): {(0,): 1, (1,): 0, (2,): 0, (3,): 0}}
assert solutions[0].period_string == "sqrt(z0)"
assert ideal.find_power_series_solution(indicials=[0], order=3) == []
def test_ideal_finds_power_series_solution_and_recovers_first_operator():
# Picard--Fuchs ideal for P_{22211}[8]:
# M1 = theta2*(-2*theta1+2*theta2-1) + 2*(theta1-2*theta2-1)*(theta1-2*theta2)*z2
# M2 = theta1^2*(2*(theta1-2*theta2)*z2-theta2) - 16*(2*theta1+1)*(4*theta1+1)*(4*theta1+3)*z1*z2
M1 = PFOperator(
"theta1*(-2*theta0 + 2*theta1 - 1) + 2*(theta0 - 2*theta1 - 1)*(theta0 - 2*theta1)*z1",
no_variables=2,
)
M2 = PFOperator(
"theta0^2*(2*(theta0 - 2*theta1)*z1 - theta1) - 16*(2*theta0 + 1)*(4*theta0 + 1)*(4*theta0 + 3)*z0*z1",
no_variables=2,
)
ideal = PFIdeal([M1, M2])
solutions = ideal.find_power_series_solution(indicials=[0, QQ(1) / 2], order=6)
assert len(solutions) == 1
period = solutions[0]
assert period.apply_operator(M1).coefficients == {}
assert period.apply_operator(M2).coefficients == {}
# Normalisation convention: the free parameter at the leading (0, 0) coefficient is 1.
assert period.coefficients[(0, 0)][(0, 0)] == 1
assert period.coefficients[(0, 0)][(1, 1)] == -32
# M1 lives entirely within z_degree <= 1, theta_degree <= 2 (a single bare factor of
# z2, quadratic in theta), so this is the natural Ansatz level to look for it at.
recovered = period.find_annihilating_operators(z_degree=1, theta_degree=2)
assert len(recovered) == 1
# recovered[0] should be a scalar multiple of M1 - compare via the coefficient of the
# bare theta2 (theta1 in code) monomial, which is nonzero in M1.
ratio = (
recovered[0].operator.monomial_coefficient(M1.theta_gens[1])
/ M1.operator.monomial_coefficient(M1.theta_gens[1])
)
assert ratio != 0
assert recovered[0].operator == ratio * M1.operator
def test_single_operator_recovers_itself_from_its_holomorphic_period():
# Ensuring methods work for operator 4.2.1:
L = PFOperator(
"theta0^4 - 4*z0*(2*theta0 + 1)^2*(7*theta0^2 + 7*theta0 + 2) "
"- 128*z0^2*(2*theta0 + 1)^2*(2*theta0 + 3)^2",
no_variables=1,
)
ideal = PFIdeal([L])
solutions = ideal.find_power_series_solution(indicials=[0], order=15)
assert len(solutions) == 1
period = solutions[0]
assert period.apply_operator(L).coefficients == {}
assert period.coefficients[(0,)][(0,)] == 1
assert period.coefficients[(0,)][(1,)] == 8
assert period.coefficients[(0,)][(2,)] == 360
# L lives entirely within z_degree <= 2, theta_degree <= 4, its stated level.
recovered = period.find_annihilating_operators(z_degree=2, theta_degree=4)
assert len(recovered) == 1
# recovered[0] should be a scalar multiple of L - compare via the coefficient of the
# bare theta0^4 monomial, which is 1 in L.
ratio = recovered[0].operator.monomial_coefficient(L.theta_gens[0] ** 4) / L.operator.monomial_coefficient(
L.theta_gens[0] ** 4
)
assert ratio != 0
assert recovered[0].operator == ratio * L.operator