Skip to main content

This package implements algorithms relying on Lefschetz fibration theory to compute periods of algebraic varieties.

Project description

lefschetz-family

Description

This Sage package provides a means of efficiently computing periods of complex projective hypersurfaces and elliptic surfaces over $\mathbb P^1$ with certified rigorous precision bounds. It implements the methods described in https://arxiv.org/abs/2306.05263. Here is a runtime benchmark for various examples:

Variety (generic) Time (on 4 M1 cores)
Elliptic curve 8 seconds
Quartic curve 4 minutes
Cubic surface 4 minutes 20
Quartic surface 1 hour (for holomorphic periods)
Cubic threefold 1 hour (for holomorphic periods)

How to install

Follow the installation procedure of the numperiods package, then run

sage -pip install lefschetz-family

or

sage -pip install --user lefschetz-family

Requirements

Sage 9.0 and above is recommended. Furthermore, this package has the following dependencies:

Usage

Hypersurface

The first step is to define the polynomial $P$ defining the projective hypersurface $X=V(P)$. For instance, the following gives the Fermat elliptic curve:

R.<X,Y,Z> = PolynomialRing(QQ)
P = X**3+Y**3+Z**3

Then the following creates an object representing the hypersurface:

from period import LefschetzFamily
X = Hypersurface(P)

The period matrix of $X$ is the simply given by:

X.period_matrix

See the computation of the periods of the Fermat quartic surface for a usage example.

See the computation of the periods of the Fermat quartic surface for a usage example.

Options

The object Hypersurface can be called with several options:

  • method ("voronoi" by default/"delaunay"/"delaunay_dual"): the method used for computing a basis of homotopy. voronoi uses integration along paths in the voronoi graph of the critical points; delaunay uses integration along paths along the delaunay triangulation of the critical points; delaunay_dual paths are along the segments connecting the barycenter of a triangle of the Delaunay triangulation to the middle of one of its edges. In practice, delaunay is more efficient for low dimension and low order varieties (such as degree 3 curves and surfaces, and degree 4 curves). This gain in performance is however hindered in higher dimensions because of the algebraic complexity of the critical points (which are defined as roots of high order polynomials, with very large integer coefficients). "delaunay" method is not working for now
  • nbits (positive integer, 400 by default): the number of bits of precision used as input for the computations. If a computation fails to recover the integral monodromy matrices, you should try to increase this precision. The output precision seems to be roughly linear with respect to the input precision.
  • debug (boolean, False by default): whether coherence checks should be done earlier rather than late. We recommend setting to true only if the computation failed in normal mode.

Properties

The object Hypersurface has several properties. Fibration related properties, in positive dimension:

  • fibration: a list of independant hyperplanes defining the iterative pencils. The first two element of the list generate the pencil used for the fibration.
  • critical_values: the list critical values of that map.
  • basepoint: the basepoint of the fibration (i.e. a non critical value).
  • fiber: the fiber above the basepoint as a Hypersurface object.
  • fundamental_group: the class computing representants of the fundamental group of $\mathbb P^1$ punctured at the critical values.
  • paths: the list of simple loops around each point of critical_values. When this is called, the ordering of critical_values changes so that the composition of these loops is the loop around infinity.
  • family: the one parameter family corresponding to the fibration.

Homology related properties:

  • monodromy_matrices: the matrices of the monodromy action of paths on $H_{n-1}(X_b)$.
  • vanishing_cycles: the vanshing cycles at each point of critical_values along paths.
  • thimbles: the thimbles of $H_n(Y,Y_b)$. They are represented by a starting cycle in $H_n(Y_b)$ and a loop in $\mathbb C$ avoiding critical_values and pointed at basepoint.
  • kernel_boundary: linear combinations of thimbles with empty boundary.
  • extensions: integer linear combinations of thimbles with vanishing boundary.
  • infinity_loops: extensions around the loop at infinity.
  • homology_modification: a basis of $H_n(Y)$.
  • intersection_product_modification: the intersection product of $H_n(Y)$.
  • fibre_class: the class of the fibre in $H_n(Y)$.
  • section: the class of a section in $H_n(Y)$.
  • thimble_extensions: couples (t, T) such that T is the homology class in $H_n(Y)$ representing the extension of a thimble $\Delta \in H_{n-1}(X_b, X_{bb'})$ over all of $\mathbb P^1$, with $\delta\Delta =$t. Futhermore, the ts define a basis of the boundary map $\delta$.
  • invariant: the intersection of section with the fibre above the basepoint, as a cycle in $H_{n-2}({X_b}_{b'})$.
  • exceptional_divisors: the exceptional cycles coming from the modification $Y\to X$, given in the basis homology_modification.
  • homology: a basis of $H_n(X)$, given as its embedding in $H_2(Y)$.
  • intersection_product: the intersection product of $H_n(X)$.
  • lift: a map taking a linear combination of thimbles with zero boundary (i.e. an element of $\ker\left(\delta:H_n(Y, Y_b)\to H_{n-1}(Y_b)\right)$) and returning the homology class of its lift in $H_2(Y)$, in the basis homology_modification.
  • lift_modification: a map taking an element of $H_n(Y)$ given by its coordinates in homology_modification, and returning its homology class in $H_n(X)$ in the basis homology.

Cohomology related properties:

  • cohomology: a basis of $PH^n(X)$, represented by the numerators of the rational fractions.
  • holomorphic_forms: the indices of the forms in cohomology that form a basis of holomorphic forms.
  • picard_fuchs_equation(i): the picard fuchs equation of the parametrization of i-th element of cohomology by the fibration

Period related properties

  • period_matrix: the period matrix of $X$ in the aforementioned bases homology and cohomology
  • period__matrix_modification: the period matrix of the modification $Y$ in the aforementioned bases homology_modification and cohomology
  • holomorphic_periods: the periods of holomorphic_forms in the basis homology.
  • holomorphic_periods: the periods of the pushforwards of holomorphic_forms in the basis homology_modification.

Miscellaneous properties:

  • P: the defining equation of $X$.
  • dim: the dimension of $X$.
  • degree: the degree of $X$.
  • ctx: the options of $X$, see related section above.

The computation of the exceptional divisors can be costly, and is not always necessary. For example, the Picard rank of a quartic surface can be recovered with holomorphic_periods_modification.

EllipticSurface

The defining equation for the elliptic surface should be given as a univariate polynomial over a trivariate polynomial ring. The coefficients should be homogeneous of degree $3$.

R.<X,Y,Z> = PolynomialRing(QQ)
S.<t> = PolynomialRing(R)
P = X^2*Y+Y^2*Z+Z^2*X+t*X*Y*Z 

Then the following creates an object representing the hypersurface:

from period import LefschetzFamily
X = EllipticSurface(P)

Options

The options are the same as those for Hypersurface (see above).

Properties

The object ElliptcSurface has several properties. Fibration related properties, in positive dimension:

  • fibration: the two linear maps defining the map $X\dashrightarrow \mathbb P^1$.
  • critical_values: the list critical values of that map.
  • basepoint: the basepoint of the fibration (i.e. a non critical value).
  • fiber: the fiber above the basepoint as a LefschetzFamily object.
  • paths: the list of simple loops around each point of critical_points. When this is called, the ordering of critical_points changes so that the composition of these loops is the loop around infinity.
  • family: the one parameter family corresponding to the fibration.

Homology related properties:

  • extensions: the extensions of the fibration.
  • extensions_morsification: the extensions of the morsification of the fibration.
  • homology: the homology of $X$.
  • singular_components: a list of lists of combinations of thimbles of the morsification, such that the elements of singular_components[i] form a basis of the singular components of the fibre above critical_values[i]. To get their coordinates in the basis homology, use X.lift(X.singular_components[i][j]).
  • fibre_class: the class of the fibre in homology.
  • section: the class of the zero section in homology.
  • intersection_product: the intersection matrix of the surface in the basis homology.
  • morsify: a map taking a combination of extensions and returning its coordinates on the basis of thimbles of the morsification.
  • lift: a map taking a combination of thimbles of the morsification with empty boundary and returning its class in homology.

Cohomology related properties:

  • holomorphic_forms: a basis of rational functions $f(t)$ such that $f(t) \operatorname{Res}\frac{\Omega_2}{P_t}\wedge\mathrm dt$ is a holomorphic form of $S$.
  • picard_fuchs_equations: the list of the Picard-Fuchs equations of the holomorphic forms mentionned previously.

Period related properties:

  • period_matrix: the holomorphic periods of $X$ in the bases self.homology and self.holomorphic_forms.
  • effective_periods: the holomorphic periods $X$ in the bases self.effective_lattice and self.holomorphic_forms

Lattices. Unless stated otherwise, lattices are given the coordinates of a basis in the basis homology.

  • effective_lattice: The lattice of effective cycles of $X$, consisting of the concatenation of extensions, singular_components, fibre_class and section.
  • neron_severi: the Néron-Severi group of $X$.
  • trivial: the trivial lattice.
  • essential_lattice: the essential lattice.
  • mordell_weil: the Mordell-Weil group of $X$, described as the quotient module neron_severi/trivial.
  • mordell_weil_lattice: the intersection matrix of the Mordell-Weil lattice of $X$.

Miscellaneous properties:

  • dim: the dimension of $X$.
  • ctx: the options of $X$, see related section above.

Contact

For any question, bug or remark, please contact eric.pichon@polytechnique.edu.

Roadmap

Near future milestones:

  • Encapsulate integration step in its own class
  • Certified computation of the exceptional divisors
  • Making Delaunay triangulation functional again
  • Saving time on differential operator by precomputing cache before parallelization

Middle term goals include:

  • Having own implementation of 2D voronoi graphs/Delaunay triangulation

Long term goals include:

  • Computing periods of elliptic fibrations.
  • Tackling higher dimensional varieties (most notably cubics in $\mathbb P^5$).
  • Computing periods of singular varieties.
  • Computing periods of complete intersections.
  • Computing periods of weighted projective hypersurfaces, notably double covers of $\mathbb P^2$ ramified along a cubic.

Other directions include:

  • Computation of homology through braid groups instead of monodromy of differential operators.

Project status

This project is actively being developped.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

lefschetz_family-0.1.6.tar.gz (36.9 kB view details)

Uploaded Source

Built Distribution

lefschetz_family-0.1.6-py3-none-any.whl (44.4 kB view details)

Uploaded Python 3

File details

Details for the file lefschetz_family-0.1.6.tar.gz.

File metadata

  • Download URL: lefschetz_family-0.1.6.tar.gz
  • Upload date:
  • Size: 36.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.11.4

File hashes

Hashes for lefschetz_family-0.1.6.tar.gz
Algorithm Hash digest
SHA256 2f518c5b66891e74b6cc6c31196db612bc21982a64ec8b0b0a9a151eea091ba3
MD5 4da08966ba847e4ba92f00b7dfd64b31
BLAKE2b-256 fd8b20f96a4f1690c196d49860818664cdd01e7916e93f55db1371bf13df40b4

See more details on using hashes here.

File details

Details for the file lefschetz_family-0.1.6-py3-none-any.whl.

File metadata

File hashes

Hashes for lefschetz_family-0.1.6-py3-none-any.whl
Algorithm Hash digest
SHA256 90e83c4455c48d7fc7eed0833213015171638324e57e1699b5d93eeb0973f410
MD5 0941b9786569e8c8b2600ea24ba41bbe
BLAKE2b-256 0e4affb1d5caf259cb758aa0b48a849b6eb9b389a0bf1543bac346c014200bc4

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page