Sage 9.2 Release Tour

released October 24, 2020

Contents

  1. Sage 9.2 Release Tour
    1. Python 3 transition completed
      1. Support for Python 2 removed
      2. Support for Python 3.6, 3.8, and 3.9 added
      3. For developers: Using Python 3.6+ features in sagelib
      4. More details
    2. Package upgrades
      1. matplotlib
      2. rpy2 and R
      3. sphinx
      4. SymPy
      5. IPython, Jupyter notebook, JupyterLab
      6. Normaliz
      7. SageTeX
      8. sws2rst + usage example
      9. Other package updates
      10. For developers: Upgrading packages
      11. For packagers: Changes to packages
    3. Graphics
      1. New features
      2. Implementation improvements
      3. For developers
    4. Linear and multilinear algebra
      1. One free module constructor to rule them all
      2. Connecting FiniteRankFreeModule and free modules with distinguished basis
      3. Eigenvalues and eigenvectors
      4. Other improvements
    5. Polyhedral geometry
      1. New features
      2. Implementation improvements
    6. Combinatorics
      1. Reduction from Dancing links to SAT or MILP
      2. Polyomino tilings
      3. Fully commutative elements of Coxeter groups
      4. BIBDs with lambda>1
      5. Finite generalized polygons
    7. Graph theory
      1. Distance regular graphs generators
      2. Diameter, radius, eccentricities
      3. More constructions
      4. More iterators
      5. Implementation improvements
    8. Commutative algebra
      1. Laurent polynomials
      2. Motivic multiple zetas
      3. Power series
      4. Ring homomorphisms
    9. Manifolds
      1. diff function for exterior derivatives
      2. Dot and cross products of vector fields along a curve
      3. Orientability of manifolds and vector bundles
      4. Euclidean spaces as metric spaces
      5. Bundle connections
      6. Internal code improvements and bug fixes
    10. Algebra
      1. Lie Conformal Algebras
      2. Differential Weyl algebra
    11. Improved Unicode support
      1. Unicode identifiers
      2. Unicode characters allowed in tensor index notation
      3. Unicode art
    12. Configuration and build changes
      1. Initial configuration with ./configure required
      2. Support for gcc/gfortran 10.x added
      3. Selecting a system Python to use for Sage's venv
      4. System package information for more distributions
      5. System package information for optional packages at runtime
      6. For developers: Changes to the build system of sagelib
    13. New development tools
      1. Testing and linting with tox
      2. Reusable wheels for the Python packages built by the Sage distribution
    14. Cleaning
    15. Availability of Sage 9.2 and installation help
      1. Binaries
      2. Availability in distributions
      3. Installation FAQ
    16. More details

Python 3 transition completed

SageMath 9.0 was the first version of Sage running on Python 3 by default. SageMath 9.1 continued to support Python 2.

Support for Python 2 removed

Sage 9.2 has removed support for Python 2. The Sage library now makes use of Python language and library features that are only available in Python 3.6 or newer; and large amounts of compatibility code have been removed.

However, note that this is unrelated to the minimal requirements for a source installation of the Sage distribution: Sage 9.2 is still able to build on a system that only provides Python 2.x or Python 3.5 or older. In this case, the SageMath distribution builds its own copy of Python 3.

Support for Python 3.6, 3.8, and 3.9 added

Sage 9.2 has added support for Python 3.8 in #27754 and Python 3.9 in #30184.

Sage 9.2 has also added support for Python 3.6. This allows Sage to use the system Python on some older Linux distributions that are still in widespread use in scientific computing, including centos-8 and fedora-{26,27,28} (although Python 3.7.x packages are also available for these). See #29033 for more details.

Hence, Sage 9.2 conforms to (and exceeds) NumPy Enhancement Proposal 29 regarding Python version support policies.

If no suitable system Python, versions 3.6.x, 3.7.x, 3.8.x, or 3.9.x is found, Sage installs its own copy of Python 3 from source. The version of Python shipped with the Sage distribution has been upgraded from 3.7.3 to 3.8.5.

For developers: Using Python 3.6+ features in sagelib

Meta-ticket #29756 provides a starting point for a discussion of new features of the Python language and standard library to bring them to systematic use in sagelib. All features provided by Python 3.6 can be used immediately; features introduced in Python 3.7 or later will require backporting or a decision to drop the goal of supporting Python 3.6.

More details

Package upgrades

The removal of support for Python 2 has enabled major package upgrades.

Major user-visible package upgrades below...

matplotlib

Dropping Python 2 support allowed us to make a major jump from matplotlib 2.2.5 to 3.3.1. See matplotlib's release notes for 3.0, 3.1, 3.2,3.3. In addition to improved output, this update will likely enable Sage developers to implement new features for plotting and graphics.

rpy2 and R

The rpy2 Python package is the foundation for SageMath's interface to R. Dropping Python 2 support allowed us to make the major upgrade from 2.8.2 to 3.3.5 in #29441; see the release notes for details.

We only did a minor upgrade of R itself in the Sage distribution, to 3.6.3, the latest in the 3.6.x series. Of course, if R 4.0.x is installed in the system, Sage will use it instead of building its own copy.

The SageMath developers are eager to learn from users how they use the SageMath-R interface, and what needs to be added to it to become more powerful. Let us know at sage-devel.

sphinx

Sage uses Sphinx to build its documentation. Sage 9.2 has updated Sphinx from 1.8.5 to 3.1.2; see Sphinx release notes for more information.

SymPy

SymPy has been updated from 1.5 to 1.6.2 in #29730, #30425. See the Release notes.

IPython, Jupyter notebook, JupyterLab

Dropping support for Python 2 allowed us to upgrade IPython from 5.8.0 to 7.13.0 in #28197. See the release notes for the 6.x and 7.x series.

We have also upgraded the Jupyter notebook from 5.7.6 to 6.1.1 in #26919; see the notebook changelog for more information. Besides, the pdf export of Jupyter notebooks has been fixed, so that LaTeX-typeset outputs are now rendered in the pdf file (#23330).

JupyterLab is now fully supported as an optional, alternative interface #30246, including interacts. To use it, install it first, using the command sage -i jupyterlab_widgets. Then you can start it using ./sage -n jupyterlab.

Normaliz

The optional package Normaliz, a tool for computations in affine monoids, vector configurations, lattice polytopes, rational cones, and algebraic polyhedra has been upgraded from 3.7.2 to 3.8.8, and PyNormaliz to version 2.12.

The upgrade adds support for incremental ("dynamic") computations, the computation of automorphism groups and refined triangulations of cones and polyhedra, and limited support for semiopen cones and polyhedra.

To install Normaliz and PyNormaliz, use sage -i pynormaliz.

SageTeX

Updated to version 3.5, improving Python 3 compatibility, also updated to version 3.5 on CTAN.

sws2rst + usage example

In ticket #28838, the command sage -sws2rst was resurrected via a new pip-installable package sage-sws2rst. It can be installed in Sage 9.2 using sage -i sage_sws2rst.

Below is an example of usage. First we download a sage worksheet (.sws) prepared for Sage Days 20 at CIRM (Marseille, 2010):

$ wget http://slabbe.org/Sage/2010-perpignan/CIRM_Tutorial_1.sws
$ ls
CIRM_Tutorial_1.sws

We translate the sws worksheet into a ReStructuredText syntax file (.rst) using sage -sws2rst. This creates also a directory of images:

$ sage -sws2rst CIRM_Tutorial_1.sws
Processing CIRM_Tutorial_1.sws
File at CIRM_Tutorial_1.rst
Image directory at CIRM_Tutorial_1_media
$ ls
CIRM_Tutorial_1_media  CIRM_Tutorial_1.rst  CIRM_Tutorial_1.sws

Then, we can check that it works properly by looking at the generated rst file. Alternatively, we can translate it to a basic html file using rst2html:

$ rst2html.py CIRM_Tutorial_1.rst CIRM_Tutorial_1.html
CIRM_Tutorial_1.rst:176: (WARNING/2) Explicit markup ends without a blank line; unexpected unindent.
CIRM_Tutorial_1.rst:334: (WARNING/2) Inline strong start-string without end-string.

As seen above, there are few warnings sometimes because the translation made by sws2rst is not 100% perfect, but most of it is okay:

$ firefox CIRM_Tutorial_1.html

Moreover, one can use the sage -rst2ipynb script to translate the rst file obtained above to a Jupyter notebook:

$ sage -rst2ipynb CIRM_Tutorial_1.rst CIRM_Tutorial_1.ipynb

One can check the result:

$ sage -n jupyter

Note that to translate old .sws files to .ipynb, you may also use the export notebook:

$ sage -n -h
[...]
* List available legacy Sage notebooks:

      sage --notebook=export --list

* Export a legacy Sage notebook as a Jupyter notebook:

      sage --notebook=export --ipynb=Output.ipynb admin:10

Other package updates

From the changelog of the release of 9.2, the following list of upgrades made in 9.2 was extracted:

#3360: sympow 2.023.6 (for GCC 10 support)
#22191: ECL 20.4.24
#26891: Nauty 2.7
#26919: Jupyter notebook to latest (6.1.1) and its dependencies to latest
#27309: FriCAS 1.3.6
#27754: Python 3.8.5
#27880: Kenzo and its interface
#27952: Normaliz 3.8.8, PyNormaliz 2.12, add script package libnauty
#28197: ipython 7
#28856: sphinx 3
#28959: zn_poly v0.9.2
#29061: symmetrica-3.0.1
#29240: pexpect 4.8
#29313: pari 2.11.4
#29441: rpy2 package 2.8.2 -> 3.3.5, Update R 3.6.3, add new dependencies
#29480: Cython 0.29.17
#29483: gsl 2.6
#29547: matplotlib 3
#29552: giac 1.5.0-87
#29658: BRiAl 1.2.8
#29730: sympy 1.6
#29766: NumPy 1.19.1, scipy 1.5.2, networkx 2.4, add pybind11 package
#29803: setuptools, setuptools_scm, pip (2020-06), add package wheel; remove zope_interface
#29809: r117
#29826: e-antic 0.1.7
#29859: palp 2.11 (for GCC 10 compatibility)
#29861: Cython 0.29.21
#30001: sphinx 3.1
#30063: Maxima 5.44.0
#30150: cmake 3.18.2
#30176: matplotlib 3.3
#30185: pillow 7.2.0
#30262: e-antic 0.1.8
#30317: pip 20.2.2, setuptools 49.6.0
#30338: libhomfly to the latest version
#30342: sagetex 3.5
#30358: matplotlib 3.3.1, certifi 2020.6.20
#30412: gf2x 1.3
#30425: misc pip Update 2020-08: SymPy, pip, six
#30583: gmpy2 2.1.0.b5
#30603: readline 8.0

For developers: Upgrading packages

Upgrading Python packages in the Sage distribution from PyPI has again become easier, thanks to #20104. You can now do:

$ sage --package update-latest matplotlib
Updating matplotlib: 3.3.0 -> 3.3.1
Downloading tarball to ...matplotlib-3.3.1.tar.bz2
[...............................................................]

When you do this, please remember to check that the checksums.ini file has an upstream_url in the format upstream_url=https://pypi.io/packages/source/m/matplotlib/matplotlib-VERSION.tar.gz. (This is not needed for updated-latest to work, but helps with automated tests of the upgrade ticket -- see Sage 9.1 release tour on this topic.)

For packagers: Changes to packages

The packages giacpy_sage and sage_brial have been merged into sagelib as sage.libs.giac and sage.rings.polynomial.pbori.

The directory build/pkgs/sage_sws2rst/src contains a new pip-installable package, providing the script sage-sws2rst.

The Sage library is now built out of the directory build/pkgs/sagelib/src/. A pip-installable source distribution (sdist) can be built using the script build/pkgs/sagelib/spkg-src (#29411, #29950).

The scripts in src/bin/ are now installed by sagelib's setup.py (#21559). Also several scripts have been moved to build/bin/, and some obsolete scripts have been removed (#29825, #27171).

Many build-related functions of the main Sage script, src/bin/sage (installed as sage), have been moved to a script build/bin/sage-site (not installed) in #29111. It is hoped that downstream distribution packaging is able to use this cleaned up script instead of replacing it with an ad-hoc distribution-specific script -- so that users can rely on a consistent interface. Contributions of further clean ups and refactoring of the script are welcome.

Graphics

New features

Implementation improvements

For developers

Linear and multilinear algebra

One free module constructor to rule them all

Sage has several specialized implementation classes for free modules and vector spaces. The factory functions FreeModule and VectorSpace select the appropriate class depending on the base ring and other parameters:

   1 sage: FreeModule(ZZ, 10)
   2 Ambient free module of rank 10 
   3 over the principal ideal domain Integer Ring
   4 sage: FreeModule(FiniteField(5), 10)
   5 Vector space of dimension 10 over Finite Field of size 5
   6 sage: QQ^10 is VectorSpace(QQ, 10)
   7 True

The free modules (vector spaces) created here have a distinguished standard basis indexed by range(rank).

In Sage 9.2, these factory functions have been extended in #30194 so that they cover two more cases:

1. If a sequence or family of indices is passed instead of the rank (dimension), then a CombinatorialFreeModule is created instead. These modules underly SageMath's facilities for algebraic combinatorics.

   1 sage: U = FreeModule(AA, ['x', 'y', 'z']); U
   2 Free module generated by {'x', 'y', 'z'} over Algebraic Real Field
   3 sage: V = VectorSpace(QQ, ZZ); V
   4 sage: V.basis()                                                                                          
   5 Lazy family 
   6 (Term map from Integer Ring 
   7  to Free module generated by Integer Ring over Rational Field(i))
   8 _{i in Integer Ring}
   9 sage: QQ^SymmetricGroup(4)                                                                                                       
  10 Free module generated by 
  11 Symmetric group of order 4! as a permutation group over Rational Field

2. If the factory function is invoked with the parameter with_basis=None, then a free module of the given rank without distinguished basis is created.

   1 sage: W = FreeModule(AA, 3, with_basis=None); W
   2 3-dimensional vector space over the Algebraic Real Field
   3 sage: W.category()
   4 Category of finite dimensional vector spaces over Algebraic Real Field
   5 sage: W.tensor_module(2, 2)                                                                                                      
   6 Free module of type-(2,2) tensors 
   7 on the 3-dimensional vector space over the Algebraic Real Field

It is represented by an instance of the class FiniteRankFreeModule from sage.tensor.modules. These modules are the foundation for the multilinear algebra developed by the SageManifolds project.

Connecting FiniteRankFreeModule and free modules with distinguished basis

Given a basis of a FiniteRankFreeModule, the new method isomorphism_with_fixed_basis (#30094) constructs an isomorphism from the FiniteRankFreeModule to a free module in the category ModulesWithBasis. By default, it uses a CombinatorialFreeModule:

   1 sage: V = FiniteRankFreeModule(QQ, 3, start_index=1); V
   2 3-dimensional vector space over the Rational Field
   3 sage: basis = e = V.basis("e"); basis
   4 Basis (e_1,e_2,e_3) on the 3-dimensional vector space over the
   5 Rational Field
   6 sage: phi_e = V.isomorphism_with_fixed_basis(basis); phi_e
   7 Generic morphism:
   8   From: 3-dimensional vector space over the Rational Field
   9   To:   Free module generated by {1, 2, 3} over Rational Field
  10 sage: phi_e(e[1] + 2 * e[2])
  11 e[1] + 2*e[2]

Eigenvalues and eigenvectors

Other improvements

Sage 9.2 has also merged a number of improvements to sage.tensor.modules: #30094, #30169, #30179, #30181, #30194, #30250, #30251, #30254, #30255, #30287

Polyhedral geometry

New features

It is now possible to choose which backend to use to compute regions of hyperplane arrangements 29506:

   1 sage: R.<sqrt5> = QuadraticField(5)
   2 sage: H = HyperplaneArrangements(R, names='xyz')
   3 sage: x,y,z = H.gens()
   4 sage: A = H(sqrt5*x+2*y+3*z, backend='normaliz')
   5 sage: A.backend()
   6 'normaliz'
   7 sage: A.regions()[0].backend()  # optional - pynormaliz
   8 'normaliz'

It is now possible to compute the slack matrix of a polyhedron 29838:

   1 sage: P = polytopes.cube(intervals='zero_one')
   2 sage: P.slack_matrix()
   3 0 1 1 1 0 0
   4 0 0 1 1 0 1
   5 0 0 0 1 1 1
   6 0 1 0 1 1 0
   7 1 1 0 0 1 0
   8 1 1 1 0 0 0
   9 1 0 1 0 0 1
  10 1 0 0 0 1 1

It is now possible to apply an affine transformation on a polyhedron 30327:

   1 sage: M = random_matrix(QQ,3,3) 
   2 sage: v = vector(QQ,(1,2,3)) 
   3 sage: F = AffineGroup(3, QQ) 
   4 sage: f = F(M, v); f                                                                    
   5       [  0   0  -2]     [1]
   6 x |-> [  0   1   0] x + [2]
   7       [ -1  -1 1/2]     [3]
   8 sage: cube = polytopes.cube() 
   9 sage: f * cube                                                            
  10 A 3-dimensional polyhedron in QQ^3 defined as the convex hull of 8 vertices
  11 sage: f(cube)                     # also works                                                        
  12 A 3-dimensional polyhedron in QQ^3 defined as the convex hull of 8 vertices

Implementation improvements

There are also some bug fixes and other improvements. For more details see the release notes for optimization and polyhedral geometry software interactions in Sage.

Combinatorics

It is now possible to solve an instance of an exact cover problem using a reduction from a dancing links instance to SAT 29338 or MILP 29955:

   1 sage: from sage.combinat.matrices.dancing_links import dlx_solver
   2 sage: rows = [[0,1,2], [3,4,5], [0,1], [2,3,4,5], [0], [1,2,3,4,5]]
   3 sage: d = dlx_solver(rows)
   4 sage: d.one_solution()
   5 [1, 0]
   6 sage: d.one_solution_using_sat_solver('cryptominisat')
   7 [2, 3]
   8 sage: d.one_solution_using_sat_solver('glucose')
   9 [2, 3]
  10 sage: d.one_solution_using_sat_solver('glucose-syrup')
  11 [2, 3]
  12 sage: d.one_solution_using_sat_solver('picosat')
  13 [4, 5]
  14 sage: d.one_solution_using_milp_solver()
  15 [0, 1]
  16 sage: d.one_solution_using_milp_solver('Gurobi')
  17 [0, 1]

Polyomino tilings

It is now possible to find a surrounding of a polyomino with copies of itself, see 29160. This is based on the dancing links solver in Sage. This is motivated by the Heesch's problem. An example is below:

sage: from sage.combinat.tiling import Polyomino
sage: H = Polyomino([(-1, 1), (-1, 4), (-1, 7), (0, 0), (0, 1), (0, 2),
....: (0, 3), (0, 4), (0, 5), (0, 6), (0, 7), (0, 8), (1, 1), (1, 2),
....: (1, 3), (1, 4), (1, 5), (1, 6), (1, 7), (1, 8), (2, 0), (2, 2),
....: (2, 3), (2, 5), (2, 6), (2, 8)])
sage: H.show2d()

H.png

sage: %time solution = H.self_surrounding(10, ncpus=8)
CPU times: user 1.69 s, sys: 1.08 s, total: 2.77 s
Wall time: 3.85 s
sage: G = sum([p.show2d() for p in solution], Graphics())
sage: G

G.png

Fully commutative elements of Coxeter groups

It is now possible by 30243 to enumerate and work with the fully commutative elements of a Coxeter group. Methods to compute *star operations* and plot the *heaps* of such elements are also included.

   1 sage: A3 = CoxeterGroup(['A', 3])
   2 sage: FCA3 = A3.fully_commutative_elements()
   3 sage: FCA3.category()
   4 Category of finite enumerated sets
   5 sage: FCA3.list()
   6 [[],
   7  [1],
   8  [2],
   9  ...
  10  [1, 3, 2],
  11  [1, 2, 3],
  12  [2, 1, 3, 2]]
  13 sage: B8 = CoxeterGroup(['B', 8])
  14 sage: FCB8 = B8.fully_commutative_elements()
  15 sage: len(FCB8)    # long time (7 seconds)
  16 14299
  17 sage: B6 = CoxeterGroup(['B', 6])
  18 sage: FCB6 = B6.fully_commutative_elements()
  19 sage: w = FCB6([1, 6, 2, 5, 4, 6, 5])
  20 sage: w.coset_decomposition({5, 6})
  21 ([6, 5, 6], [1, 2, 4, 5])
  22 sage: w.star_operation({5,6}, 'lower')
  23 [1, 5, 2, 4, 6, 5]
  24 sage: FCB6([3, 2, 4, 3, 1]).plot_heap()

heap.png

BIBDs with lambda>1

Sage can now construct a number of balanced incomplete block designs (BIBDs) with lambda>1, in particular, all the known biplanes (i.e. symmetric BIBDs with lambda=2).

Finite generalized polygons

Sage can now construct generalized quadrangles, hexagons, and octagons, and generalized quadrangles with a spread.

Graph theory

Distance regular graphs generators

A small database of constructions for distance regular graphs is now available. 15 sporadic distance regular graphs and 8 infinite families can now be constructed. Now Sage is capable of constructing all known families of distance-regular graphs with unbounded diameter.

Some code examples:

   1 sage: graphs.vanLintSchrijverGraph().is_distance_regular(parameters=True)
   2 ([6, 5, 5, 4, None], [None, 1, 1, 2, 6])
   3 sage: graphs.DoubleOddGraph(5).is_distance_regular(parameters=True)
   4 ([6, 5, 5, 4, 4, 3, 3, 2, 2, 1, 1, None],
   5  [None, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6])

Diameter, radius, eccentricities

State-of-the-art algorithms for computing the diameter, the radius and the eccentricities of (directed) (weighted) graphs are now available 29657.

   1 sage: G = graphs.RandomBarabasiAlbert(10000, 2)
   2 sage: %time G.diameter(algorithm='DHV')  # now default for undirected unweighted graphs
   3 CPU times: user 74.4 ms, sys: 2.81 ms, total: 77.2 ms
   4 Wall time: 75.8 ms
   5 9
   6 sage: %time G.diameter(algorithm='BFS')
   7 CPU times: user 573 ms, sys: 4.04 ms, total: 577 ms
   8 Wall time: 576 ms
   9 9

More constructions

linear codes got a function to compute its coset graph; undirected graphs got a method to quickly compute their antipodal quotients.

More iterators

Some methods have been turned to iterators to avoid returning long lists 23002, 30470.

   1 sage: G = graphs.Grid2dGraph(3, 3)
   2 sage: cpt = 0
   3 sage: for _ in G.spanning_trees():
   4 ....:     cpt += 1
   5 sage: cpt
   6 192
   7 sage: G.spanning_trees_count()
   8 192
   9 sage: S = G.spanning_trees()
  10 sage: next(S)
  11 Graph on 9 vertices
  12 sage: next(S)
  13 Graph on 9 vertices

   1 sage: g = graphs.PathGraph(5)
   2 sage: bridges = g.bridges()
   3 sage: next(bridges)
   4 (3, 4, None)
   5 sage: next(bridges)
   6 (2, 3, None)
   7 sage: next(bridges)
   8 (1, 2, None)
   9 sage: next(bridges)
  10 (0, 1, None)
  11 sage: next(bridges)
  12 ---------------------------------------------------------------------------
  13 StopIteration                             Traceback (most recent call last)
  14 <ipython-input-38-cf858e1a0a30> in <module>
  15 ----> 1 next(bridges)
  16 
  17 StopIteration: 

Implementation improvements

Commutative algebra

Laurent polynomials

Rings of Laurent polynomials now support ideal creation and manipulation 29512:

sage: L.<x,y,z> = LaurentPolynomialRing(QQ, 3)                                                
sage: I = L.ideal([(x+y+z)^3+x*y, x^2+y^2+z^2])                                               
sage: I.groebner_basis()                                                                      
(y^4 + 4*x*y*z^2 + y^2*z^2 + 2*x*z^3 + 2*y*z^3 - z^4 + 3/2*x*y*z + 1/4*x*z^2 + 1/4*y*z^2 - 1/4*z^3 + 1/8*x*y,
 x*y^2 - y^3 + 3*x*y*z + x*z^2 - z^3 + 1/2*x*y,
 x^2 + y^2 + z^2)
sage: (x^3+y^3+z^3) in I                                                                      
False
sage: x + x^-1*y^2 + x^-1*z^2 in I                                                            
True

Motivic multiple zetas

The ring of motivic multiple zeta values has been implemented, using algorithms of Francis Brown. It allows to compute at least up to weight 12 22713.

sage: Multizeta(1,2)**2                                                         
12*ζ(1,1,1,3) + 6*ζ(1,1,2,2) + 2*ζ(1,2,1,2)
sage: Multizeta(1,2)==Multizeta(3)                                              
True
sage: Multizeta(2,3,4).n(100)                                                   
0.0029375850405618494701189454256

The numerical evaluation is based on PARI implementation.

Power series

There is a new method to compute the coefficients in the Jacobi continued fraction expansion of a power series 29789.

sage: t = QQ[['t']].0                                                             
sage: f = sum(factorial(n)*t**n for n in range(20)).O(20)                         
sage: f.jacobi_continued_fraction()                                             
((-1, -1),
 (-3, -4),
 (-5, -9),
 (-7, -16),
 (-9, -25),
 (-11, -36),
 (-13, -49),
 (-15, -64),
 (-17, -81))

Ring homomorphisms

For many polynomial ring homomorphisms, the methods inverse, is_invertible, is_injective, is_surjective, kernel and inverse_image have been implemented. This covers not only polynomial rings, but also quotient rings, number fields and Galois fields. #9792 #29723

sage: R.<x,y,z> = QQ[]
sage: sigma = R.hom([x - 2*y*(z*x+y^2) - z*(z*x+y^2)^2, y + z*(z*x+y^2),
z], R)
sage: tau = sigma.inverse(); tau
Ring endomorphism of Multivariate Polynomial Ring in x, y, z over
Rational Field
  Defn: x |--> -y^4*z - 2*x*y^2*z^2 - x^2*z^3 + 2*y^3 + 2*x*y*z + x
        y |--> -y^2*z - x*z^2 + y
        z |--> z
sage: (tau * sigma).is_identity()
True

The method inverse_image can be used to test whether an element is contained in a subalgebra:

sage: R.<s,t> = PolynomialRing(QQ)
sage: S.<x,y,z,w> = PolynomialRing(QQ)
sage: f = S.hom([s^4, s^3*t, s*t^3, t^4], R)
sage: f.inverse_image(R.ideal(0))
Ideal (y*z - x*w, z^3 - y*w^2, x*z^2 - y^2*w, y^3 - x^2*z) of
Multivariate Polynomial Ring in x, y, z, w over Rational Field
sage: f.inverse_image(s^3*t^4*(s+t))
x*w + y*w
sage: f.inverse_image(s^2*t^2)
...
ValueError: element s^2*t^2 does not have preimage

Manifolds

diff function for exterior derivatives

It is now possible to invoke diff to compute the differential (exterior derivative) of a differentiable form (#29953). For instance, for a scalar field:

sage: M = Manifold(2, 'M')
sage: X.<x,y> = M.chart()
sage: f = M.scalar_field(x^2*y, name='f')
sage: diff(f)
1-form df on the 2-dimensional differentiable manifold M
sage: diff(f).display()
df = 2*x*y dx + x^2 dy

and for a 1-form:

sage: a = M.one_form(-y, x, name='a'); a.display()
a = -y dx + x dy
sage: diff(a)
2-form da on the 2-dimensional differentiable manifold M
sage: diff(a).display()
da = 2 dx/\dy

Dot and cross products of vector fields along a curve

The methods dot_product(), cross_product() and norm() can be now be used for vector fields defined along a differentiable map, the codomain of which is a Riemannian manifold (#30318). Previously, these methods worked only for vector fields on a Riemannian manifold, i.e. along the identity map. An important subcase is of course that of a curve in a Riemannian manifold. For instance, let us consider a helix C in the Euclidean space E3 parametrized by its arc length s:

sage: E.<x,y,z> = EuclideanSpace()
sage: R.<s> = RealLine()
sage: C = E.curve((2*cos(s/3), 2*sin(s/3), sqrt(5)*s/3), (s, -oo, +oo),
....:             name='C')
sage: C.display()                                                               
C: R --> E^3
   s |--> (x, y, z) = (2*cos(1/3*s), 2*sin(1/3*s), 1/3*sqrt(5)*s)

The tangent vector field T=C' has a unit norm since the parameter s is the arc length:

sage: T = C.tangent_vector_field()
sage: T.display()
C' = -2/3*sin(1/3*s) e_x + 2/3*cos(1/3*s) e_y + 1/3*sqrt(5) e_z
sage: norm(T)
Scalar field |C'| on the Real interval (0, 6*pi)
sage: norm(T).expr()
1

We introduce the unit normal vector N via the derivative of T:

sage: T_prime = R.vector_field([diff(T[i], s) for i in E.irange()], dest_map=C,
....:                          name="T'")
sage: N = T_prime / norm(T_prime) 
sage: N.display()                                                               
-cos(1/3*s) e_x - sin(1/3*s) e_y

and we get the binormal vector B as the cross product of T and N:

sage: B = T.cross_product(N)
sage: B
Vector field along the Real number line R with values on the Euclidean space E^3
sage: B.display()                                                               
1/3*sqrt(5)*sin(1/3*s) e_x - 1/3*sqrt(5)*cos(1/3*s) e_y + 2/3 e_z

We can then form the Frenet-Serret frame:

sage: FS = R.vector_frame(('T', 'N', 'B'), (T, N, B),
....:                     symbol_dual=('t', 'n', 'b'))
sage: FS
Vector frame (R, (T,N,B)) with values on the Euclidean space E^3

and check that it is orthonormal:

sage: matrix([[u.dot(v).expr() for v in FS] for u in FS])                       
[1 0 0]
[0 1 0]
[0 0 1]

The Frenet-Serret formulas, expressing the curvature and torsion of C, are obtained as:

sage: N_prime = R.vector_field([diff(N[i], s) for i in E.irange()],
....:                          dest_map=C, name="N'")
sage: B_prime = R.vector_field([diff(B[i], s) for i in E.irange()],
....:                          dest_map=C, name="B'")
sage: for v in (T_prime, N_prime, B_prime): 
....:     v.display(FS) 
....:                                                                           
T' = 2/9 N
N' = -2/9 T + 1/9*sqrt(5) B
B' = -1/9*sqrt(5) N

Orientability of manifolds and vector bundles

It is now possible to define an orientation on a differentiable manifold and on a vector bundle (#30178). Orientations of topological manifolds have also been introduced, according to this definition.

Euclidean spaces as metric spaces

Euclidean spaces have been endowed with a distance function and have been set in the category of complete metric spaces (#30062):

sage: E.<x,y> = EuclideanSpace()
sage: p = E((1,0))  # the point of coordinates (1,0)
sage: q = E((0,2))  # the point of coordinates (0,2)
sage: d = E.dist  # the distance function
sage: d(p,q)
sqrt(5)
sage: p.dist(q)
sqrt(5)
sage: E.category()
Join of Category of smooth manifolds over Real Field with 53 bits of precision and Category of complete metric spaces

Bundle connections

Bundle connections have been improved (#30208) and their action on vector fields and sections has been implemented (#30209).

Internal code improvements and bug fixes

Many improvements/refactoring of the code have been performed in this release:

In addition, various bugs have been fixed: #30108, #30112, #30191, #30275, #30289, #30320, #30401, #30519.

Algebra

Lie Conformal Algebras

Implemented Lie conformal algebras and superalgebras. Here are some examples of their usage:

sage: V = lie_conformal_algebras.Virasoro(QQ); V
The Virasoro Lie conformal algebra over Rational Field
sage: V.inject_variables()
Defining L, C
sage: L.bracket(L)
{0: TL, 1: 2*L, 3: 1/2*C}
sage: L.T(2).bracket(L)
{2: 2*TL, 3: 12*L, 5: 10*C}

sage: V = lie_conformal_algebras.NeveuSchwarz(QQ)
sage: V.some_elements()
[L, G, C, TG, TG + 4*T^(2)G, 4*T^(2)G]

sage: W = lie_conformal_algebras.FreeFermions(QQbar, 2); W
The free Fermions super Lie conformal algebra with generators (psi_0, psi_1, K) over Algebraic Field
sage: W.inject_variables()
Defining psi_0, psi_1, K
sage: psi_0.bracket(psi_1.T())
{}
sage: psi_0.bracket(psi_0.T())
{1: K}
sage: psi_0.is_even_odd()
1

For documentation on implemented features see Lie Conformal Algebra. For a list of implemented examples see Lie Conformal Algebra Examples.

Differential Weyl algebra

The action of differential operators from the Weyl algebra on polynomials has been implemented #29928:

sage: W.<x,y> = DifferentialWeylAlgebra(QQ)
sage: dx, dy = W.differentials()
sage: dx.diff(x^3)
3*x^2
sage: (x*dx + dy + 1).diff(x^4*y^4 + 1)
5*x^4*y^4 + 4*x^4*y^3 + 1

Improved Unicode support

Unicode identifiers

Python 3 made much improved support for Unicode available, and Sage 9.2 has merged several Unicode improvements. Note that Python does not allow arbitrary Unicode characters in identifiers but only word constituents. So before you get excited about using emojis... note that they cannot be used:

   1 sage: K.<🍎,🥝> = QQ[]
   2 SyntaxError: invalid character in identifier

However, we can use letters from various alphabets. The updated IPython allows us to type them using latex and unicode tab completion:

   1 sage: μ, ν, ξ = 1, 2, 3       # type \mu<TAB>, 
   2                               #      \nu<TAB>, ...
   3 sage: SR('λ + 2λ')
   4 3
   5 sage: var('α', domain='real')
   6 α
   7 sage: Ш = EllipticCurve('389a').sha()   
   8                               # type \CYR<TAB> CAP<TAB>
   9                               #      LET<TAB> SHA<TAB><ENTER>
  10 sage: Ш
  11 Tate-Shafarevich group for the Elliptic Curve
  12 defined by y^2 + y = x^3 + x^2 - 2*x over Rational Field
  13 sage: GelʹfandT͡setlinPattern = GelfandTsetlinPattern
  14                               # type \MODIFIER LETTER 
  15                               #      PRIME<TAB><ENTER>
  16                               # for the romanized soft mark
  17 sage: ГельфандЦетлинPattern = GelʹfandT͡setlinPattern
  18 sage: ГельфандЦетлинPattern([[3, 2, 1], [2, 1], [1]]).pp()
  19   3     2     1
  20      2     1
  21         1
  22 sage: 四次方(x) = x^4
  23 sage: 四次方(3)
  24 81

We can use math accents...

   1 sage: a = 1
   2 sage: â = 2                   # type a\hat<TAB><ENTER>
   3 sage: ā = 3                   # type a\bar<TAB><ENTER>
   4 sage: a, â, ā
   5 (1, 2, 3)
   6 sage: s(t) = t^3; s
   7 t |--> t^3
   8 sage: ṡ = diff(s, t); ṡ       # type s\dot<TAB><ENTER>                                                                                
   9 t |--> 3*t^2
  10 sage: s̈ = diff(ṡ, t); s̈       # type s\ddot<TAB><ENTER>                                                                                                   
  11 t |--> 6*t

... and have fun with modifier letters:

   1 sage: ℚ̄ = QQbar               # type \bbQ<TAB>\bar<TAB>
   2 sage: %display unicode_art
   3 sage: A = matrix(ℚ̄, [[1, 2*I], [3*I, 4]]); A
   4 1 2*I
   5 3*I   4
   6 sage: Aᵀ = A.transpose()      # type A\^T<TAB><ENTER>
   7 sage: A
   8 1 3*I
   9 2*I   4
  10 sage: Aᴴ = A.conjugate_transpose()
  11                               # type A\^H<TAB><ENTER>
  12 sage: A
  13 1 -3*I
  14 ⎝-2*I    4
  15 sage: C = Cone([[1, 1], [0, 1]])                                                                           
  16 sage: Cᵒ = C.dual(); C# type C\^o<TAB><ENTER>
  17 2-d cone in 2-d lattice M                                                                                        

But note that Python normalizes identifiers, so the following variants are not distinguished:

   1 sage: AT == Aᵀ, AH == Aᴴ, Co == C
   2 (True, True, True)
   3 sage: ℚ = QQ                  # type \bbQ<TAB><ENTER>
   4 sage: ℚ
   5 Rational Field
   6 sage: Q = 42
   7 sage: ℚ
   8 42
   9 sage: F = 1
  10 sage: 𝐹, 𝐅, 𝓕, 𝕱, 𝗙, 𝘍, 𝙁, 𝙵 # type \itF<TAB>, \bfF<TAB>,
  11                               #      \scrF<TAB>, \frakF<TAB>,
  12                               #      \sansF<TAB>, ...
  13 (1, 1, 1, 1, 1, 1, 1, 1)

We have also added a few Unicode aliases for global constants and functions.

   1 sage: π
   2 pi
   3 sage: _.n()
   4 3.14159265358979
   5 sage: Γ(5/2)                                                                                                                    
   6 3/4*sqrt(pi)
   7 sage: ζ(-1)
   8 -1/12

See Meta-ticket #30111: Unicode support for more information.

Unicode characters allowed in tensor index notation

Greek letters (and more generally any Unicode non-digit word-constituent character) are now allowed in index notation for tensors (#29248). For instance, taking the trace of a type-(1,1) tensor field:

sage: E.<x,y> = EuclideanSpace()                                                
sage: t = E.tensor_field(1, 1, [[x, 1], [0, y]])                                
sage: t['^μ_μ']                                                                 
Scalar field on the Euclidean plane E^2
sage: t['^μ_μ'] == t.trace()                                                    
True

Unicode art

Configuration and build changes

Initial configuration with ./configure required

Sage 9.1 introduced informational messages at the end of a ./configure run regarding system packages. To make sure that these messages are not overlooked, Sage 9.2 no longer invokes ./configure when you type make in an unconfigured source tree. See sage-devel: require "./configure" before "make", #29316.

Support for gcc/gfortran 10.x added

All standard Sage packages have been upgraded in Sage 9.2 so that they build correctly using gcc/gfortran 10.x. The Sage ./configure script therefore now accepts these compiler versions.

Selecting a system Python to use for Sage's venv

The configure script in Sage 9.2 has been changed so it only looks for a binary named python3 in your PATH. If Sage cannot find a suitable python3 in your PATH, it will build its own copy of Python 3.8.5. Sage no longer looks for versioned Python binaries such as python3.7, #30546.

To configure Sage to use a specific Python installation, you can use ./configure --with-python=/path/to/python3. The configure script will test whether this installation is suitable for Sage and will exit with an error otherwise.

System package information for more distributions

System package information has been added for Gentoo Linux, FreeBSD, Void Linux, and NixOS.

System package information for optional packages at runtime

When a user tries to use a feature depending on an optional package is not installed, Sage now issues advice regarding the packages that should be installed to provide the feature -- using either the system package manager, pip, or (in the Sage distribution) sage -i #30606.

(For packagers: For this to work, either SAGE_ROOT/build/pkgs/*/distros/ and SAGE_ROOT/build/bin/{sage-get-system-packages, sage-guess-package-system, sage-print-system-package-command} need to be installed, or alternative implementations of these scripts need to be provided.)

For developers: Changes to the build system of sagelib

Let's talk about setup.py. The build system of the Sage library, in build/pkgs/sagelib/src/setup.py, is based on distutils (not setuptools); it is implemented in the package sage_setup. In particular, it implements its own version of source code discovery methods similar to setuptools.find_packages: sage_setup.find.find_python_sources. Because of source discovery, developers can add new Python modules and packages under src/sage/ simply by creating files and directories; it is not necessary to edit setup.py.

Prior to Sage 9.2, the situation was different for Cython extensions. They had to be listed in src/module_list.py, either one by one, or using glob patterns such as * and **. Sage 9.2 has eliminated the need for src/module_list.py by extending sage_setup.find.find_python_sources; it now also finds Cython modules in the source tree (Trac #29701).

Some Cython modules need specific compiler and linker flags. Sage 9.2 has moved all of these flags from Extension options in src/module_list.py to distutils: directives in the individual .pyx source files, see #29706 and Cython documentation.

Sage 9.2 has also changed the mechanism for conditionalizing a Cython extension module on the presence of a Sage package. Consider the module sage.graphs.graph_decompositions.tdlib as an example. Prior to Sage 9.2, this module was declared as an OptionalExtension, conditional on the SPKG tdlib, in src/module_list.py. The new mechanism is as follows. src/setup.py maps the SPKG name tdlib to the "distribution name" sage-tdlib. At the top of the Cython source file src/sage/graphs/graph_decompositions/tdlib.pyx, there is a new directive sage_setup: distribution = sage-tdlib. Now the source discovery in sage_setup.find.find_python_sources includes this Cython module only if the SPKG tdlib is installed and current.

New development tools

Testing and linting with tox

tox is a popular package that is used by a large number of Python projects as the standard entry point for testing and linting.

Sage 9.1 started to use tox for portability testing of the Sage distribution, which requires an installation of tox in the system python.

Sage 9.2 has added a tox configuration (src/tox.ini) for the (more typical) use of tox for testing and linting of the Sage library #30453. This provides an entry point for various testing/linting methods that is more idiomatic from the viewpoint of the Python community.

The commands sage -t, sage -coverage, sage -coverageall, and sage -startuptime are repackaged as sage --tox, as the following output from sage -advanced indicates:

$ ./sage -advanced
SageMath version 9.2
...
Testing files:
...
  --tox [options] <files|dirs> -- general entry point for testing
                                  and linting of the Sage library
     -e <envlist>     -- run specific test environments (default: run all)
        doctest          -- run the Sage doctester 
                            (same as "sage -t")
        coverage         -- information about doctest coverage of files 
                            (same as "sage --coverage[all]")
        startuptime      -- display how long each component of Sage 
                            takes to start up 
                            (same as "sage --startuptime")

Three new linting methods are added:

        pycodestyle      -- check against the Python style conventions of PEP8
        relint           -- check whether some forbidden patterns appear 
                            (includes all patchbot pattern-exclusion plugins)
        codespell        -- check for misspelled words in source code

This functionality is available after installing the optional tox package using sage -i tox (or having tox available in the system).

Reusable wheels for the Python packages built by the Sage distribution

Sage 9.2 has changed the build process of all Python packages in the Sage distribution so that wheels are built and stored in $SAGE_LOCAL/var/lib/sage/wheels/ #29500.

Users can install these wheels into virtual environments that use the same base python version, using standard tools such as pip install --find-links. Example:

$ local/bin/python3 -m venv some-venv
$ cd some-venv/
$ bin/pip3 install -v --no-index --find-links=../local/var/lib/sage/wheels/ Pillow    

The installation is very fast because it does not involve compilation.

Note: Many of these wheels include extension modules that refer to libraries in the Sage installation in $SAGE_LOCAL with hard-coded paths. Therefore the wheels are not immediately suitable for distribution.

Cleaning

Availability of Sage 9.2 and installation help

SageMath 9.2 was released on 2020-10-24; it is available in the sage git repository, and the self-contained source tarballs are available for download.

Sage 9.2 has been tested to compile from source on a wide variety of platforms, including:

Binaries

Availability in distributions

See repology.org: sagemath

Installation FAQ

See README.md in the source distribution for installation instructions.

See sage-release, sage-devel.

More details

ReleaseTours/sage-9.2 (last edited 2020-10-30 02:53:32 by mkoeppe)