v0.0.6

What's Changed

Features

• Added Birkhoff-Trjitzinsky asymptotics to extract a formal basis of asymptotic solutions for scalar linear difference equations (LinearRecurrence.asymptotics()).
• Added new classes to support asymptotic calculations: SeriesMatrix, GrowthRate, and Reducer.
• Added calculation of the determinant of a pFq CMF axis matrix via a hardcoded formula.
• Made determinant calculations a built-in CMF functionality.

Bug Fixes

• Fixed broken LaTeX rendering of objects in JupyterLab (such as PCF, CMF, and LinearRecurrence) by correctly wrapping the output in math delimiters.
• Fixed an equality evaluation bug in FlintRational occurring when negations were present in both the numerator and denominator.
• Fixed pickling cache issues for CMF, pFq, and MeijerG objects by correctly implementing their respective __getstate__ and __setstate__ methods.

Refactors

• Replaced custom _repr_latex_ methods with sympy.printing.defaults.Printable inheritance for cleaner and more standardized LaTeX representation.

New Contributors

• @UriKH made their first contribution in #143
• @tsvikas made their first contribution in #146
• @talimonodromy made their first contribution in #145

Full Changelog: v0.0.5...v0.0.6

v0.0.5

Changelog:

• Features

  • Added the Meijer G CMF
  • Added the DFinite base class (which inherits from CMF) for constructing D-finite CMFs. Consists methods such as:
    • differential equation
    • construct matrix (in the future could be derived automatically from contiguous relations)
  • Added support for caching negative matrices when using them while constructing the CMF (For example, in the MeijerG CMF we first calculate the a matrices in the inverse direction as this is the direction we get from the contiguous relations).
  • Added the CMF.coboundary method which computes coboundary on the enitre CMF.
  • Added the CMF.dual method which returns the dual CMF (all matrices are inverse-transposed).
  • Added the CMF.sub_cmf method which projects the CMF into a smaller subspace.
  • Added the LinearRecurrence.compose function

• Refactors

  • LinearRecurrence now prints as sum = 0 and not as p(n+1) = sum.
  • pFq and MeijerG now inherit from the D-finite CMF.

v0.0.4

Changelog:

• Bugfixes:
  • Fixed the bug in Limit.identify where sometimes it just returns the zero matrix.

v0.0.3

Changelog:

• Bugfixes:
  • Fixed a bug in the LaTeX representations of LinearRecurrence, PCF and CMF.
  • Limit.delta now returns a float instead of an mp.mpf

v0.0.2

Changelog:

• Refactors:

  • Flattened the cmf module. The contents of ffbar and known_cmfs are directly in cmf now.
    • Notice that import now looks as from ramanujantools.cmf import pFq
  • Grouped all solvers in a new solvers module
  • Removed multimethod usage (and dependency)
  • PCF now uses LinearRecurrence for key logic, such as inflate, deflate, and especially conversion from matrix form
    • PCFFromMatrix was deleted completely
  • Merged NumericMatrix.walk and NumericMatrix.walk_list into a single NumericMatrix.walk that accepts both int and list[int]

• Bugfixes:

  • Matrix and LinearRecurrence now accept initial_values in their walk and limit methods
  • Fixed a bug where Matrix.factor would crash when the matrix was numeric (no symbols)

• Features:

  • Add a distinct repr to pFq
  • Add LaTeX pretty printing to CMF, PCF and LinearRecurrence:

v0.0.1 – Initial PyPi Release

Initial public release of ramanujantools on PyPi! 🎉

This version includes:

• Core symbolic and numeric tools used by the Ramanujan Machine group
• Contains the key classes
  • Matrix
  • PCF (Polynomial Continued Fraction)
  • LinearRecurrence
  • CMF (Conservative Matrix Field)
  • Limit
• Python 3.11+ compatibility
• Optimized python-flint backend