Morhpomatics 4.1.2 Trained Tiberius.

Fixing and improving bi-invariant two-smaple tests.

See also Commits.

Morhpomatics 4.1.1 Trained Tiberius.

Notable changes:

• new manifolds: size-and-shape space, simplex, diffeomorphism group (mostly restricted to action on point sets)
• tools for correpondence estimation based on functional maps
• new bi-invariant geodesic regression for Lie-group valued data
• LieGroup (like Metric) now inheriting Connection

See Commits for a detailed list of changes.

Fourth major release (aka Trained Tiberius) of the Morphomatics library.

New features in this release:

• new sub-packages providing geometric neural networks and graph operators on manfild-valued data
• added ProductManifold complementing PowerManifold for products consisting of different manifolds
• implementation of pole ladder for approximation of parallel transport
• added estimator for geometric median

Major improvements:

• Manifolds now implement pytree interface, hence, no longer have to be treated as static arguments in jax transforms like jax.jit
• batch feature provided by some manifolds removed as this use case is covered efficiently by PowerManifold via jax.vmap
• file names python modules now in snake case
• reduced potetial for cyclic dependencies

See Commits for a detailed list of changes.

Second microrelease of Morphomatics 3 Lucid Livia.

Notable changes:

• Added affine invariant metric for space of symmetric positive definite matrices
• Numerical safeguards for automatic differentiation

Complete list of changes here.

Third major release (aka Lucid Livia) of the Morphomatics library.

Major additions revolve around longitudinal modeling including:

• generic tangent bundle manifold with natural Sasaki metric,
• new Bézierfold of cubic splines based on product Sasaki metric,
• hyperbolic space, and
• Metric now subclass of Connection.

See Commits for a detailed list of changes.

Patch release fixing PointDistributionModel and dropping Python 3.7 support.

Minor release containing among others:

• new opt package with very basic (but jax compatible) optimization and root finding routines
• time-discretze Riemannian exp/log in Bezierfold
• performance improvements due to just-in-time compatible solver in RiemannianRegression and Bezierfold
• pymanopt now optional dependency

Detailed list of changes: Commits

Second major release (aka Accelerated Augustus) of the Morphomatics library.

Biggest change is the adoption of the jax framework for computing and statistics on manifolds, bringing

• automatic differentiation, and
• compilation and execution on accelerators (GPUs and TPUs).

See Commits for a detailed list of changes.

Major additions include:

• Bézierfold: Manifold of Riemannian Bézier Splines
• Higher-order hierarchichal model for analysis of longitudinal data (i.e. collection of time-dependent objects)
• n-dim. Kendall shape space
  For a full list of changes see Commits.

See morphomatics.github.io for an overview of features and tutorials.

First release (aka Curious Ceasar) of the Morphomatics library.

See morphomatics.github.io for an overview of features and tutorials.

Full Changelog: https://github.com/morphomatics/morphomatics/commits/v1.0