graph LR
geomstats_numerics_geodesic["geomstats.numerics.geodesic"]
geomstats_numerics_optimization["geomstats.numerics.optimization"]
geomstats_numerics_ivp["geomstats.numerics.ivp"]
geomstats_numerics_interpolation["geomstats.numerics.interpolation"]
geomstats_numerics_path["geomstats.numerics.path"]
geomstats_numerics__common["geomstats.numerics._common"]
geomstats_numerics_geodesic -- "uses" --> geomstats_numerics_ivp
geomstats_numerics_geodesic -- "uses" --> geomstats_numerics_optimization
geomstats_numerics_optimization -- "uses" --> geomstats_numerics__common
geomstats_numerics_ivp -- "might use" --> geomstats_numerics__common
geomstats_numerics_path -- "might use" --> geomstats_numerics_interpolation
This subsystem provides the core computational methods for solving various geometric problems within geomstats, focusing on numerical techniques for differential geometry and related machine learning tasks. It adheres to the Scientific Computing/Machine Learning Toolkit pattern by offering specialized, modular numerical algorithms.
Orchestrates the computation of geodesics, including exponential maps, logarithmic maps, and solving geodesic boundary value problems. It serves as the primary interface for geodesic-related numerical tasks.
Related Classes/Methods:
Offers a suite of optimization algorithms, crucial for minimizing objective functions that arise in various geometric problems, particularly in solving boundary value problems. It abstracts different optimization backends.
Related Classes/Methods:
Provides fundamental numerical integration capabilities for solving ordinary differential equations (ODEs) as initial value problems. It is a low-level workhorse for time-evolution problems.
Related Classes/Methods:
Manages the interpolation of data points, essential for generating smooth paths or approximating functions within geometric spaces.
Related Classes/Methods:
Computes the energy of a given path, a fundamental concept in geometric analysis and optimization, often used in variational problems.
Related Classes/Methods:
Provides foundational utilities for ensuring backend compatibility and consistent data type handling across different numerical operations, acting as a crucial support component.
Related Classes/Methods: