Geometric_Distributions.md

graph LR
    BrownianMotion["BrownianMotion"]
    LogNormal["LogNormal"]
    LogNormalEuclidean["LogNormalEuclidean"]
    LogNormalSPD["LogNormalSPD"]
    LogNormal -- "instantiates" --> LogNormalEuclidean
    LogNormal -- "instantiates" --> LogNormalSPD
    LogNormalEuclidean -- "inherits from" --> LogNormal
    LogNormalSPD -- "inherits from" --> LogNormal

Loading

Details

The Geometric Distributions subsystem (geomstats.distributions) defines and implements various probability distributions on geometric manifolds.

BrownianMotion

Manages the simulation and generation of Brownian motion paths on geometric manifolds. It provides the core functionality for stochastic processes within geometric spaces.

Related Classes/Methods:

• BrownianMotion:6-110

LogNormal

Serves as a polymorphic entry point and factory for Log-Normal distributions. Its key role is to abstract specific manifold implementation details, dynamically instantiating the appropriate specialized Log-Normal class based on the underlying manifold type. It also handles general initialization and sampling logic, acting as a base for manifold-specific implementations.

Related Classes/Methods:

• LogNormal:92-167

LogNormalEuclidean

Implements the Log-Normal distribution specifically for Euclidean manifolds, encapsulating the mathematical logic and sampling procedures relevant to this space. It specializes the general LogNormal behavior for Euclidean geometry.

Related Classes/Methods:

• LogNormalEuclidean:68-89

LogNormalSPD

Implements the Log-Normal distribution specifically for Symmetric Positive Definite (SPD) matrix manifolds, containing the specialized mathematical logic and sampling methods required for SPD matrices. It specializes the general LogNormal behavior for SPD geometry.

Related Classes/Methods:

• LogNormalSPD:15-65

FAQ