Module dense 

Dense matrix decompositions and solvers.

This module provides GPU-accelerated implementations of the core dense linear algebra algorithms: LU, QR, Cholesky, SVD, eigendecomposition, matrix inverse, determinant, and least squares.

Re-exports

pub use band::BandMatrix;

pub use band::band_cholesky;

pub use band::band_lu;

pub use band::band_solve;

pub use batched::BatchAlgorithm;

pub use batched::BatchConfig;

pub use batched::BatchedResult;

pub use batched::BatchedSolver;

pub use cholesky::cholesky;

pub use cholesky::cholesky_solve;

pub use dc_svd::DcSvdConfig;

pub use dc_svd::dc_svd;

pub use det::determinant;

pub use det::log_determinant;

pub use eig::EigJob;

pub use eig::syevd;

pub use inverse::inverse;

pub use ldlt::LdltResult;

pub use ldlt::ldlt;

pub use ldlt::ldlt_solve;

pub use lstsq::lstsq;

pub use lu::LuResult;

pub use lu::lu_factorize;

pub use lu::lu_solve;

pub use matrix_functions::MatrixExpConfig;

pub use matrix_functions::MatrixExpPlan;

pub use matrix_functions::MatrixLogConfig;

pub use matrix_functions::MatrixLogPlan;

pub use matrix_functions::MatrixSqrtConfig;

pub use matrix_functions::MatrixSqrtPlan;

pub use ode_pde::AdvectionEquation1D;

pub use ode_pde::Bdf2Solver;

pub use ode_pde::BoundaryCondition;

pub use ode_pde::EulerSolver;

pub use ode_pde::Grid1D;

pub use ode_pde::Grid2D;

pub use ode_pde::HeatEquation1D;

pub use ode_pde::ImplicitEulerSolver;

pub use ode_pde::OdeConfig;

pub use ode_pde::OdeMethod;

pub use ode_pde::OdeSolution;

pub use ode_pde::OdeSystem;

pub use ode_pde::PdeConfig;

pub use ode_pde::Poisson1D;

pub use ode_pde::Rk4Solver;

pub use ode_pde::Rk45Solver;

pub use ode_pde::StepResult;

pub use ode_pde::WaveEquation1D;

pub use ode_pde::numerical_jacobian;

pub use ode_pde::solve_tridiagonal as ode_solve_tridiagonal;

pub use qr::qr_factorize;

pub use qr::qr_generate_q;

pub use qr::qr_solve;

pub use qz::BalanceStrategy;

pub use qz::EigenvalueType;

pub use qz::QzConfig;

pub use qz::QzPlan;

pub use qz::QzResult;

pub use qz::QzStep;

pub use qz::ShiftStrategy;

pub use qz::classify_eigenvalue;

pub use qz::estimate_qz_flops;

pub use qz::plan_qz;

pub use qz::qz_host;

pub use qz::validate_qz_config;

pub use randomized_svd::RandomizedSvdConfig;

pub use randomized_svd::RandomizedSvdResult;

pub use randomized_svd::randomized_svd;

pub use svd::SvdJob;

pub use svd::SvdResult;

pub use svd::svd;

pub use tensor_decomp::CpAlsConfig;

pub use tensor_decomp::CpDecomposition;

pub use tensor_decomp::Matrix;

pub use tensor_decomp::Tensor;

pub use tensor_decomp::TtConfig;

pub use tensor_decomp::TtDecomposition;

pub use tensor_decomp::TuckerConfig;

pub use tensor_decomp::TuckerDecomposition;

pub use tensor_decomp::cp_als;

pub use tensor_decomp::hadamard_product;

pub use tensor_decomp::khatri_rao_product;

pub use tensor_decomp::mode_n_product;

pub use tensor_decomp::tt_svd;

pub use tensor_decomp::tucker_hooi;

pub use tensor_decomp::tucker_hosvd;

pub use tridiagonal::batched_tridiagonal_solve;

pub use tridiagonal::tridiagonal_solve;

Modules

band

Band Matrix Solvers.

batched

Batched matrix factorization for many small matrices in a single kernel launch.

cholesky

Cholesky Decomposition for symmetric positive definite matrices.

dc_svd

Divide-and-Conquer SVD.

det

Determinant computation via LU factorization.

eig

Symmetric eigenvalue decomposition.

inverse

Matrix inverse via LU factorization.

ldlt

Symmetric Indefinite Factorization (LDL^T / Bunch-Kaufman).

lstsq

Least squares solver.

lu

LU Factorization with partial pivoting.

matrix_functions

Matrix functions: exponential, logarithm, and square root.

ode_pde

ODE and PDE solver kernels.

qr

QR Factorization via blocked Householder reflections.

qz

Non-symmetric generalized eigenvalue solver (QZ algorithm).

randomized_svd

Randomized low-rank SVD algorithm (Halko, Martinsson, Tropp 2011).

svd

Singular Value Decomposition (SVD).

tensor_decomp

Tensor decomposition algorithms: CP, Tucker, and Tensor-Train (TT).

tridiagonal

Tridiagonal system solver.