Package jdistlib.accelerator

package jdistlib.accelerator

Related Packages

Package	Description
jdistlib

Class	Description
CholeskyFactor	Immutable lower Cholesky factor with reusable SPD solves.
Compute	Selects automatic, CPU, or required accelerator execution.
ComputeApi	Concrete compute API or native library family.
ComputeBackend	Optional backend for vector, dense-linear-algebra, and batched likelihood work.
ComputeBackends	Discovers optional accelerator providers without making them core dependencies.
ComputeCapabilities	Immutable compute-device capabilities relevant to statistical kernels.
ComputeDeviceInfo	Immutable backend, runtime, driver, and device provenance.
ComputeSelection	Selected compute backend, device provenance, and ownership for one workflow.
CpuComputeBackend	Deterministic CPU reference implementation for every accelerated primitive.
ExecutionKind	How an inspected operation is expected to execute.
ExecutionPlan	Immutable prediction of where and how one operation will execute.
FloatCholeskyFactor	Immutable FP32 lower Cholesky factor with reusable SPD solves.
FloatLuFactor	Immutable FP32 partial-pivoted LU factorization of a square matrix.
FloatPivotedQrFactor	Immutable FP32 column-pivoted Householder QR factorization.
FloatSingularValueDecomposition	Immutable FP32 thin singular-value decomposition A = U*S*Vt.
FloatSparseCholeskyFactor	Immutable FP32 sparse Cholesky factor with reusable solves.
FloatSymmetricEigenDecomposition	Immutable FP32 eigendecomposition of a real symmetric matrix.
FloatSymmetricIndefiniteFactor	FP32 pivoted P*A*P' = L*D*L' factorization with 1x1/2x2 D blocks.
LinearAlgebraBackend	Backend-neutral FP64 BLAS, sparse-BLAS, and reusable factorization surface.
LinearAlgebraOperation	Backend-neutral operation identifiers used for capability and routing inspection.
LogisticRegressionBatchResult	Batched logistic-regression log densities and gradients.
LuFactor	Immutable FP64 partial-pivoted LU factorization of a square matrix.
MatrixDiagonal	Whether a triangular matrix stores or implies its diagonal.
MatrixSide	Side on which a triangular matrix operates.
MatrixTranspose	Transposition applied to a row-major dense matrix operand.
MatrixTriangle	Stored triangle of a triangular matrix.
NumericPrecision	Storage and arithmetic precision requested for a linear-algebra operation.
PivotedQrFactor	Immutable column-pivoted Householder QR factorization.
PreparedCholesky	Reusable FP64 Cholesky factor that can solve multiple right sides in place.
PreparedCsrMatrix	Reusable FP64 CSR handle; capabilities report whether storage is provider-resident.
PreparedDenseMatrix	Reusable FP64 dense matrix handle; providers may retain storage on device.
PreparedFloatCholesky	Reusable FP32 Cholesky factor that can solve multiple right sides in place.
PreparedFloatCsrMatrix	Reusable FP32 CSR handle; capabilities report whether storage is provider-resident.
PreparedFloatDenseMatrix	Reusable FP32 dense matrix handle; providers may retain storage on device.
PreparedFloatSparseCholesky	Symbolically analyzed FP32 sparse Cholesky handle with reusable numeric factors.
PreparedLogisticRegression	A logistic-regression data set prepared for repeated batched evaluation.
PreparedSparseCholesky	Symbolically analyzed FP64 sparse Cholesky handle with reusable numeric factors.
PreparedTransposeProduct	A row-by-feature matrix retained for repeated X'v score batches.
SinglePrecisionLinearAlgebraBackend	Backend-neutral FP32 BLAS, sparse-BLAS, and reusable factorization surface.
SingularValueDecomposition	Immutable FP64 thin singular-value decomposition A = U*S*Vt.
SparseCholeskyFactor	Immutable FP64 sparse Cholesky factor with reusable solves.
SparseOrdering	Fill-reducing ordering used before a sparse symmetric factorization.
SymmetricEigenDecomposition	Immutable FP64 eigendecomposition of a real symmetric matrix.
SymmetricIndefiniteFactor	Immutable pivoted P*A*P' = L*D*L' factorization with 1x1/2x2 D blocks.
UnaryOperation	Portable element-wise operations supported by accelerator backends.