Fuaran.Compute.LinearAlgebra
0.43.0
Prefix Reserved
dotnet add package Fuaran.Compute.LinearAlgebra --version 0.43.0
NuGet\Install-Package Fuaran.Compute.LinearAlgebra -Version 0.43.0
<PackageReference Include="Fuaran.Compute.LinearAlgebra" Version="0.43.0" />
<PackageVersion Include="Fuaran.Compute.LinearAlgebra" Version="0.43.0" />
<PackageReference Include="Fuaran.Compute.LinearAlgebra" />
paket add Fuaran.Compute.LinearAlgebra --version 0.43.0
#r "nuget: Fuaran.Compute.LinearAlgebra, 0.43.0"
#:package Fuaran.Compute.LinearAlgebra@0.43.0
#addin nuget:?package=Fuaran.Compute.LinearAlgebra&version=0.43.0
#tool nuget:?package=Fuaran.Compute.LinearAlgebra&version=0.43.0
Fuaran.Compute.LinearAlgebra — a small, dense, column-major linear-algebra contract whose every operation runs in one pinned floating-point order, so its answers are byte-identical on .NET, under Fable and in every host that ports it: an opaque immutable Matrix, vectors as Fuaran.Core's Vector<float> (a float column is an input without a copy), the level-1 operations (dot, axpy, scale, add, sub and the scaled norm2), gemv in column-axpy order, gemm in per-element k order, the Gram matrix, triangular solves over a checked UpperTriangular, the one Givens rotation, least squares by a streaming Givens QR whose O(p²) state survives a restart through its wire codec, batch and pivoted Householder factorizations, the pinned elementary functions exp, ln, log10 and float-exponent pow behind one special-value table, Poisson and Gamma log-link generalised linear models by IRLS with the same bits and iteration count on every host, typed errors and canonical NaN out. No fused multiply-add and no host mathematics call on any host. Fable-clean.
| Product | Versions Compatible and additional computed target framework versions. |
|---|---|
| .NET | net10.0 is compatible. net10.0-android was computed. net10.0-browser was computed. net10.0-ios was computed. net10.0-maccatalyst was computed. net10.0-macos was computed. net10.0-tvos was computed. net10.0-windows was computed. |
-
net10.0
- FSharp.Core (>= 10.1.300)
- Fuaran.Core.Column (>= 0.40.2)
- Fuaran.Core.Wire (>= 0.40.2)
NuGet packages (1)
Showing the top 1 NuGet packages that depend on Fuaran.Compute.LinearAlgebra:
| Package | Downloads |
|---|---|
|
Fuaran.Compute.DataFrame
Fuaran.Compute.DataFrame — the declarative-compute layer over Fuaran.Core.Column: a serializable dataframe-transform algebra (Transform + ColExpr, full v1 verb set — filter/project/derive/groupBy/join/window/pivot/unpivot/sort/distinct/limit/union), a pure reference evaluator with pinned null/coercion/order/float semantics (the cross-host parity contract), and a canonical wire codec. transformLaws (in Fuaran.Compute.Conformance) certifies any host evaluator byte-identical to the reference. Also carries the typed delta representation for the column layer (TableDelta: rows added/changed/removed addressed by a caller-supplied identity witness, column invalidation, and full-refresh as the top element) with an associative composition and a canonical wire codec, and the incremental evaluation seam over it: a pipeline evaluated against a delta rather than from scratch, with per-step incrementality declared as data (propagate / maintain-state / typed fall-back), a measured recompute footprint, and certified equality with the reference evaluator for every (base, delta) pair. Since 0.43.0 a pipeline also fits a regression per group (`Fit`) over Fuaran.Compute.LinearAlgebra. Fable-clean. |
GitHub repositories
This package is not used by any popular GitHub repositories.
| Version | Downloads | Last Updated |
|---|---|---|
| 0.43.0 | 0 | 11/5/2026 |