RMC.Numerics
2.3.0
dotnet add package RMC.Numerics --version 2.3.0
NuGet\Install-Package RMC.Numerics -Version 2.3.0
<PackageReference Include="RMC.Numerics" Version="2.3.0" />
<PackageVersion Include="RMC.Numerics" Version="2.3.0" />
<PackageReference Include="RMC.Numerics" />
paket add RMC.Numerics --version 2.3.0
#r "nuget: RMC.Numerics, 2.3.0"
#:package RMC.Numerics@2.3.0
#addin nuget:?package=RMC.Numerics&version=2.3.0
#tool nuget:?package=RMC.Numerics&version=2.3.0
Numerics
Numerics is a free and open-source numerical computing library for .NET developed by the U.S. Army Corps of Engineers Risk Management Center (USACE-RMC). It provides methods and algorithms for probability distributions, statistical analysis, numerical methods, optimization, machine learning, and Bayesian MCMC sampling, with a focus on hydrological and risk assessment applications.
Supported Frameworks
| Framework | Version |
|---|---|
| .NET | 10.0 |
| .NET Framework | 4.8.1 |
The .NET 10 build runs on Windows and Linux. Version 2.3.0 targets net10.0 and net481; applications on .NET 8 or .NET 9 can stay on RMC.Numerics 2.2.0.
Install via NuGet:
dotnet add package RMC.Numerics --version 2.3.0
Or search for RMC.Numerics in the NuGet Package Manager.
Python
pynumerics ships the same Numerics.dll with typed Python wrappers over every Numerics area. It requires CPython 3.10 to 3.14 on Linux or Windows and the .NET 10 runtime:
pip install pynumerics
What's New in 2.3.0
- Hosking-Wallis regional frequency analysis: discordancy, heterogeneity and goodness-of-fit measures, and regional growth curves.
- Machine learning: Gaussian process regression, kriging, elastic net regression, and principal component analysis.
- Optimization: L-BFGS-B, CMA-ES, dynamically dimensioned search, Pareto fronts, dynamic programming, and knapsack solvers.
- Numerical methods: a QL eigensolver, an opt-in QUADPACK QAGS integrator, multi-source shortest paths, column statistics, and compensated sums.
- Accuracy fixes in the Student-t distribution, the incomplete beta and inverse incomplete gamma functions, the Archimedean copula densities, and the GEV, Gamma and Pearson Type III L-moments.
- The pynumerics Python package, published to PyPI with each release.
See the v2.3.0 release notes for the full list, including breaking changes and the results that move.
Documentation
User Guide and API Documentation - Comprehensive documentation with code examples and mathematical explanations.
| Section | Topics |
|---|---|
| Mathematics | Adaptive one- and two-dimensional integration, differentiation, optimization, root finding, linear algebra, ODE solvers, special functions, Fourier methods |
| Data | Interpolation, linear regression, time series analysis |
| Statistics | Descriptive and weighted statistics, goodness-of-fit metrics, hypothesis tests, global sensitivity analysis |
| Distributions | 43 univariate distributions, parameter estimation, uncertainty analysis, copulas, multivariate distributions, regional frequency analysis |
| Functions | Uncertain univariate function forms, composition, posterior ensembles, link functions |
| Machine Learning | GLM, elastic net, Gaussian process regression, kriging, decision trees, random forests, KNN, naive Bayes, k-means, GMM, PCA |
| Sampling | MCMC (RWMH, ARWMH, DE-MCz, DE-MCzs, HMC, NUTS, Gibbs), random generation, scrambled quasi-random sequences, convergence diagnostics |
| References | Consolidated bibliography |
Prerequisites
- .NET 10 runtime on Windows or Linux, or .NET Framework 4.8.1 on Windows. Install the .NET SDK if you don't already have it.
- The Microsoft .NET SDK is available on the App Portal for Corps users.
Support
USACE-RMC is committed to maintaining and supporting the library with regular updates, bug fixes, and enhancements.
The repository includes a unit testing library with over 4,000 tests that also serve as usage examples for the classes and methods in the library, and the pynumerics package adds over 5,000 Python tests.
Contributing
Contributions are welcome. Please see CONTRIBUTING.md for guidelines on bug reports, feature requests, and pull requests.
License
See LICENSE for details.
| 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. |
| .NET Framework | net481 is compatible. |
-
.NETFramework 4.8.1
- Microsoft.Bcl.HashCode (>= 6.0.0)
- System.Text.Encodings.Web (>= 10.0.1)
- System.Text.Json (>= 10.0.1)
- System.Threading.Tasks.Extensions (>= 4.6.3)
- System.ValueTuple (>= 4.6.1)
-
net10.0
- No dependencies.
NuGet packages (2)
Showing the top 2 NuGet packages that depend on RMC.Numerics:
| Package | Downloads |
|---|---|
|
RMC.Wpf.Framework.Controls
WPF controls, framework UI, DAG controls, database controls, numeric controls, OxyPlot controls, and AvalonDock assemblies for the RMC WPF Framework. |
|
|
RMC.BestFit
Portable model library for RMC-BestFit Bayesian estimation, distribution fitting, and frequency analysis in flood risk and water resources engineering. |
GitHub repositories
This package is not used by any popular GitHub repositories.
Version 2.3.0 adds Hosking-Wallis regional frequency analysis (Numerics.Distributions.Regional), Gaussian process regression, kriging, elastic net and PCA, the L-BFGS-B, CMA-ES and DDS optimizers with Pareto-front, dynamic programming and knapsack utilities, a QL eigensolver, an opt-in QAGS integrator, multi-source CSR shortest paths, column statistics, compensated sums, and per-element Poisson and Binomial array members. It fixes accuracy defects in StudentT, the incomplete beta and inverse incomplete gamma functions, the Joe, Clayton, Frank and Gumbel copula densities, GEV, Gamma and Pearson Type III L-moments, variance, covariance and Pearson correlation on offset data, and Kernel Density and Truncated modes. GoodnessOfFit.KolmogorovSmirnov and AndersonDarling now sort a copy of unsorted input, with new dataIsSorted overloads. GEV ParametersFromLinearMoments now throws for a non-positive L-scale, and Dijkstra and Network reject negative edge weights. The net10.0 build is tested on Windows and Linux, and the pynumerics Python package, which ships this assembly, is published to PyPI with each release. Breaking: the package targets net10.0 and net481 only; .NET 8 and .NET 9 reach end of support on November 10, 2026, so projects on those runtimes can stay on 2.2.0. See the v2.3.0 GitHub release notes for the full list and the results that move.