What's New in Version 8.0
.NET Core and .NET Standard support
- Support for .NET Core 3.1.
- Support for .NET Standard 2.0.
- Support for .NET Framework 4.0 and 4.6.
- Support for .NET 5.0.
- All packages are available on the Nuget Gallery.
Linear algebra
Major enhancements
- Upgraded native libraries to Intel® OneAPI version 2023 Update 1.
- Upgraded managed linear algebra library to LAPACK 3.9.1.
- Support for indexing using range and index operators.
Other improvements
- Improve performance for managed complex linear algebra.
- Code quality and performance improvements in the machine translation of LAPACK.
- Deconstruction of matrix decompositions into their component matrices.
Mathematics
General improvements
- Complex number arithmetic is optimized to be on par with non-generic code.
- Easier creation of complex numbers thanks to an implicit conversion from a 2-tuple to a complex number.
- Deconstructors for complex and rational numbers.
- New integration algorithm: double-exponential quadrature.
- Automatic differentiation supports the new System.Math methods.
Equation solvers
- New one-dimensional equation solver based on TOMS algorithm 487.
- One-dimensional equation solvers can now restrict their search to a user-defined interval.
- Improved robustness of one-dimensional equation solvers.
Special functions
- New: Struve functions, which are the solutions of the inhomogeneous Bessel differential equations.
- Improved accuracy and performance of various gamma-related functions.
Statistics and data analysis
Hypothesis tests
- New: Fisher’s Exact Test for 2x2 and RxC contingency tables.
- Breusch-Godfrey test for serial correlation in the residuals of a linear regression.
Probability distributions
- New: DiscreteDistribution<T> that can sample from a finite set of arbitrary objects.
- New continuous distributions: Johnson family of distributions, arcsine distribution, PERT distribution.
- New properties of distributions: Entropy, Median, Mode, IsMultimodal, GetAllModes()
- New LogProbability method for discrete distributions.
- Gumbel (extreme value) distribution now allows modelling of the smallest or the largest extreme value.
- New: Estimation of inverse Weibull and Gumbel distributions.
Want to go further back? See what was new in version 7.0.