Release v5.0.0-alpha01

RELEASENOTES.md

@@ -1,3 +1,22 @@
+### 5.0.0-alpha01 - 2021-06-27
+* COMPATIBILITY: net5.0, net48 better supported with explicit builds
+* COMPATIBILITY: netstandard1.x, net40, net45 no longer supported
+* BREAKING: drop all which was marked as obsolete
+* BREAKING: all native provider adapters moved out to separate NuGet packages
+* BREAKING: switch many usages of tuples to value tuples (experimental)
+* Distributions: Logistic *~Bobby Ingram*
+* Distributions: Perf: Cauchy avoid duplicate evaluation *~Febin*
+* Precision: Perf: pre-compute negative powers *~Febin*
+* Optimizations: Remove static properties in LevenbergMarquardtMinimizer *~Jong Hyun Kim*
+* Fit.Curve and FindMinimum extended to accept two more parameters
+* Series: stable series summation
+* Providers: drop managed reference linear algebra provider
+* Providers: native providers no longer inherit managed providers, managed now sealed
+* Providers: MKL provider compilation switched to Intel oneAPI MKL
+* Lots of internal cleanup, leveraging newer language features
+* Data: now released always together with Numerics (no longer separate versioning)
+* Control.Describe now includes CPU architecture and family identifier if know
+
 ### 4.15.0 - 2021-01-07
 * Precision: Round (with integer part rounding) *~Jon Larborn*
 * Precision: RoundToMultiple, RoundToPower

src/Data.Matlab/Data.Matlab.csproj

@@ -8,13 +8,27 @@
         <IsPackable>true</IsPackable>
         <PackageId>MathNet.Numerics.Data.Matlab</PackageId>
         <PackageId Condition="'$(StrongName)'=='True'">MathNet.Numerics.Data.Matlab.Signed</PackageId>
-        <VersionPrefix>4.15.0</VersionPrefix>
+        <VersionPrefix>5.0.0</VersionPrefix>
-        <VersionSuffix></VersionSuffix>
+        <VersionSuffix>alpha01</VersionSuffix>
         <Title>Math.NET Numerics - MATLAB Data I/O Extensions$(TitleSuffix)</Title>
         <Description>MathWorks MATLAB Data Input/Output Extensions for Math.NET Numerics, the numerical foundation of the Math.NET project, aiming to provide methods and algorithms for numerical computations in science, engineering and every day use.$(DescriptionSuffix)</Description>
-        <PackageReleaseNotes>Precision: Round (with integer part rounding) ~Jon Larborn
+        <PackageReleaseNotes>COMPATIBILITY: net5.0, net48 better supported with explicit builds
-Precision: RoundToMultiple, RoundToPower
+COMPATIBILITY: netstandard1.x, net40, net45 no longer supported
-F#: BigRational.FromDecimal ~Brian Berns</PackageReleaseNotes>
+BREAKING: drop all which was marked as obsolete
+BREAKING: all native provider adapters moved out to separate NuGet packages
+BREAKING: switch many usages of tuples to value tuples (experimental)
+Distributions: Logistic ~Bobby Ingram
+Distributions: Perf: Cauchy avoid duplicate evaluation ~Febin
+Precision: Perf: pre-compute negative powers ~Febin
+Optimizations: Remove static properties in LevenbergMarquardtMinimizer ~Jong Hyun Kim
+Fit.Curve and FindMinimum extended to accept two more parameters
+Series: stable series summation
+Providers: drop managed reference linear algebra provider
+Providers: native providers no longer inherit managed providers, managed now sealed
+Providers: MKL provider compilation switched to Intel oneAPI MKL
+Lots of internal cleanup, leveraging newer language features
+Data: now released always together with Numerics (no longer separate versioning)
+Control.Describe now includes CPU architecture and family identifier if know</PackageReleaseNotes>
         <PackageTags>math numeric data matlab</PackageTags>
         <IsTool>false</IsTool>
         <RepositoryUrl>https://github.com/mathnet/mathnet-numerics</RepositoryUrl>

src/Data.Tests/Properties/AssemblyInfo.cs

@@ -14,6 +14,6 @@ using System.Runtime.InteropServices;
 [assembly: ComVisible(false)]
 [assembly: Guid("bb41bee6-eb4c-48c9-8edd-7a5ae2422567")]
 
-[assembly: AssemblyVersion("4.15.0.0")]
+[assembly: AssemblyVersion("5.0.0.0")]
-[assembly: AssemblyFileVersion("4.15.0.0")]
+[assembly: AssemblyFileVersion("5.0.0.0")]
-[assembly: AssemblyInformationalVersion("4.15.0")]
+[assembly: AssemblyInformationalVersion("5.0.0-alpha01")]

src/Data.Text/Data.Text.csproj

@@ -8,13 +8,27 @@
         <IsPackable>true</IsPackable>
         <PackageId>MathNet.Numerics.Data.Text</PackageId>
         <PackageId Condition="'$(StrongName)'=='True'">MathNet.Numerics.Data.Text.Signed</PackageId>
-        <VersionPrefix>4.15.0</VersionPrefix>
+        <VersionPrefix>5.0.0</VersionPrefix>
-        <VersionSuffix></VersionSuffix>
+        <VersionSuffix>alpha01</VersionSuffix>
         <Title>Math.NET Numerics - Text Data I/O Extensions$(TitleSuffix)</Title>
         <Description>Text Data Input/Output Extensions for Math.NET Numerics, the numerical foundation of the Math.NET project, aiming to provide methods and algorithms for numerical computations in science, engineering and every day use.$(DescriptionSuffix)</Description>
-        <PackageReleaseNotes>Precision: Round (with integer part rounding) ~Jon Larborn
+        <PackageReleaseNotes>COMPATIBILITY: net5.0, net48 better supported with explicit builds
-Precision: RoundToMultiple, RoundToPower
+COMPATIBILITY: netstandard1.x, net40, net45 no longer supported
-F#: BigRational.FromDecimal ~Brian Berns</PackageReleaseNotes>
+BREAKING: drop all which was marked as obsolete
+BREAKING: all native provider adapters moved out to separate NuGet packages
+BREAKING: switch many usages of tuples to value tuples (experimental)
+Distributions: Logistic ~Bobby Ingram
+Distributions: Perf: Cauchy avoid duplicate evaluation ~Febin
+Precision: Perf: pre-compute negative powers ~Febin
+Optimizations: Remove static properties in LevenbergMarquardtMinimizer ~Jong Hyun Kim
+Fit.Curve and FindMinimum extended to accept two more parameters
+Series: stable series summation
+Providers: drop managed reference linear algebra provider
+Providers: native providers no longer inherit managed providers, managed now sealed
+Providers: MKL provider compilation switched to Intel oneAPI MKL
+Lots of internal cleanup, leveraging newer language features
+Data: now released always together with Numerics (no longer separate versioning)
+Control.Describe now includes CPU architecture and family identifier if know</PackageReleaseNotes>
         <PackageTags>math numeric data text csv tsv json xml</PackageTags>
         <IsTool>false</IsTool>
         <RepositoryUrl>https://github.com/mathnet/mathnet-numerics</RepositoryUrl>

src/FSharp.Tests/AssemblyInfo.fs

@@ -10,9 +10,9 @@ open System.Runtime.InteropServices
 [<assembly: AssemblyProduct("Math.NET Numerics")>]
 [<assembly: AssemblyCopyright("Copyright (c) Math.NET Project")>]
 
-[<assembly: AssemblyVersion("4.15.0.0")>]
+[<assembly: AssemblyVersion("5.0.0.0")>]
-[<assembly: AssemblyFileVersion("4.15.0.0")>]
+[<assembly: AssemblyFileVersion("5.0.0.0")>]
-[<assembly: AssemblyInformationalVersion("4.15.0")>]
+[<assembly: AssemblyInformationalVersion("5.0.0-alpha01")>]
 
 [<assembly: ComVisible(false)>]
 [<assembly: Guid("C9AA6156-F799-42E4-B50D-2E88AD7D1750")>]

src/FSharp/AssemblyInfo.fs

@@ -44,9 +44,9 @@ open System.Runtime.InteropServices
 [<assembly: AssemblyCulture("")>]
 [<assembly: NeutralResourcesLanguage("en")>]
 
-[<assembly: AssemblyVersion("4.15.0.0")>]
+[<assembly: AssemblyVersion("5.0.0.0")>]
-[<assembly: AssemblyFileVersion("4.15.0.0")>]
+[<assembly: AssemblyFileVersion("5.0.0.0")>]
-[<assembly: AssemblyInformationalVersion("4.15.0")>]
+[<assembly: AssemblyInformationalVersion("5.0.0-alpha01")>]
 
 [<assembly: ComVisible(false)>]
 [<assembly: Guid("048BC4EB-CE2B-4040-9967-4784F5405B0F")>]

src/FSharp/FSharp.fsproj

@@ -8,13 +8,27 @@
         <IsPackable>true</IsPackable>
         <PackageId>MathNet.Numerics.FSharp</PackageId>
         <PackageId Condition="'$(StrongName)'=='True'">MathNet.Numerics.FSharp.Signed</PackageId>
-        <VersionPrefix>4.15.0</VersionPrefix>
+        <VersionPrefix>5.0.0</VersionPrefix>
-        <VersionSuffix></VersionSuffix>
+        <VersionSuffix>alpha01</VersionSuffix>
         <Title>Math.NET Numerics for F#$(TitleSuffix)</Title>
         <Description>F# Modules for Math.NET Numerics, the numerical foundation of the Math.NET project, aiming to provide methods and algorithms for numerical computations in science, engineering and every day use. Supports .Net 5.0 or higher, .NET Standard 2.0 and .NET Framework 4.6.1 or higher, on Windows, Linux and Mac.$(DescriptionSuffix)</Description>
-        <PackageReleaseNotes>Precision: Round (with integer part rounding) ~Jon Larborn
+        <PackageReleaseNotes>COMPATIBILITY: net5.0, net48 better supported with explicit builds
-Precision: RoundToMultiple, RoundToPower
+COMPATIBILITY: netstandard1.x, net40, net45 no longer supported
-F#: BigRational.FromDecimal ~Brian Berns</PackageReleaseNotes>
+BREAKING: drop all which was marked as obsolete
+BREAKING: all native provider adapters moved out to separate NuGet packages
+BREAKING: switch many usages of tuples to value tuples (experimental)
+Distributions: Logistic ~Bobby Ingram
+Distributions: Perf: Cauchy avoid duplicate evaluation ~Febin
+Precision: Perf: pre-compute negative powers ~Febin
+Optimizations: Remove static properties in LevenbergMarquardtMinimizer ~Jong Hyun Kim
+Fit.Curve and FindMinimum extended to accept two more parameters
+Series: stable series summation
+Providers: drop managed reference linear algebra provider
+Providers: native providers no longer inherit managed providers, managed now sealed
+Providers: MKL provider compilation switched to Intel oneAPI MKL
+Lots of internal cleanup, leveraging newer language features
+Data: now released always together with Numerics (no longer separate versioning)
+Control.Describe now includes CPU architecture and family identifier if know</PackageReleaseNotes>
         <PackageTags>fsharp F# math numeric statistics probability integration interpolation regression solve fit linear algebra matrix fft</PackageTags>
         <GenerateAssemblyInfo>false</GenerateAssemblyInfo>
         <GeneratePackageOnBuild>false</GeneratePackageOnBuild>

src/Numerics.Tests/Properties/AssemblyInfo.cs

@@ -9,8 +9,8 @@ using MathNet.Numerics.UnitTests;
 [assembly: ComVisible(false)]
 [assembly: Guid("04157581-63f3-447b-a277-83c6e69126a4")]
 
-[assembly: AssemblyVersion("4.15.0.0")]
+[assembly: AssemblyVersion("5.0.0.0")]
-[assembly: AssemblyFileVersion("4.15.0.0")]
+[assembly: AssemblyFileVersion("5.0.0.0")]
-[assembly: AssemblyInformationalVersion("4.15.0")]
+[assembly: AssemblyInformationalVersion("5.0.0-alpha01")]
 
 [assembly: UseLinearAlgebraProvider]

src/Numerics/Numerics.csproj

@@ -8,13 +8,27 @@
     <IsPackable>true</IsPackable>
     <PackageId>MathNet.Numerics</PackageId>
     <PackageId Condition="'$(StrongName)'=='True'">MathNet.Numerics.Signed</PackageId>
-    <VersionPrefix>4.15.0</VersionPrefix>
+    <VersionPrefix>5.0.0</VersionPrefix>
-    <VersionSuffix></VersionSuffix>
+    <VersionSuffix>alpha01</VersionSuffix>
     <Title>Math.NET Numerics$(TitleSuffix)</Title>
     <Description>Math.NET Numerics is the numerical foundation of the Math.NET project, aiming to provide methods and algorithms for numerical computations in science, engineering and every day use. Supports .Net 5.0 or higher, .NET Standard 2.0 and .NET Framework 4.6.1 or higher, on Windows, Linux and Mac.$(DescriptionSuffix)</Description>
-    <PackageReleaseNotes>Precision: Round (with integer part rounding) ~Jon Larborn
+    <PackageReleaseNotes>COMPATIBILITY: net5.0, net48 better supported with explicit builds
-Precision: RoundToMultiple, RoundToPower
+COMPATIBILITY: netstandard1.x, net40, net45 no longer supported
-F#: BigRational.FromDecimal ~Brian Berns</PackageReleaseNotes>
+BREAKING: drop all which was marked as obsolete
+BREAKING: all native provider adapters moved out to separate NuGet packages
+BREAKING: switch many usages of tuples to value tuples (experimental)
+Distributions: Logistic ~Bobby Ingram
+Distributions: Perf: Cauchy avoid duplicate evaluation ~Febin
+Precision: Perf: pre-compute negative powers ~Febin
+Optimizations: Remove static properties in LevenbergMarquardtMinimizer ~Jong Hyun Kim
+Fit.Curve and FindMinimum extended to accept two more parameters
+Series: stable series summation
+Providers: drop managed reference linear algebra provider
+Providers: native providers no longer inherit managed providers, managed now sealed
+Providers: MKL provider compilation switched to Intel oneAPI MKL
+Lots of internal cleanup, leveraging newer language features
+Data: now released always together with Numerics (no longer separate versioning)
+Control.Describe now includes CPU architecture and family identifier if know</PackageReleaseNotes>
     <PackageTags>math numeric statistics probability integration interpolation regression solve fit linear algebra matrix fft</PackageTags>
     <GenerateAssemblyInfo>true</GenerateAssemblyInfo>
     <GeneratePackageOnBuild>false</GeneratePackageOnBuild>

src/Providers.CUDA/Providers.CUDA.csproj

@@ -8,13 +8,27 @@
     <IsPackable>true</IsPackable>
     <PackageId>MathNet.Numerics.Providers.CUDA</PackageId>
     <PackageId Condition="'$(StrongName)'=='True'">MathNet.Numerics.Providers.CUDA.Signed</PackageId>
-    <VersionPrefix>4.15.0</VersionPrefix>
+    <VersionPrefix>5.0.0</VersionPrefix>
-    <VersionSuffix></VersionSuffix>
+    <VersionSuffix>alpha01</VersionSuffix>
     <Title>Math.NET Numerics CUDA Provider$(TitleSuffix)</Title>
     <Description>Math.NET Numerics is the numerical foundation of the Math.NET project, aiming to provide methods and algorithms for numerical computations in science, engineering and every day use.$(DescriptionSuffix)</Description>
-    <PackageReleaseNotes>Precision: Round (with integer part rounding) ~Jon Larborn
+    <PackageReleaseNotes>COMPATIBILITY: net5.0, net48 better supported with explicit builds
-Precision: RoundToMultiple, RoundToPower
+COMPATIBILITY: netstandard1.x, net40, net45 no longer supported
-F#: BigRational.FromDecimal ~Brian Berns</PackageReleaseNotes>
+BREAKING: drop all which was marked as obsolete
+BREAKING: all native provider adapters moved out to separate NuGet packages
+BREAKING: switch many usages of tuples to value tuples (experimental)
+Distributions: Logistic ~Bobby Ingram
+Distributions: Perf: Cauchy avoid duplicate evaluation ~Febin
+Precision: Perf: pre-compute negative powers ~Febin
+Optimizations: Remove static properties in LevenbergMarquardtMinimizer ~Jong Hyun Kim
+Fit.Curve and FindMinimum extended to accept two more parameters
+Series: stable series summation
+Providers: drop managed reference linear algebra provider
+Providers: native providers no longer inherit managed providers, managed now sealed
+Providers: MKL provider compilation switched to Intel oneAPI MKL
+Lots of internal cleanup, leveraging newer language features
+Data: now released always together with Numerics (no longer separate versioning)
+Control.Describe now includes CPU architecture and family identifier if know</PackageReleaseNotes>
     <GenerateAssemblyInfo>true</GenerateAssemblyInfo>
     <GeneratePackageOnBuild>false</GeneratePackageOnBuild>
     <GenerateDocumentationFile>true</GenerateDocumentationFile>

src/Providers.MKL/Providers.MKL.csproj

@@ -8,13 +8,27 @@
     <IsPackable>true</IsPackable>
     <PackageId>MathNet.Numerics.Providers.MKL</PackageId>
     <PackageId Condition="'$(StrongName)'=='True'">MathNet.Numerics.Providers.MKL.Signed</PackageId>
-    <VersionPrefix>4.15.0</VersionPrefix>
+    <VersionPrefix>5.0.0</VersionPrefix>
-    <VersionSuffix></VersionSuffix>
+    <VersionSuffix>alpha01</VersionSuffix>
     <Title>Math.NET Numerics MKL Provider$(TitleSuffix)</Title>
     <Description>Math.NET Numerics is the numerical foundation of the Math.NET project, aiming to provide methods and algorithms for numerical computations in science, engineering and every day use.$(DescriptionSuffix)</Description>
-    <PackageReleaseNotes>Precision: Round (with integer part rounding) ~Jon Larborn
+    <PackageReleaseNotes>COMPATIBILITY: net5.0, net48 better supported with explicit builds
-Precision: RoundToMultiple, RoundToPower
+COMPATIBILITY: netstandard1.x, net40, net45 no longer supported
-F#: BigRational.FromDecimal ~Brian Berns</PackageReleaseNotes>
+BREAKING: drop all which was marked as obsolete
+BREAKING: all native provider adapters moved out to separate NuGet packages
+BREAKING: switch many usages of tuples to value tuples (experimental)
+Distributions: Logistic ~Bobby Ingram
+Distributions: Perf: Cauchy avoid duplicate evaluation ~Febin
+Precision: Perf: pre-compute negative powers ~Febin
+Optimizations: Remove static properties in LevenbergMarquardtMinimizer ~Jong Hyun Kim
+Fit.Curve and FindMinimum extended to accept two more parameters
+Series: stable series summation
+Providers: drop managed reference linear algebra provider
+Providers: native providers no longer inherit managed providers, managed now sealed
+Providers: MKL provider compilation switched to Intel oneAPI MKL
+Lots of internal cleanup, leveraging newer language features
+Data: now released always together with Numerics (no longer separate versioning)
+Control.Describe now includes CPU architecture and family identifier if know</PackageReleaseNotes>
     <GenerateAssemblyInfo>true</GenerateAssemblyInfo>
     <GeneratePackageOnBuild>false</GeneratePackageOnBuild>
     <GenerateDocumentationFile>true</GenerateDocumentationFile>

src/Providers.OpenBLAS/Providers.OpenBLAS.csproj

@@ -8,13 +8,27 @@
     <IsPackable>true</IsPackable>
     <PackageId>MathNet.Numerics.Providers.OpenBLAS</PackageId>
     <PackageId Condition="'$(StrongName)'=='True'">MathNet.Numerics.Providers.OpenBLAS.Signed</PackageId>
-    <VersionPrefix>4.15.0</VersionPrefix>
+    <VersionPrefix>5.0.0</VersionPrefix>
-    <VersionSuffix></VersionSuffix>
+    <VersionSuffix>alpha01</VersionSuffix>
     <Title>Math.NET Numerics OpenBLAS Provider$(TitleSuffix)</Title>
     <Description>Math.NET Numerics is the numerical foundation of the Math.NET project, aiming to provide methods and algorithms for numerical computations in science, engineering and every day use.$(DescriptionSuffix)</Description>
-    <PackageReleaseNotes>Precision: Round (with integer part rounding) ~Jon Larborn
+    <PackageReleaseNotes>COMPATIBILITY: net5.0, net48 better supported with explicit builds
-Precision: RoundToMultiple, RoundToPower
+COMPATIBILITY: netstandard1.x, net40, net45 no longer supported
-F#: BigRational.FromDecimal ~Brian Berns</PackageReleaseNotes>
+BREAKING: drop all which was marked as obsolete
+BREAKING: all native provider adapters moved out to separate NuGet packages
+BREAKING: switch many usages of tuples to value tuples (experimental)
+Distributions: Logistic ~Bobby Ingram
+Distributions: Perf: Cauchy avoid duplicate evaluation ~Febin
+Precision: Perf: pre-compute negative powers ~Febin
+Optimizations: Remove static properties in LevenbergMarquardtMinimizer ~Jong Hyun Kim
+Fit.Curve and FindMinimum extended to accept two more parameters
+Series: stable series summation
+Providers: drop managed reference linear algebra provider
+Providers: native providers no longer inherit managed providers, managed now sealed
+Providers: MKL provider compilation switched to Intel oneAPI MKL
+Lots of internal cleanup, leveraging newer language features
+Data: now released always together with Numerics (no longer separate versioning)
+Control.Describe now includes CPU architecture and family identifier if know</PackageReleaseNotes>
     <GenerateAssemblyInfo>true</GenerateAssemblyInfo>
     <GeneratePackageOnBuild>false</GeneratePackageOnBuild>
     <GenerateDocumentationFile>true</GenerateDocumentationFile>

src/TestData/Properties/AssemblyInfo.cs

@@ -14,5 +14,5 @@ using System.Runtime.InteropServices;
 
 [assembly: Guid("a4a6a08e-5265-4608-a43d-e4f2e210ba2d")]
 
-[assembly: AssemblyVersion("4.15.0.0")]
+[assembly: AssemblyVersion("5.0.0.0")]
-[assembly: AssemblyFileVersion("4.15.0.0")]
+[assembly: AssemblyFileVersion("5.0.0.0")]