Source C# 11 Generic Math: Writing Universal Vector Math with INumber<T> - UnreliableCode Forums

MathMagician

Vector Math & Physics

MEMBER

Rep: 219

Join Date: Nov 2024

Posts: 8

Thanks: 19

2y ago · Nov 20, 2023 12:45 PM

#1

In older C#, you had to write separate classes for

CODE

Vector3f

(floats),

CODE

Vector3d

(doubles), and

CODE

Vector3i

(integers) because operators (+, -, *, /) could not be used in generic constraints.

C# 11 static abstract interfaces in Generic Math (
CODE
INumber<T>
):

CSHARP

using System.Numerics;

public record struct Vector3<T>(T X, T Y, T Z) where T : INumber<T>, IRootFunctions<T>
{
    public static Vector3<T> operator +(Vector3<T> a, Vector3<T> b) =>
        new(a.X + b.X, a.Y + b.Y, a.Z + b.Z);

    public static Vector3<T> operator *(Vector3<T> a, T scalar) =>
        new(a.X * scalar, a.Y * scalar, a.Z * scalar);

    public T LengthSquared() =>
        (X * X) + (Y * Y) + (Z * Z);

    public T Length() =>
        T.Sqrt(LengthSquared()); // Generic square root!
}

Now a single struct definition works seamlessly with

CODE

float

,

CODE

double

,

CODE

Half

, or

CODE

decimal

with zero boxing overhead!

In older C#, you had to write separate classes for [code]Vector3f[/code] (floats), [code]Vector3d[/code] (doubles), and [code]Vector3i[/code] (integers) because operators (+, -, *, /) could not be used in generic constraints.

[b]C# 11 static abstract interfaces in Generic Math ([code]INumber<T>[/code]):[/b]

[code=csharp]using System.Numerics;

public record struct Vector3<T>(T X, T Y, T Z) where T : INumber<T>, IRootFunctions<T>
{
    public static Vector3<T> operator +(Vector3<T> a, Vector3<T> b) =>
        new(a.X + b.X, a.Y + b.Y, a.Z + b.Z);

public static Vector3<T> operator *(Vector3<T> a, T scalar) =>
        new(a.X * scalar, a.Y * scalar, a.Z * scalar);

public T LengthSquared() =>
        (X * X) + (Y * Y) + (Z * Z);

public T Length() =>
        T.Sqrt(LengthSquared()); // Generic square root!
}[/code]

Now a single struct definition works seamlessly with [code]float[/code], [code]double[/code], [code]Half[/code], or [code]decimal[/code] with zero boxing overhead!

MathMagician · Vector Math & Physics
Quaternions, transformation matrices, and physics simulation...

The following users thanked MathMagician for this post:

GenericGuru

Templates & Generics

MEMBER

Rep: 154

Join Date: Apr 2023

Posts: 8

Thanks: 35

2y ago · Nov 20, 2023 4:14 PM

#2

Static abstract interface members (

CODE

static abstract T operator +

) was one of the biggest additions in .NET 7. The JIT specializes and inlines the math operations directly into native CPU registers with zero virtual dispatch!

GenericGuru · Templates & Generics
Curiously Recurring Template Pattern (CRTP), C++ concepts, a...

CSharpNinja

Senior .NET Developer

VIP

Rep: 329

Join Date: Jan 2020

Posts: 16

Thanks: 75

2y ago · Nov 21, 2023 12:14 AM

#3

We converted our procedural terrain generator to generic math and could switch between

CODE

float

for real-time rendering and

CODE

double

for 64-bit planetary coordinates without duplicating 2,000 lines of math code.

CSharpNinja · Senior .NET Developer
C# enthusiast, building distributed backend services and hig...

UnreliableCode Community

C# 11 Generic Math: Writing Universal Vector Math with INumber<T>

Create Forum Thread

Add Reputation