axiosengine/axios/Common/Math.cs
2012-03-19 18:57:59 -05:00

638 lines
20 KiB
C#

/*
* Farseer Physics Engine based on Box2D.XNA port:
* Copyright (c) 2010 Ian Qvist
*
* Box2D.XNA port of Box2D:
* Copyright (c) 2009 Brandon Furtwangler, Nathan Furtwangler
*
* Original source Box2D:
* Copyright (c) 2006-2009 Erin Catto http://www.gphysics.com
*
* This software is provided 'as-is', without any express or implied
* warranty. In no event will the authors be held liable for any damages
* arising from the use of this software.
* Permission is granted to anyone to use this software for any purpose,
* including commercial applications, and to alter it and redistribute it
* freely, subject to the following restrictions:
* 1. The origin of this software must not be misrepresented; you must not
* claim that you wrote the original software. If you use this software
* in a product, an acknowledgment in the product documentation would be
* appreciated but is not required.
* 2. Altered source versions must be plainly marked as such, and must not be
* misrepresented as being the original software.
* 3. This notice may not be removed or altered from any source distribution.
*/
using System;
using System.Diagnostics;
using System.Runtime.InteropServices;
using Microsoft.Xna.Framework;
namespace FarseerPhysics.Common
{
public static class MathUtils
{
public static float Cross(Vector2 a, Vector2 b)
{
return a.X * b.Y - a.Y * b.X;
}
public static Vector2 Cross(Vector2 a, float s)
{
return new Vector2(s * a.Y, -s * a.X);
}
public static Vector2 Cross(float s, Vector2 a)
{
return new Vector2(-s * a.Y, s * a.X);
}
public static Vector2 Abs(Vector2 v)
{
return new Vector2(Math.Abs(v.X), Math.Abs(v.Y));
}
public static Vector2 Multiply(ref Mat22 A, Vector2 v)
{
return Multiply(ref A, ref v);
}
public static Vector2 Multiply(ref Mat22 A, ref Vector2 v)
{
return new Vector2(A.Col1.X * v.X + A.Col2.X * v.Y, A.Col1.Y * v.X + A.Col2.Y * v.Y);
}
public static Vector2 MultiplyT(ref Mat22 A, Vector2 v)
{
return MultiplyT(ref A, ref v);
}
public static Vector2 MultiplyT(ref Mat22 A, ref Vector2 v)
{
return new Vector2(v.X * A.Col1.X + v.Y * A.Col1.Y, v.X * A.Col2.X + v.Y * A.Col2.Y);
}
public static Vector2 Multiply(ref Transform T, Vector2 v)
{
return Multiply(ref T, ref v);
}
public static Vector2 Multiply(ref Transform T, ref Vector2 v)
{
return new Vector2(T.Position.X + T.R.Col1.X * v.X + T.R.Col2.X * v.Y,
T.Position.Y + T.R.Col1.Y * v.X + T.R.Col2.Y * v.Y);
}
public static Vector2 MultiplyT(ref Transform T, Vector2 v)
{
return MultiplyT(ref T, ref v);
}
public static Vector2 MultiplyT(ref Transform T, ref Vector2 v)
{
Vector2 tmp = Vector2.Zero;
tmp.X = v.X - T.Position.X;
tmp.Y = v.Y - T.Position.Y;
return MultiplyT(ref T.R, ref tmp);
}
// A^T * B
public static void MultiplyT(ref Mat22 A, ref Mat22 B, out Mat22 C)
{
C = new Mat22();
C.Col1.X = A.Col1.X * B.Col1.X + A.Col1.Y * B.Col1.Y;
C.Col1.Y = A.Col2.X * B.Col1.X + A.Col2.Y * B.Col1.Y;
C.Col2.X = A.Col1.X * B.Col2.X + A.Col1.Y * B.Col2.Y;
C.Col2.Y = A.Col2.X * B.Col2.X + A.Col2.Y * B.Col2.Y;
}
// v2 = A.R' * (B.R * v1 + B.p - A.p) = (A.R' * B.R) * v1 + (B.p - A.p)
public static void MultiplyT(ref Transform A, ref Transform B, out Transform C)
{
C = new Transform();
MultiplyT(ref A.R, ref B.R, out C.R);
C.Position.X = B.Position.X - A.Position.X;
C.Position.Y = B.Position.Y - A.Position.Y;
}
public static void Swap<T>(ref T a, ref T b)
{
T tmp = a;
a = b;
b = tmp;
}
/// <summary>
/// This function is used to ensure that a floating point number is
/// not a NaN or infinity.
/// </summary>
/// <param name="x">The x.</param>
/// <returns>
/// <c>true</c> if the specified x is valid; otherwise, <c>false</c>.
/// </returns>
public static bool IsValid(float x)
{
if (float.IsNaN(x))
{
// NaN.
return false;
}
return !float.IsInfinity(x);
}
public static bool IsValid(this Vector2 x)
{
return IsValid(x.X) && IsValid(x.Y);
}
/// <summary>
/// This is a approximate yet fast inverse square-root.
/// </summary>
/// <param name="x">The x.</param>
/// <returns></returns>
public static float InvSqrt(float x)
{
FloatConverter convert = new FloatConverter();
convert.x = x;
float xhalf = 0.5f * x;
convert.i = 0x5f3759df - (convert.i >> 1);
x = convert.x;
x = x * (1.5f - xhalf * x * x);
return x;
}
public static int Clamp(int a, int low, int high)
{
return Math.Max(low, Math.Min(a, high));
}
public static float Clamp(float a, float low, float high)
{
return Math.Max(low, Math.Min(a, high));
}
public static Vector2 Clamp(Vector2 a, Vector2 low, Vector2 high)
{
return Vector2.Max(low, Vector2.Min(a, high));
}
public static void Cross(ref Vector2 a, ref Vector2 b, out float c)
{
c = a.X * b.Y - a.Y * b.X;
}
/// <summary>
/// Return the angle between two vectors on a plane
/// The angle is from vector 1 to vector 2, positive anticlockwise
/// The result is between -pi -> pi
/// </summary>
public static double VectorAngle(ref Vector2 p1, ref Vector2 p2)
{
double theta1 = Math.Atan2(p1.Y, p1.X);
double theta2 = Math.Atan2(p2.Y, p2.X);
double dtheta = theta2 - theta1;
while (dtheta > Math.PI)
dtheta -= (2 * Math.PI);
while (dtheta < -Math.PI)
dtheta += (2 * Math.PI);
return (dtheta);
}
public static double VectorAngle(Vector2 p1, Vector2 p2)
{
return VectorAngle(ref p1, ref p2);
}
/// <summary>
/// Returns a positive number if c is to the left of the line going from a to b.
/// </summary>
/// <returns>Positive number if point is left, negative if point is right,
/// and 0 if points are collinear.</returns>
public static float Area(Vector2 a, Vector2 b, Vector2 c)
{
return Area(ref a, ref b, ref c);
}
/// <summary>
/// Returns a positive number if c is to the left of the line going from a to b.
/// </summary>
/// <returns>Positive number if point is left, negative if point is right,
/// and 0 if points are collinear.</returns>
public static float Area(ref Vector2 a, ref Vector2 b, ref Vector2 c)
{
return a.X * (b.Y - c.Y) + b.X * (c.Y - a.Y) + c.X * (a.Y - b.Y);
}
/// <summary>
/// Determines if three vertices are collinear (ie. on a straight line)
/// </summary>
/// <param name="a">First vertex</param>
/// <param name="b">Second vertex</param>
/// <param name="c">Third vertex</param>
/// <returns></returns>
public static bool Collinear(ref Vector2 a, ref Vector2 b, ref Vector2 c)
{
return Collinear(ref a, ref b, ref c, 0);
}
public static bool Collinear(ref Vector2 a, ref Vector2 b, ref Vector2 c, float tolerance)
{
return FloatInRange(Area(ref a, ref b, ref c), -tolerance, tolerance);
}
public static void Cross(float s, ref Vector2 a, out Vector2 b)
{
b = new Vector2(-s * a.Y, s * a.X);
}
public static bool FloatEquals(float value1, float value2)
{
return Math.Abs(value1 - value2) <= Settings.Epsilon;
}
/// <summary>
/// Checks if a floating point Value is equal to another,
/// within a certain tolerance.
/// </summary>
/// <param name="value1">The first floating point Value.</param>
/// <param name="value2">The second floating point Value.</param>
/// <param name="delta">The floating point tolerance.</param>
/// <returns>True if the values are "equal", false otherwise.</returns>
public static bool FloatEquals(float value1, float value2, float delta)
{
return FloatInRange(value1, value2 - delta, value2 + delta);
}
/// <summary>
/// Checks if a floating point Value is within a specified
/// range of values (inclusive).
/// </summary>
/// <param name="value">The Value to check.</param>
/// <param name="min">The minimum Value.</param>
/// <param name="max">The maximum Value.</param>
/// <returns>True if the Value is within the range specified,
/// false otherwise.</returns>
public static bool FloatInRange(float value, float min, float max)
{
return (value >= min && value <= max);
}
#region Nested type: FloatConverter
[StructLayout(LayoutKind.Explicit)]
private struct FloatConverter
{
[FieldOffset(0)]
public float x;
[FieldOffset(0)]
public int i;
}
#endregion
}
/// <summary>
/// A 2-by-2 matrix. Stored in column-major order.
/// </summary>
public struct Mat22
{
public Vector2 Col1, Col2;
/// <summary>
/// Construct this matrix using columns.
/// </summary>
/// <param name="c1">The c1.</param>
/// <param name="c2">The c2.</param>
public Mat22(Vector2 c1, Vector2 c2)
{
Col1 = c1;
Col2 = c2;
}
/// <summary>
/// Construct this matrix using scalars.
/// </summary>
/// <param name="a11">The a11.</param>
/// <param name="a12">The a12.</param>
/// <param name="a21">The a21.</param>
/// <param name="a22">The a22.</param>
public Mat22(float a11, float a12, float a21, float a22)
{
Col1 = new Vector2(a11, a21);
Col2 = new Vector2(a12, a22);
}
/// <summary>
/// Construct this matrix using an angle. This matrix becomes
/// an orthonormal rotation matrix.
/// </summary>
/// <param name="angle">The angle.</param>
public Mat22(float angle)
{
// TODO_ERIN compute sin+cos together.
float c = (float)Math.Cos(angle), s = (float)Math.Sin(angle);
Col1 = new Vector2(c, s);
Col2 = new Vector2(-s, c);
}
/// <summary>
/// Extract the angle from this matrix (assumed to be
/// a rotation matrix).
/// </summary>
/// <value></value>
public float Angle
{
get { return (float)Math.Atan2(Col1.Y, Col1.X); }
}
public Mat22 Inverse
{
get
{
float a = Col1.X, b = Col2.X, c = Col1.Y, d = Col2.Y;
float det = a * d - b * c;
if (det != 0.0f)
{
det = 1.0f / det;
}
Mat22 result = new Mat22();
result.Col1.X = det * d;
result.Col1.Y = -det * c;
result.Col2.X = -det * b;
result.Col2.Y = det * a;
return result;
}
}
/// <summary>
/// Initialize this matrix using columns.
/// </summary>
/// <param name="c1">The c1.</param>
/// <param name="c2">The c2.</param>
public void Set(Vector2 c1, Vector2 c2)
{
Col1 = c1;
Col2 = c2;
}
/// <summary>
/// Initialize this matrix using an angle. This matrix becomes
/// an orthonormal rotation matrix.
/// </summary>
/// <param name="angle">The angle.</param>
public void Set(float angle)
{
float c = (float)Math.Cos(angle), s = (float)Math.Sin(angle);
Col1.X = c;
Col2.X = -s;
Col1.Y = s;
Col2.Y = c;
}
/// <summary>
/// Set this to the identity matrix.
/// </summary>
public void SetIdentity()
{
Col1.X = 1.0f;
Col2.X = 0.0f;
Col1.Y = 0.0f;
Col2.Y = 1.0f;
}
/// <summary>
/// Set this matrix to all zeros.
/// </summary>
public void SetZero()
{
Col1.X = 0.0f;
Col2.X = 0.0f;
Col1.Y = 0.0f;
Col2.Y = 0.0f;
}
/// <summary>
/// Solve A * x = b, where b is a column vector. This is more efficient
/// than computing the inverse in one-shot cases.
/// </summary>
/// <param name="b">The b.</param>
/// <returns></returns>
public Vector2 Solve(Vector2 b)
{
float a11 = Col1.X, a12 = Col2.X, a21 = Col1.Y, a22 = Col2.Y;
float det = a11 * a22 - a12 * a21;
if (det != 0.0f)
{
det = 1.0f / det;
}
return new Vector2(det * (a22 * b.X - a12 * b.Y), det * (a11 * b.Y - a21 * b.X));
}
public static void Add(ref Mat22 A, ref Mat22 B, out Mat22 R)
{
R.Col1 = A.Col1 + B.Col1;
R.Col2 = A.Col2 + B.Col2;
}
}
/// <summary>
/// A 3-by-3 matrix. Stored in column-major order.
/// </summary>
public struct Mat33
{
public Vector3 Col1, Col2, Col3;
/// <summary>
/// Construct this matrix using columns.
/// </summary>
/// <param name="c1">The c1.</param>
/// <param name="c2">The c2.</param>
/// <param name="c3">The c3.</param>
public Mat33(Vector3 c1, Vector3 c2, Vector3 c3)
{
Col1 = c1;
Col2 = c2;
Col3 = c3;
}
/// <summary>
/// Set this matrix to all zeros.
/// </summary>
public void SetZero()
{
Col1 = Vector3.Zero;
Col2 = Vector3.Zero;
Col3 = Vector3.Zero;
}
/// <summary>
/// Solve A * x = b, where b is a column vector. This is more efficient
/// than computing the inverse in one-shot cases.
/// </summary>
/// <param name="b">The b.</param>
/// <returns></returns>
public Vector3 Solve33(Vector3 b)
{
float det = Vector3.Dot(Col1, Vector3.Cross(Col2, Col3));
if (det != 0.0f)
{
det = 1.0f / det;
}
return new Vector3(det * Vector3.Dot(b, Vector3.Cross(Col2, Col3)),
det * Vector3.Dot(Col1, Vector3.Cross(b, Col3)),
det * Vector3.Dot(Col1, Vector3.Cross(Col2, b)));
}
/// <summary>
/// Solve A * x = b, where b is a column vector. This is more efficient
/// than computing the inverse in one-shot cases. Solve only the upper
/// 2-by-2 matrix equation.
/// </summary>
/// <param name="b">The b.</param>
/// <returns></returns>
public Vector2 Solve22(Vector2 b)
{
float a11 = Col1.X, a12 = Col2.X, a21 = Col1.Y, a22 = Col2.Y;
float det = a11 * a22 - a12 * a21;
if (det != 0.0f)
{
det = 1.0f / det;
}
return new Vector2(det * (a22 * b.X - a12 * b.Y), det * (a11 * b.Y - a21 * b.X));
}
}
/// <summary>
/// A transform contains translation and rotation. It is used to represent
/// the position and orientation of rigid frames.
/// </summary>
public struct Transform
{
public Vector2 Position;
public Mat22 R;
/// <summary>
/// Initialize using a position vector and a rotation matrix.
/// </summary>
/// <param name="position">The position.</param>
/// <param name="r">The r.</param>
public Transform(ref Vector2 position, ref Mat22 r)
{
Position = position;
R = r;
}
/// <summary>
/// Calculate the angle that the rotation matrix represents.
/// </summary>
/// <value></value>
public float Angle
{
get { return (float)Math.Atan2(R.Col1.Y, R.Col1.X); }
}
/// <summary>
/// Set this to the identity transform.
/// </summary>
public void SetIdentity()
{
Position = Vector2.Zero;
R.SetIdentity();
}
/// <summary>
/// Set this based on the position and angle.
/// </summary>
/// <param name="position">The position.</param>
/// <param name="angle">The angle.</param>
public void Set(Vector2 position, float angle)
{
Position = position;
R.Set(angle);
}
}
/// <summary>
/// This describes the motion of a body/shape for TOI computation.
/// Shapes are defined with respect to the body origin, which may
/// no coincide with the center of mass. However, to support dynamics
/// we must interpolate the center of mass position.
/// </summary>
public struct Sweep
{
/// <summary>
/// World angles
/// </summary>
public float A;
public float A0;
/// <summary>
/// Fraction of the current time step in the range [0,1]
/// c0 and a0 are the positions at alpha0.
/// </summary>
public float Alpha0;
/// <summary>
/// Center world positions
/// </summary>
public Vector2 C;
public Vector2 C0;
/// <summary>
/// Local center of mass position
/// </summary>
public Vector2 LocalCenter;
/// <summary>
/// Get the interpolated transform at a specific time.
/// </summary>
/// <param name="xf">The transform.</param>
/// <param name="beta">beta is a factor in [0,1], where 0 indicates alpha0.</param>
public void GetTransform(out Transform xf, float beta)
{
xf = new Transform();
xf.Position.X = (1.0f - beta) * C0.X + beta * C.X;
xf.Position.Y = (1.0f - beta) * C0.Y + beta * C.Y;
float angle = (1.0f - beta) * A0 + beta * A;
xf.R.Set(angle);
// Shift to origin
xf.Position -= MathUtils.Multiply(ref xf.R, ref LocalCenter);
}
/// <summary>
/// Advance the sweep forward, yielding a new initial state.
/// </summary>
/// <param name="alpha">new initial time..</param>
public void Advance(float alpha)
{
Debug.Assert(Alpha0 < 1.0f);
float beta = (alpha - Alpha0) / (1.0f - Alpha0);
C0.X = (1.0f - beta) * C0.X + beta * C.X;
C0.Y = (1.0f - beta) * C0.Y + beta * C.Y;
A0 = (1.0f - beta) * A0 + beta * A;
Alpha0 = alpha;
}
/// <summary>
/// Normalize the angles.
/// </summary>
public void Normalize()
{
float d = MathHelper.TwoPi * (float)Math.Floor(A0 / MathHelper.TwoPi);
A0 -= d;
A -= d;
}
}
}