I came across a situation doing some advanced collision detection, where I needed to calculate the roots of a quartic function.
I wrote a function that seems to work fine using Ferrari's general solution as seen here: http://en.wikipedia.org/wiki/Quartic_function#Ferrari.27s_solution.
Here's my function:
private function solveQuartic(A:Number, B:Number, C:Number, D:Number, E:Number):Array{
// For paramters: Ax^4 + Bx^3 + Cx^2 + Dx + E
var solution:Array = new Array(4);
// Using Ferrari's formula: http://en.wikipedia.org/wiki/Quartic_function#Ferrari.27s_solution
var Alpha:Number = ((-3 * (B * B)) / (8 * (A * A))) + (C / A);
var Beta:Number = ((B * B * B) / (8 * A * A * A)) - ((B * C) / (2 * A * A)) + (D / A);
var Gamma:Number = ((-3 * B * B * B * B) / (256 * A * A * A * A)) + ((C * B * B) / (16 * A * A * A)) - ((B * D) / (4 * A * A)) + (E / A);
var P:Number = ((-1 * Alpha * Alpha) / 12) - Gamma;
var Q:Number = ((-1 * Alpha * Alpha * Alpha) / 108) + ((Alpha * Gamma) / 3) - ((Beta * Beta) / 8);
var PreRoot1:Number = ((Q * Q) / 4) + ((P * P * P) / 27);
var R:ComplexNumber = ComplexNumber.add(new ComplexNumber((-1 * Q) / 2), ComplexNumber.sqrt(new ComplexNumber(PreRoot1)));
var U:ComplexNumber = ComplexNumber.pow(R, 1/3);
var preY1:Number = (-5 / 6) * Alpha;
var RedundantY:ComplexNumber = ComplexNumber.add(new ComplexNumber(preY1), U);
var Y:ComplexNumber;
if(U.isZero()){
var preY2:ComplexNumber = ComplexNumber.pow(new ComplexNumber(Q), 1/3);
Y = ComplexNumber.subtract(RedundantY, preY2);
} else{
var preY3:ComplexNumber = ComplexNumber.multiply(new ComplexNumber(3), U);
var preY4:ComplexNumber = ComplexNumber.divide(new ComplexNumber(P), preY3);
Y = ComplexNumber.subtract(RedundantY, preY4);
}
var W:ComplexNumber = ComplexNumber.sqrt(ComplexNumber.add(new ComplexNumber(Alpha), ComplexNumber.multiply(new ComplexNumber(2), Y)));
var Two:ComplexNumber = new ComplexNumber(2);
var NegativeOne:ComplexNumber = new ComplexNumber(-1);
var NegativeBOverFourA:ComplexNumber = new ComplexNumber((-1 * B) / (4 * A));
var NegativeW:ComplexNumber = ComplexNumber.multiply(W, NegativeOne);
var ThreeAlphaPlusTwoY:ComplexNumber = ComplexNumber.add(new ComplexNumber(3 * Alpha), ComplexNumber.multiply(new ComplexNumber(2), Y));
var TwoBetaOverW:ComplexNumber = ComplexNumber.divide(new ComplexNumber(2 * Beta), W);
solution["root1"] = ComplexNumber.add(NegativeBOverFourA, ComplexNumber.divide(ComplexNumber.add(W, ComplexNumber.sqrt(ComplexNumber.multiply(NegativeOne, ComplexNumber.add(ThreeAlphaPlusTwoY, TwoBetaOverW)))), Two));
solution["root2"] = ComplexNumber.add(NegativeBOverFourA, ComplexNumber.divide(ComplexNumber.subtract(NegativeW, ComplexNumber.sqrt(ComplexNumber.multiply(NegativeOne, ComplexNumber.subtract(ThreeAlphaPlusTwoY, TwoBetaOverW)))), Two));
solution["root3"] = ComplexNumber.add(NegativeBOverFourA, ComplexNumber.divide(ComplexNumber.subtract(W, ComplexNumber.sqrt(ComplexNumber.multiply(NegativeOne, ComplexNumber.add(ThreeAlphaPlusTwoY, TwoBetaOverW)))), Two));
solution["root4"] = ComplexNumber.add(NegativeBOverFourA, ComplexNumber.divide(ComplexNumber.add(NegativeW, ComplexNumber.sqrt(ComplexNumber.multiply(NegativeOne, ComplexNumber.subtract(ThreeAlphaPlusTwoY, TwoBetaOverW)))), Two));
return solution;
}
The only issue is that I seem to get a few exceptions. Most notably when I have two real roots, and two imaginary roots.
For example, this equation:
y = 0.9604000000000001x^4 - 5.997600000000001x^3 + 13.951750054511718x^2 - 14.326264455924333x + 5.474214401412618
Returns the roots:
1.7820304835380467 + 0i
1.34041662585388 + 0i
1.3404185025061823 + 0i
1.7820323472855648 + 0i
If I graph that particular equation, I can see that the actual roots are closer to 1.2 and 2.9 (approximately). I can't dismiss the four incorrect roots as random, because they're actually two of the roots for the equation's first derivative:
y = 3.8416x^3 - 17.9928x^2 + 27.9035001x - 14.326264455924333
Keep in mind that I'm not actually looking for the specific roots to the equation I posted. My question is whether there's some sort of special case that I'm not taking into consideration.
Any ideas?
For finding roots of polynomials of degree >= 3, I've always had better results using Jenkins-Traub ( http://en.wikipedia.org/wiki/Jenkins-Traub_algorithm ) than explicit formulas.
I do not know why Ferrari's solution does not work, but I tried to use the standard numerical method (create a companion matrix and compute its eigenvalues), and I obtain the correct solution, i.e., two real roots at 1.2 and 1.9.
This method is not for the faint of heart. After constructing the companion matrix of the polynomial, you run the QR algorithm to find the eigenvalues of that matrix. Those are the zeroes of the polynomial.
I suggest you to use an existing implementation of the QR algorithm since a good deal of it is closer to kitchen recipe than algorithmics. But it is, I believe, the most widely used algorithm to compute eigenvalues, and thereby, roots of polynomials.
You can see my answer to a related question. I support the view of Olivier: the way to go may just be the companion matrix / eigenvalue approach (very stable, simple, reliable, and fast).
Edit
I guess it does not hurt if I reproduce the answer here, for convenience:
The numerical solution for doing this many times in a reliable, stable manner, involve: (1) Form the companion matrix, (2) find the eigenvalues of the companion matrix.
You may think this is a harder problem to solve than the original one, but this is how the solution is implemented in most production code (say, Matlab).
For the polynomial:
p(t) = c0 + c1 * t + c2 * t^2 + t^3
the companion matrix is:
[[0 0 -c0],[1 0 -c1],[0 1 -c2]]
Find the eigenvalues of such matrix; they correspond to the roots of the original polynomial.
For doing this very fast, download the singular value subroutines from LAPACK, compile them, and link them to your code. Do this in parallel if you have too many (say, about a million) sets of coefficients. You could use QR decomposition, or any other stable methodology for computing eigenvalues (see the Wikipedia entry on "matrix eigenvalues").
Notice that the coefficient of t^3 is one, if this is not the case in your polynomials, you will have to divide the whole thing by the coefficient and then proceed.
Good luck.
Edit: Numpy and octave also depend on this methodology for computing the roots of polynomials. See, for instance, this link.
The other answers are good and sound advice. However, recalling my experience with the implementation of Ferrari's method in Forth, I think your wrong results are probably caused by 1. wrong implementation of the necessary and rather tricky sign combinations, 2. not realizing yet that ".. == beta" in floating-point should become "abs(.. - beta) < eps, 3. not yet having found out that there are other square roots in the code that may return complex solutions.
For this particular problem my Forth code in diagnostic mode returns:
x1 = 1.5612244897959360787072371026316680470492e+0000 -1.6542769593216835969789894020584464029664e-0001 i
--> -4.2123274051525879873007970023884313331788e-0054 3.4544674220377778501545407451201598284464e-0077 i
x2 = 1.5612244897959360787072371026316680470492e+0000 1.6542769593216835969789894020584464029664e-0001 i
--> -4.2123274051525879873007970023884313331788e-0054 -3.4544674220377778501545407451201598284464e-0077 i
x3 = 1.2078440724224197532447709413299479764843e+0000 0.0000000000000000000000000000000000000000e-0001 i
--> -4.2123274051525879873010733597821943554068e-0054 0.0000000000000000000000000000000000000000e-0001 i
x4 = 1.9146049071693819497220585618954851525216e+0000 -0.0000000000000000000000000000000000000000e-0001 i
--> -4.2123274051525879873013497171759573776348e-0054 0.0000000000000000000000000000000000000000e-0001 i
The text after "-->" follows from backsubstituting the root into the original equation.
For reference, here are Mathematica/Alpha's results to the highest possible precision I managed to set it:
Mathematica:
x1 = 1.20784407242
x2 = 1.91460490717
x3 = 1.56122449 - 0.16542770 i
x4 = 1.56122449 + 0.16542770 i
A good alternative to the methods already mentioned is the TOMS Algorithm 326, which is based on the paper "Roots of Low Order Polynomials" by Terence R.F.Nonweiler CACM (Apr 1968).
This is an algebraic solution to 3rd and 4th order polynomials that is reasonably compact, fast, and quite accurate. It is much simpler than Jenkins Traub.
Be warned however that the TOMS code doesn't work all that well.
This Iowa Hills Root Solver page has code for a Quartic / Cubic root finder that is a bit more refined. It also has a Jenkins Traub type root finder.
Related
I have a recursive function f(n) with time complexicity
O(f(n)) = O(combination(n, n/2) * f(n/2)^2)
where combination(a, b) means combination nuber a above b.
I tried to simplify it, but don't have enough mathematical skills. The only thing that I foud out is that
combination(n,n/2) = 2^n * (gamma(n/2 + 1/2)/(sqrt(1/2) * gamma(n/2 + 1)))
I don't have experience with calculation of complexities but this seems to me to be of order n!. At least for calculation of special case with n=2^i. With integers combination(n, n/2) is equal to n!/((n/2)!)^2.
f(2^i) = (2^i)! / ((2^(i-1))!)^2 * f(2^(i-1))^2
= (2^i)! / ((2^(i-1))!)^2 * (2^(i-1))! / ((2^(i-2))!)^2)^2 * f(2^(i-2))^4
= (2^i)! / ((2^(i-2))!)^4 * f(2^(i-2))^4
...
= (2^i)! / (1!)^(2^i) = n!
I have allready solved it in this thread on satck exchange:
https://math.stackexchange.com/questions/2642848/time-complexicity-of-recursive-function?noredirect=1#comment5458208_2642848
I tried to implement bessel function using that formula, this is the code:
function result=Bessel(num);
if num==0
result=bessel(0,1);
elseif num==1
result=bessel(1,1);
else
result=2*(num-1)*Bessel(num-1)-Bessel(num-2);
end;
But if I use MATLAB's bessel function to compare it with this one, I get too high different values.
For example if I type Bessel(20) it gives me 3.1689e+005 as result, if instead I type bessel(20,1) it gives me 3.8735e-025 , a totally different result.
such recurrence relations are nice in mathematics but numerically unstable when implementing algorithms using limited precision representations of floating-point numbers.
Consider the following comparison:
x = 0:20;
y1 = arrayfun(#(n)besselj(n,1), x); %# builtin function
y2 = arrayfun(#Bessel, x); %# your function
semilogy(x,y1, x,y2), grid on
legend('besselj','Bessel')
title('J_\nu(z)'), xlabel('\nu'), ylabel('log scale')
So you can see how the computed values start to differ significantly after 9.
According to MATLAB:
BESSELJ uses a MEX interface to a Fortran library by D. E. Amos.
and gives the following as references for their implementation:
D. E. Amos, "A subroutine package for Bessel functions of a complex
argument and nonnegative order", Sandia National Laboratory Report,
SAND85-1018, May, 1985.
D. E. Amos, "A portable package for Bessel functions of a complex
argument and nonnegative order", Trans. Math. Software, 1986.
The forward recurrence relation you are using is not stable. To see why, consider that the values of BesselJ(n,x) become smaller and smaller by about a factor 1/2n. You can see this by looking at the first term of the Taylor series for J.
So, what you're doing is subtracting a large number from a multiple of a somewhat smaller number to get an even smaller number. Numerically, that's not going to work well.
Look at it this way. We know the result is of the order of 10^-25. You start out with numbers that are of the order of 1. So in order to get even one accurate digit out of this, we have to know the first two numbers with at least 25 digits precision. We clearly don't, and the recurrence actually diverges.
Using the same recurrence relation to go backwards, from high orders to low orders, is stable. When you start with correct values for J(20,1) and J(19,1), you can calculate all orders down to 0 with full accuracy as well. Why does this work? Because now the numbers are getting larger in each step. You're subtracting a very small number from an exact multiple of a larger number to get an even larger number.
You can just modify the code below which is for the Spherical bessel function. It is well tested and works for all arguments and order range. I am sorry it is in C#
public static Complex bessel(int n, Complex z)
{
if (n == 0) return sin(z) / z;
if (n == 1) return sin(z) / (z * z) - cos(z) / z;
if (n <= System.Math.Abs(z.real))
{
Complex h0 = bessel(0, z);
Complex h1 = bessel(1, z);
Complex ret = 0;
for (int i = 2; i <= n; i++)
{
ret = (2 * i - 1) / z * h1 - h0;
h0 = h1;
h1 = ret;
if (double.IsInfinity(ret.real) || double.IsInfinity(ret.imag)) return double.PositiveInfinity;
}
return ret;
}
else
{
double u = 2.0 * abs(z.real) / (2 * n + 1);
double a = 0.1;
double b = 0.175;
int v = n - (int)System.Math.Ceiling((System.Math.Log(0.5e-16 * (a + b * u * (2 - System.Math.Pow(u, 2)) / (1 - System.Math.Pow(u, 2))), 2)));
Complex ret = 0;
while (v > n - 1)
{
ret = z / (2 * v + 1.0 - z * ret);
v = v - 1;
}
Complex jnM1 = ret;
while (v > 0)
{
ret = z / (2 * v + 1.0 - z * ret);
jnM1 = jnM1 * ret;
v = v - 1;
}
return jnM1 * sin(z) / z;
}
}
I'm using Physics for Games Programmers to develop a simple physics-based game.
I need to compute the resulting velocities for two spheres after an elastic collision. The book example for this in Chapter 6 assumes that the 2nd sphere is stationary, and so some of the equations are simplified to 0. I need the math to work when both bodies are in motion.
I've tried to convert the book's example to code, and puzzle out what should happen for the second sphere's line of action and normal- V2p and V2n. My code sort of works, but occasionally the velocities suddenly speed up and bounce out of control. Clearly there's something wrong with my math.
Here's what I'm using. The code is in Java, "s1" and "s2" are the spheres.
double e = 1d;
// distance of sphere centers
double dX = s2.getCenterX() - s1.getCenterX();
double dY = s2.getCenterY() - s1.getCenterY();
double tangent = dY / dX;
double angle = Math.atan(tangent);
// v1 line of action
double v1p = s1.getVelocityX() * Math.cos(angle) + s1.getVelocityY() * Math.sin(angle);
// v1 normal
double v1n = -s1.getVelocityX() * Math.sin(angle) + s1.getVelocityY() * Math.cos(angle);
// v2 line of action
double v2p = s2.getVelocityX() * Math.cos(angle) + s2.getVelocityY() * Math.sin(angle);
// v2 normal
double v2n = -s2.getVelocityX() * Math.sin(angle) + s2.getVelocityY() * Math.cos(angle);
double v1massScale = (s1.getMass() - (e * s2.getMass())) / (s1.getMass() + s2.getMass());
double v2massScale = ((1 + e) * s1.getMass()) / (s1.getMass() + s2.getMass());
// compute post-collision velocities
double v1pPrime = v1massScale * v1p + v2massScale * v2p;
double v2pPrime = v2massScale * v1p + v1massScale * v2p;
// rotate back to normal
double v1xPrime = v1pPrime * Math.cos(angle) - v1n * Math.sin(angle);
double v1yPrime = v1pPrime * Math.sin(angle) + v1n * Math.cos(angle);
double v2xPrime = v2pPrime * Math.cos(angle) - v2n * Math.sin(angle);
double v2yPrime = v2pPrime * Math.sin(angle) + v2n * Math.cos(angle);
It might be the Math.atan(y/x). You need to handle some special cases, and you usually want to use Math.atan2(y,x) instead.
See the Math.atan() and Math.atan2() docs.
It's a surprisingly difficult problem if you don't know much about physics or math.
You need to know about Newton's laws of motion. These are couple differential equations, so you'll need to know how to solve those.
The problem is easier if you assume certain things to be unimportant for the behavior you're interested in: rigid versus deformable spheres, friction, and other factors.
The answer depends a lot on what you'd like to do.
A quick glance at the code you posted suggests that the best you can hope for is that you have a correct but less than optimal implementation. It's likely that you fail to understand the physics and mathematics sufficiently well to solve it on your own.
I'd Google for a physics engine in the language of your choice and use an idealization that someone else has already coded.
Okay, this all takes place in a nice and simple 2D world... :)
Suppose I have a static object A at position Apos, and a linearly moving object B at Bpos with bVelocity, and an ammo round with velocity Avelocity...
How would I find out the angle that A has to shoot, to hit B, taking into account B's linear velocity and the speed of A's ammo ?
Right now the aim's at the current position of the object, which means that by the time my projectile gets there the unit has moved on to safer positions :)
I wrote an aiming subroutine for xtank a while back. I'll try to lay out how I did it.
Disclaimer: I may have made one or more silly mistakes anywhere in here; I'm just trying to reconstruct the reasoning with my rusty math skills. However, I'll cut to the chase first, since this is a programming Q&A instead of a math class :-)
How to do it
It boils down to solving a quadratic equation of the form:
a * sqr(x) + b * x + c == 0
Note that by sqr I mean square, as opposed to square root. Use the following values:
a := sqr(target.velocityX) + sqr(target.velocityY) - sqr(projectile_speed)
b := 2 * (target.velocityX * (target.startX - cannon.X)
+ target.velocityY * (target.startY - cannon.Y))
c := sqr(target.startX - cannon.X) + sqr(target.startY - cannon.Y)
Now we can look at the discriminant to determine if we have a possible solution.
disc := sqr(b) - 4 * a * c
If the discriminant is less than 0, forget about hitting your target -- your projectile can never get there in time. Otherwise, look at two candidate solutions:
t1 := (-b + sqrt(disc)) / (2 * a)
t2 := (-b - sqrt(disc)) / (2 * a)
Note that if disc == 0 then t1 and t2 are equal.
If there are no other considerations such as intervening obstacles, simply choose the smaller positive value. (Negative t values would require firing backward in time to use!)
Substitute the chosen t value back into the target's position equations to get the coordinates of the leading point you should be aiming at:
aim.X := t * target.velocityX + target.startX
aim.Y := t * target.velocityY + target.startY
Derivation
At time T, the projectile must be a (Euclidean) distance from the cannon equal to the elapsed time multiplied by the projectile speed. This gives an equation for a circle, parametric in elapsed time.
sqr(projectile.X - cannon.X) + sqr(projectile.Y - cannon.Y)
== sqr(t * projectile_speed)
Similarly, at time T, the target has moved along its vector by time multiplied by its velocity:
target.X == t * target.velocityX + target.startX
target.Y == t * target.velocityY + target.startY
The projectile can hit the target when its distance from the cannon matches the projectile's distance.
sqr(projectile.X - cannon.X) + sqr(projectile.Y - cannon.Y)
== sqr(target.X - cannon.X) + sqr(target.Y - cannon.Y)
Wonderful! Substituting the expressions for target.X and target.Y gives
sqr(projectile.X - cannon.X) + sqr(projectile.Y - cannon.Y)
== sqr((t * target.velocityX + target.startX) - cannon.X)
+ sqr((t * target.velocityY + target.startY) - cannon.Y)
Substituting the other side of the equation gives this:
sqr(t * projectile_speed)
== sqr((t * target.velocityX + target.startX) - cannon.X)
+ sqr((t * target.velocityY + target.startY) - cannon.Y)
... subtracting sqr(t * projectile_speed) from both sides and flipping it around:
sqr((t * target.velocityX) + (target.startX - cannon.X))
+ sqr((t * target.velocityY) + (target.startY - cannon.Y))
- sqr(t * projectile_speed)
== 0
... now resolve the results of squaring the subexpressions ...
sqr(target.velocityX) * sqr(t)
+ 2 * t * target.velocityX * (target.startX - cannon.X)
+ sqr(target.startX - cannon.X)
+ sqr(target.velocityY) * sqr(t)
+ 2 * t * target.velocityY * (target.startY - cannon.Y)
+ sqr(target.startY - cannon.Y)
- sqr(projectile_speed) * sqr(t)
== 0
... and group similar terms ...
sqr(target.velocityX) * sqr(t)
+ sqr(target.velocityY) * sqr(t)
- sqr(projectile_speed) * sqr(t)
+ 2 * t * target.velocityX * (target.startX - cannon.X)
+ 2 * t * target.velocityY * (target.startY - cannon.Y)
+ sqr(target.startX - cannon.X)
+ sqr(target.startY - cannon.Y)
== 0
... then combine them ...
(sqr(target.velocityX) + sqr(target.velocityY) - sqr(projectile_speed)) * sqr(t)
+ 2 * (target.velocityX * (target.startX - cannon.X)
+ target.velocityY * (target.startY - cannon.Y)) * t
+ sqr(target.startX - cannon.X) + sqr(target.startY - cannon.Y)
== 0
... giving a standard quadratic equation in t. Finding the positive real zeros of this equation gives the (zero, one, or two) possible hit locations, which can be done with the quadratic formula:
a * sqr(x) + b * x + c == 0
x == (-b ± sqrt(sqr(b) - 4 * a * c)) / (2 * a)
+1 on Jeffrey Hantin's excellent answer here. I googled around and found solutions that were either too complex or not specifically about the case I was interested in (simple constant velocity projectile in 2D space.) His was exactly what I needed to produce the self-contained JavaScript solution below.
The one point I would add is that there are a couple special cases you have to watch for in addition to the discriminant being negative:
"a == 0": occurs if target and projectile are traveling the same speed. (solution is linear, not quadratic)
"a == 0 and b == 0": if both target and projectile are stationary. (no solution unless c == 0, i.e. src & dst are same point.)
Code:
/**
* Return the firing solution for a projectile starting at 'src' with
* velocity 'v', to hit a target, 'dst'.
*
* #param ({x, y}) src position of shooter
* #param ({x, y, vx, vy}) dst position & velocity of target
* #param (Number) v speed of projectile
*
* #return ({x, y}) Coordinate at which to fire (and where intercept occurs). Or `null` if target cannot be hit.
*/
function intercept(src, dst, v) {
const tx = dst.x - src.x;
const ty = dst.y - src.y;
const tvx = dst.vx;
const tvy = dst.vy;
// Get quadratic equation components
const a = tvx * tvx + tvy * tvy - v * v;
const b = 2 * (tvx * tx + tvy * ty);
const c = tx * tx + ty * ty;
// Solve quadratic
const ts = quad(a, b, c); // See quad(), below
// Find smallest positive solution
let sol = null;
if (ts) {
const t0 = ts[0];
const t1 = ts[1];
let t = Math.min(t0, t1);
if (t < 0) t = Math.max(t0, t1);
if (t > 0) {
sol = {
x: dst.x + dst.vx * t,
y: dst.y + dst.vy * t
};
}
}
return sol;
}
/**
* Return solutions for quadratic
*/
function quad(a, b, c) {
let sol = null;
if (Math.abs(a) < 1e-6) {
if (Math.abs(b) < 1e-6) {
sol = Math.abs(c) < 1e-6 ? [0, 0] : null;
} else {
sol = [-c / b, -c / b];
}
} else {
let disc = b * b - 4 * a * c;
if (disc >= 0) {
disc = Math.sqrt(disc);
a = 2 * a;
sol = [(-b - disc) / a, (-b + disc) / a];
}
}
return sol;
}
// For example ...
const sol = intercept(
{x:2, y:4}, // Starting coord
{x:5, y:7, vx: 2, vy:1}, // Target coord and velocity
5 // Projectile velocity
)
console.log('Fire at', sol)
First rotate the axes so that AB is vertical (by doing a rotation)
Now, split the velocity vector of B into the x and y components (say Bx and By). You can use this to calculate the x and y components of the vector you need to shoot at.
B --> Bx
|
|
V
By
Vy
^
|
|
A ---> Vx
You need Vx = Bx and Sqrt(Vx*Vx + Vy*Vy) = Velocity of Ammo.
This should give you the vector you need in the new system. Transform back to old system and you are done (by doing a rotation in the other direction).
Jeffrey Hantin has a nice solution for this problem, though his derivation is overly complicated. Here's a cleaner way of deriving it with some of the resultant code at the bottom.
I'll be using x.y to represent vector dot product, and if a vector quantity is squared, it means I am dotting it with itself.
origpos = initial position of shooter
origvel = initial velocity of shooter
targpos = initial position of target
targvel = initial velocity of target
projvel = velocity of the projectile relative to the origin (cause ur shooting from there)
speed = the magnitude of projvel
t = time
We know that the position of the projectile and target with respect to t time can be described with some equations.
curprojpos(t) = origpos + t*origvel + t*projvel
curtargpos(t) = targpos + t*targvel
We want these to be equal to each other at some point (the point of intersection), so let's set them equal to each other and solve for the free variable, projvel.
origpos + t*origvel + t*projvel = targpos + t*targvel
turns into ->
projvel = (targpos - origpos)/t + targvel - origvel
Let's forget about the notion of origin and target position/velocity. Instead, let's work in relative terms since motion of one thing is relative to another. In this case, what we now have is relpos = targetpos - originpos and relvel = targetvel - originvel
projvel = relpos/t + relvel
We don't know what projvel is, but we do know that we want projvel.projvel to be equal to speed^2, so we'll square both sides and we get
projvel^2 = (relpos/t + relvel)^2
expands into ->
speed^2 = relvel.relvel + 2*relpos.relvel/t + relpos.relpos/t^2
We can now see that the only free variable is time, t, and then we'll use t to solve for projvel. We'll solve for t with the quadratic formula. First separate it out into a, b and c, then solve for the roots.
Before solving, though, remember that we want the best solution where t is smallest, but we need to make sure that t is not negative (you can't hit something in the past)
a = relvel.relvel - speed^2
b = 2*relpos.relvel
c = relpos.relpos
h = -b/(2*a)
k2 = h*h - c/a
if k2 < 0, then there are no roots and there is no solution
if k2 = 0, then there is one root at h
if 0 < h then t = h
else, no solution
if k2 > 0, then there are two roots at h - k and h + k, we also know r0 is less than r1.
k = sqrt(k2)
r0 = h - k
r1 = h + k
we have the roots, we must now solve for the smallest positive one
if 0<r0 then t = r0
elseif 0<r1 then t = r1
else, no solution
Now, if we have a t value, we can plug t back into the original equation and solve for the projvel
projvel = relpos/t + relvel
Now, to the shoot the projectile, the resultant global position and velocity for the projectile is
globalpos = origpos
globalvel = origvel + projvel
And you're done!
My implementation of my solution in Lua, where vec*vec represents vector dot product:
local function lineartrajectory(origpos,origvel,speed,targpos,targvel)
local relpos=targpos-origpos
local relvel=targvel-origvel
local a=relvel*relvel-speed*speed
local b=2*relpos*relvel
local c=relpos*relpos
if a*a<1e-32 then--code translation for a==0
if b*b<1e-32 then
return false,"no solution"
else
local h=-c/b
if 0<h then
return origpos,relpos/h+targvel,h
else
return false,"no solution"
end
end
else
local h=-b/(2*a)
local k2=h*h-c/a
if k2<-1e-16 then
return false,"no solution"
elseif k2<1e-16 then--code translation for k2==0
if 0<h then
return origpos,relpos/h+targvel,h
else
return false,"no solution"
end
else
local k=k2^0.5
if k<h then
return origpos,relpos/(h-k)+targvel,h-k
elseif -k<h then
return origpos,relpos/(h+k)+targvel,h+k
else
return false,"no solution"
end
end
end
end
Following is polar coordinate based aiming code in C++.
To use with rectangular coordinates you would need to first convert the targets relative coordinate to angle/distance, and the targets x/y velocity to angle/speed.
The "speed" input is the speed of the projectile. The units of the speed and targetSpeed are irrelevent, as only the ratio of the speeds are used in the calculation. The output is the angle the projectile should be fired at and the distance to the collision point.
The algorithm is from source code available at http://www.turtlewar.org/ .
// C++
static const double pi = 3.14159265358979323846;
inline double Sin(double a) { return sin(a*(pi/180)); }
inline double Asin(double y) { return asin(y)*(180/pi); }
bool/*ok*/ Rendezvous(double speed,double targetAngle,double targetRange,
double targetDirection,double targetSpeed,double* courseAngle,
double* courseRange)
{
// Use trig to calculate coordinate of future collision with target.
// c
//
// B A
//
// a C b
//
// Known:
// C = distance to target
// b = direction of target travel, relative to it's coordinate
// A/B = ratio of speed and target speed
//
// Use rule of sines to find unknowns.
// sin(a)/A = sin(b)/B = sin(c)/C
//
// a = asin((A/B)*sin(b))
// c = 180-a-b
// B = C*(sin(b)/sin(c))
bool ok = 0;
double b = 180-(targetDirection-targetAngle);
double A_div_B = targetSpeed/speed;
double C = targetRange;
double sin_b = Sin(b);
double sin_a = A_div_B*sin_b;
// If sin of a is greater than one it means a triangle cannot be
// constructed with the given angles that have sides with the given
// ratio.
if(fabs(sin_a) <= 1)
{
double a = Asin(sin_a);
double c = 180-a-b;
double sin_c = Sin(c);
double B;
if(fabs(sin_c) > .0001)
{
B = C*(sin_b/sin_c);
}
else
{
// Sin of small angles approach zero causing overflow in
// calculation. For nearly flat triangles just treat as
// flat.
B = C/(A_div_B+1);
}
// double A = C*(sin_a/sin_c);
ok = 1;
*courseAngle = targetAngle+a;
*courseRange = B;
}
return ok;
}
Here's an example where I devised and implemented a solution to the problem of predictive targeting using a recursive algorithm: http://www.newarteest.com/flash/targeting.html
I'll have to try out some of the other solutions presented because it seems more efficient to calculate it in one step, but the solution I came up with was to estimate the target position and feed that result back into the algorithm to make a new more accurate estimate, repeating several times.
For the first estimate I "fire" at the target's current position and then use trigonometry to determine where the target will be when the shot reaches the position fired at. Then in the next iteration I "fire" at that new position and determine where the target will be this time. After about 4 repeats I get within a pixel of accuracy.
I just hacked this version for aiming in 2d space, I didn't test it very thoroughly yet but it seems to work. The idea behind it is this:
Create a vector perpendicular to the vector pointing from the muzzle to the target.
For a collision to occur, the velocities of the target and the projectile along this vector (axis) should be the same!
Using fairly simple cosine stuff I arrived at this code:
private Vector3 CalculateProjectileDirection(Vector3 a_MuzzlePosition, float a_ProjectileSpeed, Vector3 a_TargetPosition, Vector3 a_TargetVelocity)
{
// make sure it's all in the horizontal plane:
a_TargetPosition.y = 0.0f;
a_MuzzlePosition.y = 0.0f;
a_TargetVelocity.y = 0.0f;
// create a normalized vector that is perpendicular to the vector pointing from the muzzle to the target's current position (a localized x-axis):
Vector3 perpendicularVector = Vector3.Cross(a_TargetPosition - a_MuzzlePosition, -Vector3.up).normalized;
// project the target's velocity vector onto that localized x-axis:
Vector3 projectedTargetVelocity = Vector3.Project(a_TargetVelocity, perpendicularVector);
// calculate the angle that the projectile velocity should make with the localized x-axis using the consine:
float angle = Mathf.Acos(projectedTargetVelocity.magnitude / a_ProjectileSpeed) / Mathf.PI * 180;
if (Vector3.Angle(perpendicularVector, a_TargetVelocity) > 90.0f)
{
angle = 180.0f - angle;
}
// rotate the x-axis so that is points in the desired velocity direction of the projectile:
Vector3 returnValue = Quaternion.AngleAxis(angle, -Vector3.up) * perpendicularVector;
// give the projectile the correct speed:
returnValue *= a_ProjectileSpeed;
return returnValue;
}
I made a public domain Unity C# function here:
http://ringofblades.com/Blades/Code/PredictiveAim.cs
It is for 3D, but you can easily modify this for 2D by replacing the Vector3s with Vector2s and using your down axis of choice for gravity if there is gravity.
In case the theory interests you, I walk through the derivation of the math here:
http://www.gamasutra.com/blogs/KainShin/20090515/83954/Predictive_Aim_Mathematics_for_AI_Targeting.php
I've seen many ways to solve this problem mathematically, but this was a component relevant to a project my class was required to do in high school, and not everyone in this programming class had a background with calculus, or even vectors for that matter, so I created a way to solve this problem with more of a programming approach. The point of intersection will be accurate, although it may hit 1 frame later than in the mathematical computations.
Consider:
S = shooterPos, E = enemyPos, T = targetPos, Sr = shooter range, D = enemyDir
V = distance from E to T, P = projectile speed, Es = enemy speed
In the standard implementation of this problem [S,E,P,Es,D] are all givens and you are solving either to find T or the angle at which to shoot so that you hit T at the proper timing.
The main aspect of this method of solving the problem is to consider the range of the shooter as a circle encompassing all possible points that can be shot at any given time. The radius of this circle is equal to:
Sr = P*time
Where time is calculated as an iteration of a loop.
Thus to find the distance an enemy travels given the time iteration we create the vector:
V = D*Es*time
Now, to actually solve the problem we want to find a point at which the distance from the target (T) to our shooter (S) is less than the range of our shooter (Sr). Here is somewhat of a pseudocode implementation of this equation.
iteration = 0;
while(TargetPoint.hasNotPassedShooter)
{
TargetPoint = EnemyPos + (EnemyMovementVector)
if(distanceFrom(TargetPoint,ShooterPos) < (ShooterRange))
return TargetPoint;
iteration++
}
Basically , intersection concept is not really needed here, As far as you are using projectile motion, you just need to hit at a particular angle and instantiate at the time of shooting so that you get the exact distance of your target from the Source and then once you have the distance, you can calculate the appropriate velocity with which it should shot in order to hit the Target.
The following link makes teh concept clear and is considered helpful, might help:
Projectile motion to always hit a moving target
I grabbed one of the solutions from here, but none of them take into account movement of the shooter. If your shooter is moving, you might want to take that into account (as the shooter's velocity should be added to your bullet's velocity when you fire). Really all you need to do is subtract your shooter's velocity from the target's velocity. So if you're using broofa's code above (which I would recommend), change the lines
tvx = dst.vx;
tvy = dst.vy;
to
tvx = dst.vx - shooter.vx;
tvy = dst.vy - shooter.vy;
and you should be all set.
I need to programmatically solve a system of linear equations in C, Objective C, or (if needed) C++.
Here's an example of the equations:
-44.3940 = a * 50.0 + b * 37.0 + tx
-45.3049 = a * 43.0 + b * 39.0 + tx
-44.9594 = a * 52.0 + b * 41.0 + tx
From this, I'd like to get the best approximation for a, b, and tx.
Cramer's Rule
and
Gaussian Elimination
are two good, general-purpose algorithms (also see Simultaneous Linear Equations). If you're looking for code, check out GiNaC, Maxima, and SymbolicC++ (depending on your licensing requirements, of course).
EDIT: I know you're working in C land, but I also have to put in a good word for SymPy (a computer algebra system in Python). You can learn a lot from its algorithms (if you can read a bit of python). Also, it's under the new BSD license, while most of the free math packages are GPL.
You can solve this with a program exactly the same way you solve it by hand (with multiplication and subtraction, then feeding results back into the equations). This is pretty standard secondary-school-level mathematics.
-44.3940 = 50a + 37b + c (A)
-45.3049 = 43a + 39b + c (B)
-44.9594 = 52a + 41b + c (C)
(A-B): 0.9109 = 7a - 2b (D)
(B-C): 0.3455 = -9a - 2b (E)
(D-E): 1.2564 = 16a (F)
(F/16): a = 0.078525 (G)
Feed G into D:
0.9109 = 7a - 2b
=> 0.9109 = 0.549675 - 2b (substitute a)
=> 0.361225 = -2b (subtract 0.549675 from both sides)
=> -0.1806125 = b (divide both sides by -2) (H)
Feed H/G into A:
-44.3940 = 50a + 37b + c
=> -44.3940 = 3.92625 - 6.6826625 + c (substitute a/b)
=> -41.6375875 = c (subtract 3.92625 - 6.6826625 from both sides)
So you end up with:
a = 0.0785250
b = -0.1806125
c = -41.6375875
If you plug these values back into A, B and C, you'll find they're correct.
The trick is to use a simple 4x3 matrix which reduces in turn to a 3x2 matrix, then a 2x1 which is "a = n", n being an actual number. Once you have that, you feed it into the next matrix up to get another value, then those two values into the next matrix up until you've solved all variables.
Provided you have N distinct equations, you can always solve for N variables. I say distinct because these two are not:
7a + 2b = 50
14a + 4b = 100
They are the same equation multiplied by two so you cannot get a solution from them - multiplying the first by two then subtracting leaves you with the true but useless statement:
0 = 0 + 0
By way of example, here's some C code that works out the simultaneous equations that you're placed in your question. First some necessary types, variables, a support function for printing out an equation, and the start of main:
#include <stdio.h>
typedef struct { double r, a, b, c; } tEquation;
tEquation equ1[] = {
{ -44.3940, 50, 37, 1 }, // -44.3940 = 50a + 37b + c (A)
{ -45.3049, 43, 39, 1 }, // -45.3049 = 43a + 39b + c (B)
{ -44.9594, 52, 41, 1 }, // -44.9594 = 52a + 41b + c (C)
};
tEquation equ2[2], equ3[1];
static void dumpEqu (char *desc, tEquation *e, char *post) {
printf ("%10s: %12.8lf = %12.8lfa + %12.8lfb + %12.8lfc (%s)\n",
desc, e->r, e->a, e->b, e->c, post);
}
int main (void) {
double a, b, c;
Next, the reduction of the three equations with three unknowns to two equations with two unknowns:
// First step, populate equ2 based on removing c from equ.
dumpEqu (">", &(equ1[0]), "A");
dumpEqu (">", &(equ1[1]), "B");
dumpEqu (">", &(equ1[2]), "C");
puts ("");
// A - B
equ2[0].r = equ1[0].r * equ1[1].c - equ1[1].r * equ1[0].c;
equ2[0].a = equ1[0].a * equ1[1].c - equ1[1].a * equ1[0].c;
equ2[0].b = equ1[0].b * equ1[1].c - equ1[1].b * equ1[0].c;
equ2[0].c = 0;
// B - C
equ2[1].r = equ1[1].r * equ1[2].c - equ1[2].r * equ1[1].c;
equ2[1].a = equ1[1].a * equ1[2].c - equ1[2].a * equ1[1].c;
equ2[1].b = equ1[1].b * equ1[2].c - equ1[2].b * equ1[1].c;
equ2[1].c = 0;
dumpEqu ("A-B", &(equ2[0]), "D");
dumpEqu ("B-C", &(equ2[1]), "E");
puts ("");
Next, the reduction of the two equations with two unknowns to one equation with one unknown:
// Next step, populate equ3 based on removing b from equ2.
// D - E
equ3[0].r = equ2[0].r * equ2[1].b - equ2[1].r * equ2[0].b;
equ3[0].a = equ2[0].a * equ2[1].b - equ2[1].a * equ2[0].b;
equ3[0].b = 0;
equ3[0].c = 0;
dumpEqu ("D-E", &(equ3[0]), "F");
puts ("");
Now that we have a formula of the type number1 = unknown * number2, we can simply work out the unknown value with unknown <- number1 / number2. Then, once you've figured that value out, substitute it into one of the equations with two unknowns and work out the second value. Then substitute both those (now-known) unknowns into one of the original equations and you now have the values for all three unknowns:
// Finally, substitute values back into equations.
a = equ3[0].r / equ3[0].a;
printf ("From (F ), a = %12.8lf (G)\n", a);
b = (equ2[0].r - equ2[0].a * a) / equ2[0].b;
printf ("From (D,G ), b = %12.8lf (H)\n", b);
c = (equ1[0].r - equ1[0].a * a - equ1[0].b * b) / equ1[0].c;
printf ("From (A,G,H), c = %12.8lf (I)\n", c);
return 0;
}
The output of that code matches the earlier calculations in this answer:
>: -44.39400000 = 50.00000000a + 37.00000000b + 1.00000000c (A)
>: -45.30490000 = 43.00000000a + 39.00000000b + 1.00000000c (B)
>: -44.95940000 = 52.00000000a + 41.00000000b + 1.00000000c (C)
A-B: 0.91090000 = 7.00000000a + -2.00000000b + 0.00000000c (D)
B-C: -0.34550000 = -9.00000000a + -2.00000000b + 0.00000000c (E)
D-E: -2.51280000 = -32.00000000a + 0.00000000b + 0.00000000c (F)
From (F ), a = 0.07852500 (G)
From (D,G ), b = -0.18061250 (H)
From (A,G,H), c = -41.63758750 (I)
Take a look at the Microsoft Solver Foundation.
With it you could write code like this:
SolverContext context = SolverContext.GetContext();
Model model = context.CreateModel();
Decision a = new Decision(Domain.Real, "a");
Decision b = new Decision(Domain.Real, "b");
Decision c = new Decision(Domain.Real, "c");
model.AddDecisions(a,b,c);
model.AddConstraint("eqA", -44.3940 == 50*a + 37*b + c);
model.AddConstraint("eqB", -45.3049 == 43*a + 39*b + c);
model.AddConstraint("eqC", -44.9594 == 52*a + 41*b + c);
Solution solution = context.Solve();
string results = solution.GetReport().ToString();
Console.WriteLine(results);
Here is the output:
===Solver Foundation Service Report===
Datetime: 04/20/2009 23:29:55
Model Name: Default
Capabilities requested: LP
Solve Time (ms): 1027
Total Time (ms): 1414
Solve Completion Status: Optimal
Solver Selected: Microsoft.SolverFoundation.Solvers.SimplexSolver
Directives:
Microsoft.SolverFoundation.Services.Directive
Algorithm: Primal
Arithmetic: Hybrid
Pricing (exact): Default
Pricing (double): SteepestEdge
Basis: Slack
Pivot Count: 3
===Solution Details===
Goals:
Decisions:
a: 0.0785250000000004
b: -0.180612500000001
c: -41.6375875
For a 3x3 system of linear equations I guess it would be okay to roll out your own algorithms.
However, you might have to worry about accuracy, division by zero or really small numbers and what to do about infinitely many solutions. My suggestion is to go with a standard numerical linear algebra package such as LAPACK.
Are you looking for a software package that'll do the work or actually doing the matrix operations and such and do each step?
The the first, a coworker of mine just used Ocaml GLPK. It is just a wrapper for the GLPK, but it removes a lot of the steps of setting things up. It looks like you're going to have to stick with the GLPK, in C, though. For the latter, thanks to delicious for saving an old article I used to learn LP awhile back, PDF. If you need specific help setting up further, let us know and I'm sure, me or someone will wander back in and help, but, I think it's fairly straight forward from here. Good Luck!
Template Numerical Toolkit from NIST has tools for doing that.
One of the more reliable ways is to use a QR Decomposition.
Here's an example of a wrapper so that I can call "GetInverse(A, InvA)" in my code and it will put the inverse into InvA.
void GetInverse(const Array2D<double>& A, Array2D<double>& invA)
{
QR<double> qr(A);
invA = qr.solve(I);
}
Array2D is defined in the library.
In terms of run-time efficiency, others have answered better than I. If you always will have the same number of equations as variables, I like Cramer's rule as it's easy to implement. Just write a function to calculate determinant of a matrix (or use one that's already written, I'm sure you can find one out there), and divide the determinants of two matrices.
Personally, I'm partial to the algorithms of Numerical Recipes. (I'm fond of the C++ edition.)
This book will teach you why the algorithms work, plus show you some pretty-well debugged implementations of those algorithms.
Of course, you could just blindly use CLAPACK (I've used it with great success), but I would first hand-type a Gaussian Elimination algorithm to at least have a faint idea of the kind of work that has gone into making these algorithms stable.
Later, if you're doing more interesting linear algebra, looking around the source code of Octave will answer a lot of questions.
From the wording of your question, it seems like you have more equations than unknowns and you want to minimize the inconsistencies. This is typically done with linear regression, which minimizes the sum of the squares of the inconsistencies. Depending on the size of the data, you can do this in a spreadsheet or in a statistical package. R is a high-quality, free package that does linear regression, among a lot of other things. There is a lot to linear regression (and a lot of gotcha's), but as it's straightforward to do for simple cases. Here's an R example using your data. Note that the "tx" is the intercept to your model.
> y <- c(-44.394, -45.3049, -44.9594)
> a <- c(50.0, 43.0, 52.0)
> b <- c(37.0, 39.0, 41.0)
> regression = lm(y ~ a + b)
> regression
Call:
lm(formula = y ~ a + b)
Coefficients:
(Intercept) a b
-41.63759 0.07852 -0.18061
function x = LinSolve(A,y)
%
% Recursive Solution of Linear System Ax=y
% matlab equivalent: x = A\y
% x = n x 1
% A = n x n
% y = n x 1
% Uses stack space extensively. Not efficient.
% C allows recursion, so convert it into C.
% ----------------------------------------------
n=length(y);
x=zeros(n,1);
if(n>1)
x(1:n-1,1) = LinSolve( A(1:n-1,1:n-1) - (A(1:n-1,n)*A(n,1:n-1))./A(n,n) , ...
y(1:n-1,1) - A(1:n-1,n).*(y(n,1)/A(n,n)));
x(n,1) = (y(n,1) - A(n,1:n-1)*x(1:n-1,1))./A(n,n);
else
x = y(1,1) / A(1,1);
end
For general cases, you could use python along with numpy for Gaussian elimination. And then plug in values and get the remaining values.