Created
July 28, 2012 02:07
-
-
Save skeeto/3191437 to your computer and use it in GitHub Desktop.
Compute local maxima using Newton's method
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#include <stdio.h> | |
#include <math.h> | |
#include <float.h> | |
#define N 3 | |
typedef struct { | |
double x[N]; | |
} vector; | |
void print_vector(vector v) | |
{ | |
putchar('('); | |
int i; | |
for (i = 0; i < N; i++) | |
printf("%f%s", v.x[i], i == N - 1 ? ")\n" : ", "); | |
} | |
/* Return true if all of v's elements are 0. */ | |
int zero(vector v) | |
{ | |
int i; | |
for (i = 0; i < N; i++) { | |
if (v.x[i] >= DBL_EPSILON) | |
return 0; | |
} | |
return 1; | |
} | |
/* Central finite difference coefficients. */ | |
const double coef[][9] = { | |
{ 1./280, -4./105, 1./5, -4./5, 0, 4./5, -1./5, 4./105, -1./280}, | |
{-1./560, 8./315, -1./5, 8./5, -205./72, 8./5, -1./5, 8./315, -1./560}, | |
}; | |
/* Compute a reasonable h for floating point precision. */ | |
#define compute_h(x) (x) != 0 ? sqrt(fabs(x) * FLT_EPSILON) : FLT_EPSILON | |
/* Compute the nth derivatives of f() at v. */ | |
vector df(int n, double (*f)(vector), vector v) | |
{ | |
vector result = {{0}}; | |
int i, d; | |
for (d = 0; d < N; d++) { | |
vector vh = v; | |
double h = compute_h(v.x[d]); | |
for (i = -4; i <= 4; i++) { | |
vh.x[d] = v.x[d] + h * i; | |
result.x[d] += coef[n - 1][i + 4] * f(vh); | |
} | |
result.x[d] /= pow(h, n); | |
} | |
return result; | |
} | |
/* Find the local minima/maxima via Newton's method and gradient descent. */ | |
vector fmin(double (*f)(vector), vector v) | |
{ | |
while (!zero(df(1, f, v))) { | |
vector d1 = df(1, f, v), d2 = df(2, f, v); | |
int i; | |
for (i = 0; i < N; i++) | |
v.x[i] -= d1.x[i] / d2.x[i]; | |
} | |
return v; | |
} | |
/* Example function (higher-order paraboloid). */ | |
double f(vector v) | |
{ | |
return pow(v.x[0], 2) + pow(v.x[1], 2) + pow(v.x[2], 2); | |
} | |
int main() | |
{ | |
vector v = {{1, 1, 1}}; | |
print_vector(fmin(f, v)); | |
return 0; | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment