public static void cvt_sample(ref CVTHaltonData data, int dim_num, int n, int n_now, int sample, bool initialize,
ref int seed, ref double[] r)
//****************************************************************************80
//
// Purpose:
//
// CVT_SAMPLE returns sample points.
//
// Discussion:
//
// N sample points are to be taken from the unit box of dimension DIM_NUM.
//
// These sample points are usually created by a pseudorandom process
// for which the points are essentially indexed by a quantity called
// SEED. To get N sample points, we generate values with indices
// SEED through SEED+N-1.
//
// It may not be practical to generate all the sample points in a
// single call. For that reason, the routine allows the user to
// request a total of N points, but to require that only N_NOW be
// generated now (on this call).
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 23 June 2005
//
// Author:
//
// John Burkardt
//
// Parameters:
//
// Input, int DIM_NUM, the spatial dimension.
//
// Input, int N, the number of Voronoi cells.
//
// Input, int N_NOW, the number of sample points to be generated
// on this call. N_NOW must be at least 1.
//
// Input, int SAMPLE, specifies how the sampling is done.
// -1, 'RANDOM', using C++ RANDOM function;
// 0, 'UNIFORM', using a simple uniform RNG;
// 1, 'HALTON', from a Halton sequence;
// 2, 'GRID', points from a grid;
// 3, 'USER', call "user" routine.
//
// Input, bool INITIALIZE, is TRUE if the pseudorandom process should be
// reinitialized.
//
// Input/output, int *SEED, the random number seed.
//
// Output, double R[DIM_NUM*N_NOW], the sample points.
//
{
switch (n_now)
{
case < 1:
Console.WriteLine("");
Console.WriteLine("CVT_SAMPLE - Fatal error!");
Console.WriteLine(" N_NOW < 1.");
return;
default:
int i;
int j;
switch (sample)
{
case -1:
{
/*
* if (initialize)
* {
* random_initialize(ref seed);
* }
*/
for (j = 0; j < n_now; j++)
{
for (i = 0; i < dim_num; i++)
{
r[i + j * dim_num] = RNG.nextdouble(); //(double) random() / (double) RAND_MAX;
}
}
seed += n_now * dim_num;
break;
}
case 0:
r = UniformRNG.r8mat_uniform_01(dim_num, n_now, ref seed);
break;
case 1:
{
data.halton_seed = new int[dim_num];
data.halton_leap = new int[dim_num];
data.halton_base = new int[dim_num];
int halton_step = seed;
for (i = 0; i < dim_num; i++)
{
data.halton_seed[i] = 0;
}
for (i = 0; i < dim_num; i++)
{
data.halton_leap[i] = 1;
}
for (i = 0; i < dim_num; i++)
{
data.halton_base[i] = Prime.prime(i + 1);
}
Halton.i4_to_halton_sequence(dim_num, n_now, halton_step, data.halton_seed,
data.halton_leap, data.halton_base, ref r);
seed += n_now;
break;
}
case 2:
{
double exponent = 1.0 / dim_num;
data.ngrid = (int)Math.Pow(n, exponent);
int rank_max = (int)Math.Pow(data.ngrid, dim_num);
data.tuple = new int[dim_num];
if (rank_max < n)
{
data.ngrid += 1;
rank_max = (int)Math.Pow(data.ngrid, dim_num);
}
switch (initialize)
{
case true:
data.rank = -1;
BTuple.tuple_next_fast(ref data.tupledata, data.ngrid, dim_num, data.rank, ref data.tuple);
break;
}
data.rank = seed % rank_max;
for (j = 0; j < n_now; j++)
{
BTuple.tuple_next_fast(ref data.tupledata, data.ngrid, dim_num, data.rank, ref data.tuple);
data.rank += 1;
data.rank %= rank_max;
for (i = 0; i < dim_num; i++)
{
r[i + j * dim_num] = (2.0 * data.tuple[i] - 1) / (2.0 * data.ngrid);
}
}
seed += n_now;
break;
}
case 3:
user(dim_num, n_now, ref seed, ref r);
break;
default:
Console.WriteLine("");
Console.WriteLine("CVT_SAMPLE - Fatal error!");
Console.WriteLine(" The value of SAMPLE = " + sample + " is illegal.");
break;
}
break;
}
}
public static void cvt(ref CVTHaltonData data, int dim_num, int n, int batch, int init, int sample, int sample_num,
int it_max, int it_fixed, ref int seed, ref double[] r, ref int it_num, ref double it_diff,
ref double energy)
//****************************************************************************80
//
// Purpose:
//
// CVT computes a Centroidal Voronoi Tessellation.
//
// Discussion:
//
// This routine initializes the data, and carries out the
// CVT iteration.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 23 June 2005
//
// Author:
//
// John Burkardt
//
// Reference:
//
// Qiang Du, Vance Faber, and Max Gunzburger,
// Centroidal Voronoi Tessellations: Applications and Algorithms,
// SIAM Review, Volume 41, 1999, pages 637-676.
//
// Parameters:
//
// Input, int DIM_NUM, the spatial dimension.
//
// Input, int N, the number of Voronoi cells.
//
// Input, int BATCH, sets the maximum number of sample points
// generated at one time. It is inefficient to generate the sample
// points 1 at a time, but memory intensive to generate them all
// at once. You might set BATCH to min ( SAMPLE_NUM, 10000 ), for instance.
// BATCH must be at least 1.
//
// Input, int INIT, specifies how the points are to be initialized.
// -1, 'RANDOM', using C++ RANDOM function;
// 0, 'UNIFORM', using a simple uniform RNG;
// 1, 'HALTON', from a Halton sequence;
// 2, 'GRID', points from a grid;
// 3, 'USER', call "user" routine;
// 4, points are already initialized on input.
//
// Input, int SAMPLE, specifies how the sampling is done.
// -1, 'RANDOM', using C++ RANDOM function;
// 0, 'UNIFORM', using a simple uniform RNG;
// 1, 'HALTON', from a Halton sequence;
// 2, 'GRID', points from a grid;
// 3, 'USER', call "user" routine.
//
// Input, int SAMPLE_NUM, the number of sample points.
//
// Input, int IT_MAX, the maximum number of iterations.
//
// Input, int IT_FIXED, the maximum number of iterations to take
// with a fixed set of sample points.
//
// Input/output, int *SEED, the random number seed.
//
// Input/output, double R[DIM_NUM*N], the approximate CVT points.
// If INIT = 4 on input, then it is assumed that these values have been
// initialized. On output, the CVT iteration has been applied to improve
// the value of the points.
//
// Output, int *IT_NUM, the number of iterations taken. Generally,
// this will be equal to IT_MAX, unless the iteration tolerance was
// satisfied early.
//
// Output, double *IT_DIFF, the L2 norm of the difference
// between the iterates.
//
// Output, double *ENERGY, the discrete "energy", divided
// by the number of sample points.
//
{
const bool DEBUG = true;
bool initialize;
int seed_base = 0;
switch (batch)
{
case < 1:
Console.WriteLine("");
Console.WriteLine("CVT - Fatal error!");
Console.WriteLine(" The input value BATCH < 1.");
return;
}
switch (seed)
{
case <= 0:
Console.WriteLine("");
Console.WriteLine("CVT - Fatal error!");
Console.WriteLine(" The input value SEED <= 0.");
return;
}
switch (DEBUG)
{
case true:
Console.WriteLine("");
Console.WriteLine(" Step SEED L2-Change Energy");
Console.WriteLine("");
break;
}
it_num = 0;
it_diff = 0.0;
energy = 0.0;
int seed_init = seed;
//
// Initialize the data, unless the user has already done that.
//
if (init != 4)
{
initialize = true;
cvt_sample(ref data, dim_num, n, n, init, initialize, ref seed, ref r);
}
switch (DEBUG)
{
case true:
Console.WriteLine(" "
+ it_num.ToString(CultureInfo.InvariantCulture).PadLeft(4) + " "
+ seed_init.ToString(CultureInfo.InvariantCulture).PadLeft(12) + "");
break;
}
//
// If the initialization and sampling steps use the same random number
// scheme, then the sampling scheme does not have to be initialized.
//
initialize = init != sample;
//
// Carry out the iteration.
//
while (it_num < it_max)
{
switch (it_num % it_fixed)
{
//
// If it's time to update the seed, save its current value
// as the starting value for all iterations in this cycle.
// If it's not time to update the seed, restore it to its initial
// value for this cycle.
//
case 0:
seed_base = seed;
break;
default:
seed = seed_base;
break;
}
it_num += 1;
seed_init = seed;
cvt_iterate(ref data, dim_num, n, batch, sample, initialize, sample_num, ref seed,
ref r, ref it_diff, ref energy);
initialize = false;
switch (DEBUG)
{
case true:
Console.WriteLine(" "
+ it_num.ToString(CultureInfo.InvariantCulture).PadLeft(4) + " "
+ seed_init.ToString(CultureInfo.InvariantCulture).PadLeft(12) + " "
+ it_diff.ToString(CultureInfo.InvariantCulture).PadLeft(14) + " "
+ energy.ToString(CultureInfo.InvariantCulture).PadLeft(14) + "");
break;
}
}
}
public static void cvt_iterate(ref CVTHaltonData data, int dim_num, int n, int batch, int sample, bool initialize,
int sample_num, ref int seed, ref double[] r, ref double it_diff, ref double energy)
//****************************************************************************80
//
// Purpose:
//
// CVT_ITERATE takes one step of the CVT iteration.
//
// Discussion:
//
// The routine is given a set of points, called "generators", which
// define a tessellation of the region into Voronoi cells. Each point
// defines a cell. Each cell, in turn, has a centroid, but it is
// unlikely that the centroid and the generator coincide.
//
// Each time this CVT iteration is carried out, an attempt is made
// to modify the generators in such a way that they are closer and
// closer to being the centroids of the Voronoi cells they generate.
//
// A large number of sample points are generated, and the nearest generator
// is determined. A count is kept of how many points were nearest to each
// generator. Once the sampling is completed, the location of all the
// generators is adjusted. This step should decrease the discrepancy
// between the generators and the centroids.
//
// The centroidal Voronoi tessellation minimizes the "energy",
// defined to be the integral, over the region, of the square of
// the distance between each point in the region and its nearest generator.
// The sampling technique supplies a discrete estimate of this
// energy.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 20 September 2004
//
// Author:
//
// John Burkardt
//
// Reference:
//
// Qiang Du, Vance Faber, and Max Gunzburger,
// Centroidal Voronoi Tessellations: Applications and Algorithms,
// SIAM Review, Volume 41, 1999, pages 637-676.
//
// Parameters:
//
// Input, int DIM_NUM, the spatial dimension.
//
// Input, int N, the number of Voronoi cells.
//
// Input, int BATCH, sets the maximum number of sample points
// generated at one time. It is inefficient to generate the sample
// points 1 at a time, but memory intensive to generate them all
// at once. You might set BATCH to min ( SAMPLE_NUM, 10000 ), for instance.
// BATCH must be at least 1.
//
// Input, int SAMPLE, specifies how the sampling is done.
// -1, 'RANDOM', using C++ RANDOM function;
// 0, 'UNIFORM', using a simple uniform RNG;
// 1, 'HALTON', from a Halton sequence;
// 2, 'GRID', points from a grid;
// 3, 'USER', call "user" routine.
//
// Input, bool INITIALIZE, is TRUE if the SEED must be reset to SEED_INIT
// before computation. Also, the pseudorandom process may need to be
// reinitialized.
//
// Input, int SAMPLE_NUM, the number of sample points.
//
// Input/output, int *SEED, the random number seed.
//
// Input/output, double R[DIM_NUM*N], the Voronoi
// cell generators. On output, these have been modified
//
// Output, double *IT_DIFF, the L2 norm of the difference
// between the iterates.
//
// Output, double *ENERGY, the discrete "energy", divided
// by the number of sample points.
//
{
int i;
int j;
//
// Take each generator as the first sample point for its region.
// This can slightly slow the convergence, but it simplifies the
// algorithm by guaranteeing that no region is completely missed
// by the sampling.
//
energy = 0.0;
double[] r2 = new double[dim_num * n];
int[] count = new int[n];
int[] nearest = new int[sample_num];
double[] s = new double[dim_num * sample_num];
for (j = 0; j < n; j++)
{
for (i = 0; i < dim_num; i++)
{
r2[i + j * dim_num] = r[i + j * dim_num];
}
}
for (j = 0; j < n; j++)
{
count[j] = 1;
}
//
// Generate the sampling points S.
//
int have = 0;
while (have < sample_num)
{
int get = Math.Min(sample_num - have, batch);
cvt_sample(ref data, dim_num, sample_num, get, sample, initialize, ref seed, ref s);
initialize = false;
have += get;
//
// Find the index N of the nearest cell generator to each sample point S.
//
find_closest(dim_num, n, get, s, r, ref nearest);
//
// Add S to the centroid associated with generator N.
//
for (j = 0; j < get; j++)
{
int j2 = nearest[j];
for (i = 0; i < dim_num; i++)
{
r2[i + j2 * dim_num] += s[i + j * dim_num];
}
for (i = 0; i < dim_num; i++)
{
energy += Math.Pow(r[i + j2 * dim_num] - s[i + j * dim_num], 2);
}
count[j2] += 1;
}
}
//
// Estimate the centroids.
//
for (j = 0; j < n; j++)
{
for (i = 0; i < dim_num; i++)
{
r2[i + j * dim_num] /= count[j];
}
}
//
// Determine the sum of the distances between generators and centroids.
//
it_diff = 0.0;
for (j = 0; j < n; j++)
{
double term = 0.0;
for (i = 0; i < dim_num; i++)
{
term += (r2[i + j * dim_num] - r[i + j * dim_num])
* (r2[i + j * dim_num] - r[i + j * dim_num]);
}
it_diff += Math.Sqrt(term);
}
//
// Replace the generators by the centroids.
//
for (j = 0; j < n; j++)
{
for (i = 0; i < dim_num; i++)
{
r[i + j * dim_num] = r2[i + j * dim_num];
}
}
//
// Normalize the discrete energy estimate.
//
energy /= sample_num;
}
public static double cvt_energy(ref CVTHaltonData data, int dim_num, int n, int batch, int sample, bool initialize,
int sample_num, ref int seed, double[] r)
//****************************************************************************80
//
// Purpose:
//
// CVT_ENERGY computes the CVT energy of a dataset.
//
// Discussion:
//
// For a given number of generators, a CVT is a minimizer (or at least
// a local minimizer) of the CVT energy. During a CVT iteration,
// it should generally be the case that the CVT energy decreases from
// step to step, and that perturbations or adjustments of an
// approximate CVT will almost always have higher CVT energy.
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 03 December 2004
//
// Author:
//
// John Burkardt
//
// Parameters:
//
// Input, int DIM_NUM, the spatial dimension.
//
// Input, int N, the number of generators.
//
// Input, int BATCH, the maximum number of sample points to generate
// at one time.
//
// Input, int SAMPLE, specifies how the sampling is done.
// -1, 'RANDOM', using C++ RANDOM function;
// 0, 'UNIFORM', using a simple uniform RNG;
// 1, 'HALTON', from a Halton sequence;
// 2, 'GRID', points from a grid;
// 3, 'USER', call "user" routine.
//
// Input, bool INITIALIZE, is TRUE if the pseudorandom process
// should be reinitialized.
//
// Input, int SAMPLE_NUM, the number of sample points to use.
//
// Input/output, int *SEED, a seed for the random number generator.
//
// Input, double R[DIM_NUM*N], the coordinates of the points.
//
// Output, double CVT_ENERGY, the estimated CVT energy.
//
{
int[] nearest = new int[batch];
double[] s = new double [dim_num * batch];
int have = 0;
double energy = 0.0;
while (have < sample_num)
{
int get = Math.Min(sample_num - have, batch);
cvt_sample(ref data, dim_num, sample_num, get, sample, initialize, ref seed, ref s);
have += get;
find_closest(dim_num, n, get, s, r, ref nearest);
int j;
for (j = 0; j < get; j++)
{
int i;
for (i = 0; i < dim_num; i++)
{
energy += (s[i + j * dim_num] - r[i + nearest[j] * dim_num])
* (s[i + j * dim_num] - r[i + nearest[j] * dim_num]);
}
}
}
energy /= sample_num;
return(energy);
}
