public static GdescentResult mlat(Matrix <Double> anchors_in, Vector <double> ranges_in,
Matrix <double> bounds_in = null, int n_trial = 100, double alpha = 0.001, double time_threshold = 0)
{
GdescentResult gdescent_result = gdescent(anchors_in, ranges_in, bounds_in, n_trial, alpha, time_threshold);
int idx = -1;
double error = double.MaxValue;
for (int i = 0; i < gdescent_result.error.Count; i++)
{
if (gdescent_result.error[i] < error)
{
idx = i;
error = gdescent_result.error[i];
}
}
gdescent_result.estimator = gdescent_result.estimator_candidate.Row(idx);
return(gdescent_result);
}
public static GdescentResult gdescent(Matrix <Double> anchors_in, Vector <double> ranges_in,
Matrix <double> bounds_in = null, int n_trial = 100, double alpha = 0.001, double time_threshold = 0)
{
System.Random random = new SystemRandomSource();
int n = anchors_in.RowCount;
int dim = anchors_in.ColumnCount;
GdescentResult gdescent_result = new GdescentResult(n_trial, dim);
if (bounds_in == null)
{
bounds_in = DenseMatrix.Build.Dense(1, dim);
}
Matrix <double> bounds_temp = anchors_in.Stack(bounds_in);
Matrix <double> bounds = DenseMatrix.Build.Dense(2, dim);
for (int i = 0; i < dim; i++)
{
bounds[0, i] = bounds_temp.Column(i).Min();
bounds[1, i] = bounds_temp.Column(i).Max();
}
if (time_threshold == 0)
{
time_threshold = 1.0 / n_trial;
}
Vector <double> ranges = DenseVector.Build.Dense(n);
for (int i = 0; i < n_trial; i++)
{
Vector <double> estimator0 = DenseVector.Build.Dense(dim);
for (int j = 0; j < dim; j++)
{
estimator0[j] = random.NextDouble() * (bounds[1, j] - bounds[0, j]) + bounds[0, j];
}
Vector <double> estimator = DenseVector.OfVector(estimator0);
Stopwatch stopwatch = new Stopwatch();
while (true)
{
for (int j = 0; j < n; j++)
{
ranges[j] = MLAT.d(anchors_in.Row(j), estimator);
}
double error = MLAT.d(ranges_in, ranges);
Vector <double> delta = DenseVector.Build.Dense(dim);
for (int j = 0; j < n; j++)
{
delta += (ranges_in[j] - ranges[j]) / ranges[j] * (estimator - anchors_in.Row(j));
}
delta *= 2 * alpha;
Vector <double> estimator_next = estimator - delta;
for (int j = 0; j < n; j++)
{
ranges[j] = MLAT.d(anchors_in.Row(j), estimator_next);
}
double error_next = MLAT.d(ranges_in, ranges);
if (error_next < error)
{
estimator = estimator_next;
}
else
{
gdescent_result.estimator_candidate.SetRow(i, estimator);
gdescent_result.error[i] = error;
break;
}
if (stopwatch.ElapsedMilliseconds > time_threshold)
{
gdescent_result.error[i] = double.MaxValue;
break;
}
}
}
return(gdescent_result);
}
