topbase.c File Reference

Detailed Description

Provides routines to determine the top-level and base-level of data.

This file contains functions to find the top-level and base-level of a dataset, as well as to identify crossing points within the data. The routines utilize histogram binning and statistical analysis to compute these levels.

Copyright

• (c) 2002 The University of Chicago, as Operator of Argonne National Laboratory.
• (c) 2002 The Regents of the University of California, as Operator of Los Alamos National Laboratory.

License

This file is distributed under the terms of the Software License Agreement found in the file LICENSE included with this distribution.

Author

M. Borland, C. Saunders, R. Soliday

Definition in file topbase.c.

#include "mdb.h"

Go to the source code of this file.

Functions
long	findTopBaseLevels (double *top, double *base, double *data, int64_t points, long bins, double sigmasRequired)
	Finds the top-level and base-level of a dataset.
int64_t	findCrossingPoint (int64_t start, double *data, int64_t points, double level, long direction, double *indepData, double *location)
	Finds the crossing point in the data where the data crosses a specified level.

Function Documentation

findCrossingPoint()

int64_t findCrossingPoint	(	int64_t	start,
		double *	data,
		int64_t	points,
		double	level,
		long	direction,
		double *	indepData,
		double *	location )

Finds the crossing point in the data where the data crosses a specified level.

Parameters

start	The starting index to search for the crossing point.
data	Pointer to the array of data points.
points	Number of data points in the array.
level	The level at which to find the crossing point.
direction	The direction of crossing (positive for upward, negative for downward).
indepData	Pointer to the independent data array corresponding to the data points. Can be NULL.
location	Pointer to store the interpolated location of the crossing point. Can be NULL.

Returns

The index of the crossing point on success,

-1 if no crossing is found or if input parameters are invalid.

Definition at line 118 of file topbase.c.

                                                               {
  double diff;
  int64_t i;
  long transitionPossible;
 
  if (start < 0 || start > (points - 1))
    return -1;
  transitionPossible = 0;
  for (i = start; i < points; i++) {
    diff = direction * (data[i] - level);
    if (diff <= 0)
      transitionPossible = 1;
    if (diff > 0 && transitionPossible)
      break;
  }
  if (i == points)
    return -1;
  if (!indepData || !location)
    return i;
  if (i == 0 || data[i] == data[i - 1]) {
    *location = indepData[i];
    return i;
  }
  *location = indepData[i - 1] + (indepData[i] - indepData[i - 1]) / (data[i] - data[i - 1]) * (level - data[i - 1]);
  return i;
}

findTopBaseLevels()

long findTopBaseLevels	(	double *	top,
		double *	base,
		double *	data,
		int64_t	points,
		long	bins,
		double	sigmasRequired )

Finds the top-level and base-level of a dataset.

Parameters

top	Pointer to store the computed top-level value.
base	Pointer to store the computed base-level value.
data	Pointer to the array of data points.
points	Number of data points in the array.
bins	Number of bins to use for histogram computation. If <= 0, a default bin factor is applied.
sigmasRequired	Number of standard deviations required for determining significant histogram peaks.

Returns

Returns 1 on success, 0 on failure.

Definition at line 36 of file topbase.c.

                                                         {
  long binned, iMaximum, i;
  static long maxBins = 0;
  static double *histogram = NULL;
  double min, max, midpoint, maxHistogram, meanBinned;
  double lower, upper, delta;
 
  if (bins <= 0)
    bins = DEFAULT_BINFACTOR * points;
  if (sigmasRequired <= 0)
    sigmasRequired = DEFAULT_SIGMAS;
  if (points < 2)
    return 0;
 
  if (bins > maxBins)
    histogram = tmalloc(sizeof(*histogram) * (maxBins = bins));
 
  if (!find_min_max(&min, &max, data, points))
    return 0;
 
  *base = min;
  *top = max;
  midpoint = (min + max) / 2;
  if (points < 10)
    return 1;
 
  delta = (midpoint - min) / (bins - 1);
  lower = min - delta / 2;
  upper = midpoint + delta / 2;
  if ((binned = make_histogram(histogram, bins, lower, upper, data, points, 1))) {
    maxHistogram = 0;
    iMaximum = -1;
    for (i = 0; i < bins; i++) {
      if (maxHistogram < histogram[i]) {
        iMaximum = i;
        maxHistogram = histogram[i];
      }
    }
    meanBinned = ((double)binned) / bins;
    if ((maxHistogram > 1) && (maxHistogram > (meanBinned + sigmasRequired * sqrt(meanBinned))))
      *base = lower + (iMaximum + 0.5) * ((upper - lower) / bins);
  }
 
  delta = (max - midpoint) / (bins - 1);
  lower = midpoint - delta / 2;
  upper = max + delta / 2;
  if ((binned = make_histogram(histogram, bins, lower, upper, data, points, 1))) {
    maxHistogram = 0;
    iMaximum = -1;
    for (i = 0; i < bins; i++) {
      if (maxHistogram < histogram[i]) {
        iMaximum = i;
        maxHistogram = histogram[i];
      }
    }
    meanBinned = ((double)binned) / bins;
    if ((maxHistogram > 1) && (maxHistogram > (meanBinned + sigmasRequired * sqrt(meanBinned))))
      *top = lower + (iMaximum + 0.5) * ((upper - lower) / bins);
  }
 
  if (*top == *base) {
    *base = min;
    *top = max;
  }
  return 1;
}