edu.stanford.nlp.math.ArrayMath Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of stanford-parser Show documentation
Show all versions of stanford-parser Show documentation
Stanford Parser processes raw text in English, Chinese, German, Arabic, and French, and extracts constituency parse trees.
The newest version!
package edu.stanford.nlp.math;
import java.io.IOException;
import java.text.NumberFormat;
import java.util.Arrays;
import java.util.Collection;
import java.util.List;
import java.util.Random;
import edu.stanford.nlp.io.IOUtils;
import edu.stanford.nlp.util.RuntimeInterruptedException;
import edu.stanford.nlp.util.StringUtils;
/**
* Methods for operating on numerical arrays as vectors and matrices.
*
* @author Teg Grenager
*/
public class ArrayMath {
private static final Random rand = new Random();
private ArrayMath() { } // not instantiable
// BASIC INFO -----------------------------------------------------------------
public static int numRows(double[] v) {
return v.length;
}
// GENERATION -----------------------------------------------------------------
/**
* Generate a range of integers from start (inclusive) to end (exclusive).
* Similar to the Python range() builtin function.
*
* @param start Beginning number (inclusive)
* @param end Ending number (exclusive)
* @return integers from [start...end)
*/
public static int[] range(int start, int end) {
assert end > start;
int len = end - start;
int[] range = new int[len];
for (int i = 0; i < range.length; ++i) {
range[i] = i+start;
}
return range;
}
// CASTS ----------------------------------------------------------------------
public static float[] doubleArrayToFloatArray(double[] a) {
float[] result = new float[a.length];
for (int i = 0; i < a.length; i++) {
result[i] = (float) a[i];
}
return result;
}
public static double[] floatArrayToDoubleArray(float[] a) {
double[] result = new double[a.length];
for (int i = 0; i < a.length; i++) {
result[i] = a[i];
}
return result;
}
public static double[][] floatArrayToDoubleArray(float[][] a) {
double[][] result = new double[a.length][];
for (int i = 0; i < a.length; i++) {
result[i] = new double[a[i].length];
for (int j = 0; j < a[i].length; j++) {
result[i][j] = a[i][j];
}
}
return result;
}
public static float[][] doubleArrayToFloatArray(double[][] a) {
float[][] result = new float[a.length][];
for (int i = 0; i < a.length; i++) {
result[i] = new float[a[i].length];
for (int j = 0; j < a[i].length; j++) {
result[i][j] = (float) a[i][j];
}
}
return result;
}
// OPERATIONS ON AN ARRAY - NONDESTRUCTIVE
public static double[] exp(double[] a) {
double[] result = new double[a.length];
for (int i = 0; i < a.length; i++) {
result[i] = Math.exp(a[i]);
}
return result;
}
public static double[] log(double[] a) {
double[] result = new double[a.length];
for (int i = 0; i < a.length; i++) {
result[i] = Math.log(a[i]);
}
return result;
}
// OPERATIONS ON AN ARRAY - DESTRUCTIVE
public static void expInPlace(double[] a) {
for (int i = 0; i < a.length; i++) {
a[i] = Math.exp(a[i]);
}
}
public static void logInPlace(double[] a) {
for (int i = 0; i < a.length; i++) {
a[i] = Math.log(a[i]);
}
}
public static double[] softmax(double[] scales) {
double[] newScales = new double[scales.length];
double sum = 0;
for (int i = 0; i < scales.length; i++) {
newScales[i] = Math.exp(scales[i]);
sum += newScales[i];
}
for (int i = 0; i < scales.length; i++) {
newScales[i] /= sum;
}
return newScales;
}
// OPERATIONS WITH SCALAR - DESTRUCTIVE
/**
* Increases the values in the first array a by b. Does it in place.
*
* @param a The array
* @param b The amount by which to increase each item
*/
public static void addInPlace(double[] a, double b) {
for (int i = 0; i < a.length; i++) {
a[i] = a[i] + b;
}
}
/**
* Increases the values in this array by b. Does it in place.
*
* @param a The array
* @param b The amount by which to increase each item
*/
public static void addInPlace(float[] a, double b) {
for (int i = 0; i < a.length; i++) {
a[i] = (float) (a[i] + b);
}
}
/**
* Add c times the array b to array a. Does it in place.
*/
public static void addMultInPlace(double[] a, double[] b, double c) {
for (int i=0; i list) {
int[] result = new int[list.size()];
int i = 0;
for (int v : list) {
result[i++] = v;
}
return result;
}
public static Double[] box(double[] assignment) {
Double[] result = new Double[assignment.length];
for (int i=0; i list) {
double[] result = new double[list.size()];
int i = 0;
for (double v : list) {
result[i++] = v;
}
return result;
}
public static int indexOf(int n, int[] a) {
for (int i=0; i 0; i--) {
int j = rand.nextInt(i+1); // a random index from 0 to i inclusive, may shuffle with itself
int tmp = a[i];
a[i] = a[j];
a[j] = tmp;
}
}
public static void reverse(int[] a) {
for (int i=0; i= testStatistic) successes++;
}
return (double) (successes + 1) / (double) (iterations + 1);
}
public static double sigLevelByApproxRand(int[] A, int[] B) {
return sigLevelByApproxRand(A, B, 1000);
}
public static double sigLevelByApproxRand(int[] A, int[] B, int iterations) {
if (A.length == 0)
throw new IllegalArgumentException("Input arrays must not be empty!");
if (A.length != B.length)
throw new IllegalArgumentException("Input arrays must have equal length!");
if (iterations <= 0)
throw new IllegalArgumentException("Number of iterations must be positive!");
double[] X = new double[A.length];
double[] Y = new double[B.length];
for (int i = 0; i < A.length; i++) {
X[i] = A[i];
Y[i] = B[i];
}
return sigLevelByApproxRand(X, Y, iterations);
}
public static double sigLevelByApproxRand(boolean[] A, boolean[] B) {
return sigLevelByApproxRand(A, B, 1000);
}
public static double sigLevelByApproxRand(boolean[] A, boolean[] B, int iterations) {
if (A.length == 0)
throw new IllegalArgumentException("Input arrays must not be empty!");
if (A.length != B.length)
throw new IllegalArgumentException("Input arrays must have equal length!");
if (iterations <= 0)
throw new IllegalArgumentException("Number of iterations must be positive!");
double[] X = new double[A.length];
double[] Y = new double[B.length];
for (int i = 0; i < A.length; i++) {
X[i] = (A[i] ? 1.0 : 0.0);
Y[i] = (B[i] ? 1.0 : 0.0);
}
return sigLevelByApproxRand(X, Y, iterations);
}
// Returns the absolute difference between the means of arrays A and B.
// If 'randomize' is true, swaps matched A & B entries with 50% probability
// Assumes input arrays have equal, non-zero length.
private static double absDiffOfMeans(double[] A, double[] B, boolean randomize) {
Random random = new Random();
double aTotal = 0.0;
double bTotal = 0.0;
for (int i = 0; i < A.length; i++) {
if (randomize && random.nextBoolean()) {
aTotal += B[i];
bTotal += A[i];
} else {
aTotal += A[i];
bTotal += B[i];
}
}
double aMean = aTotal / A.length;
double bMean = bTotal / B.length;
return Math.abs(aMean - bMean);
}
// PRINTING FUNCTIONS
public static String toBinaryString(byte[] b) {
StringBuilder s = new StringBuilder();
for (byte by : b) {
for (int j = 7; j >= 0; j--) {
if ((by & (1 << j)) > 0) {
s.append('1');
} else {
s.append('0');
}
}
s.append(' ');
}
return s.toString();
}
public static String toString(double[] a) {
return toString(a, null);
}
public static String toString(double[] a, NumberFormat nf) {
if (a == null) return null;
if (a.length == 0) return "[]";
StringBuilder b = new StringBuilder();
b.append('[');
for (int i = 0; i < a.length - 1; i++) {
String s;
if (nf == null) {
s = String.valueOf(a[i]);
} else {
s = nf.format(a[i]);
}
b.append(s);
b.append(", ");
}
String s;
if (nf == null) {
s = String.valueOf(a[a.length - 1]);
} else {
s = nf.format(a[a.length - 1]);
}
b.append(s);
b.append(']');
return b.toString();
}
public static String toString(float[] a) {
return toString(a, null);
}
public static String toString(float[] a, NumberFormat nf) {
if (a == null) return null;
if (a.length == 0) return "[]";
StringBuilder b = new StringBuilder();
b.append('[');
for (int i = 0; i < a.length - 1; i++) {
String s;
if (nf == null) {
s = String.valueOf(a[i]);
} else {
s = nf.format(a[i]);
}
b.append(s);
b.append(", ");
}
String s;
if (nf == null) {
s = String.valueOf(a[a.length - 1]);
} else {
s = nf.format(a[a.length - 1]);
}
b.append(s);
b.append(']');
return b.toString();
}
public static String toString(int[] a) {
return toString(a, null);
}
public static String toString(int[] a, NumberFormat nf) {
if (a == null) return null;
if (a.length == 0) return "[]";
StringBuilder b = new StringBuilder();
b.append('[');
for (int i = 0; i < a.length - 1; i++) {
String s;
if (nf == null) {
s = String.valueOf(a[i]);
} else {
s = nf.format(a[i]);
}
b.append(s);
b.append(", ");
}
String s;
if (nf == null) {
s = String.valueOf(a[a.length - 1]);
} else {
s = nf.format(a[a.length - 1]);
}
b.append(s);
b.append(']');
return b.toString();
}
public static String toString(byte[] a) {
return toString(a, null);
}
public static String toString(byte[] a, NumberFormat nf) {
if (a == null) return null;
if (a.length == 0) return "[]";
StringBuilder b = new StringBuilder();
b.append('[');
for (int i = 0; i < a.length - 1; i++) {
String s;
if (nf == null) {
s = String.valueOf(a[i]);
} else {
s = nf.format(a[i]);
}
b.append(s);
b.append(", ");
}
String s;
if (nf == null) {
s = String.valueOf(a[a.length - 1]);
} else {
s = nf.format(a[a.length - 1]);
}
b.append(s);
b.append(']');
return b.toString();
}
public static String toString(int[][] counts) {
return toString(counts, null, null, 10, 10, NumberFormat.getInstance(), false);
}
public static String toString(int[][] counts, Object[] rowLabels, Object[] colLabels, int labelSize, int cellSize, NumberFormat nf, boolean printTotals) {
// first compute row totals and column totals
if (counts.length==0 || counts[0].length==0) return "";
int[] rowTotals = new int[counts.length];
int[] colTotals = new int[counts[0].length]; // assume it's square
int total = 0;
for (int i = 0; i < counts.length; i++) {
for (int j = 0; j < counts[i].length; j++) {
rowTotals[i] += counts[i][j];
colTotals[j] += counts[i][j];
total += counts[i][j];
}
}
StringBuilder result = new StringBuilder();
// column labels
if (colLabels != null) {
result.append(StringUtils.padLeft("", labelSize)); // spacing for the row labels!
for (int j = 0; j < counts[0].length; j++) {
String s = (colLabels[j]==null ? "null" : colLabels[j].toString());
if (s.length() > cellSize - 1) {
s = s.substring(0, cellSize - 1);
}
s = StringUtils.padLeft(s, cellSize);
result.append(s);
}
if (printTotals) {
result.append(StringUtils.padLeftOrTrim("Total", cellSize));
}
result.append('\n');
}
for (int i = 0; i < counts.length; i++) {
// row label
if (rowLabels != null) {
String s = (rowLabels[i]==null ? "null" : rowLabels[i].toString());
s = StringUtils.padOrTrim(s, labelSize); // left align this guy only
result.append(s);
}
// value
for (int j = 0; j < counts[i].length; j++) {
result.append(StringUtils.padLeft(nf.format(counts[i][j]), cellSize));
}
// the row total
if (printTotals) {
result.append(StringUtils.padLeft(nf.format(rowTotals[i]), cellSize));
}
result.append('\n');
}
// the col totals
if (printTotals) {
result.append(StringUtils.pad("Total", labelSize));
for (int colTotal : colTotals) {
result.append(StringUtils.padLeft(nf.format(colTotal), cellSize));
}
result.append(StringUtils.padLeft(nf.format(total), cellSize));
}
return result.toString();
}
public static String toString(double[][] counts) {
return toString(counts, 10, null, null, NumberFormat.getInstance(), false);
}
public static String toString(double[][] counts, int cellSize, Object[] rowLabels, Object[] colLabels, NumberFormat nf, boolean printTotals) {
if (counts==null) return null;
// first compute row totals and column totals
double[] rowTotals = new double[counts.length];
double[] colTotals = new double[counts[0].length]; // assume it's square
double total = 0.0;
for (int i = 0; i < counts.length; i++) {
for (int j = 0; j < counts[i].length; j++) {
rowTotals[i] += counts[i][j];
colTotals[j] += counts[i][j];
total += counts[i][j];
}
}
StringBuilder result = new StringBuilder();
// column labels
if (colLabels != null) {
result.append(StringUtils.padLeft("", cellSize));
for (int j = 0; j < counts[0].length; j++) {
String s = colLabels[j].toString();
if (s.length() > cellSize - 1) {
s = s.substring(0, cellSize - 1);
}
s = StringUtils.padLeft(s, cellSize);
result.append(s);
}
if (printTotals) {
result.append(StringUtils.padLeftOrTrim("Total", cellSize));
}
result.append('\n');
}
for (int i = 0; i < counts.length; i++) {
// row label
if (rowLabels != null) {
String s = rowLabels[i].toString();
s = StringUtils.padOrTrim(s, cellSize); // left align this guy only
result.append(s);
}
// value
for (int j = 0; j < counts[i].length; j++) {
result.append(StringUtils.padLeft(nf.format(counts[i][j]), cellSize));
}
// the row total
if (printTotals) {
result.append(StringUtils.padLeft(nf.format(rowTotals[i]), cellSize));
}
result.append('\n');
}
// the col totals
if (printTotals) {
result.append(StringUtils.pad("Total", cellSize));
for (double colTotal : colTotals) {
result.append(StringUtils.padLeft(nf.format(colTotal), cellSize));
}
result.append(StringUtils.padLeft(nf.format(total), cellSize));
}
return result.toString();
}
public static String toString(float[][] counts) {
return toString(counts, 10, null, null, NumberFormat.getIntegerInstance(), false);
}
public static String toString(float[][] counts, int cellSize, Object[] rowLabels, Object[] colLabels, NumberFormat nf, boolean printTotals) {
// first compute row totals and column totals
double[] rowTotals = new double[counts.length];
double[] colTotals = new double[counts[0].length]; // assume it's square
double total = 0.0;
for (int i = 0; i < counts.length; i++) {
for (int j = 0; j < counts[i].length; j++) {
rowTotals[i] += counts[i][j];
colTotals[j] += counts[i][j];
total += counts[i][j];
}
}
StringBuilder result = new StringBuilder();
// column labels
if (colLabels != null) {
result.append(StringUtils.padLeft("", cellSize));
for (int j = 0; j < counts[0].length; j++) {
String s = colLabels[j].toString();
s = StringUtils.padLeftOrTrim(s, cellSize);
result.append(s);
}
if (printTotals) {
result.append(StringUtils.padLeftOrTrim("Total", cellSize));
}
result.append('\n');
}
for (int i = 0; i < counts.length; i++) {
// row label
if (rowLabels != null) {
String s = rowLabels[i].toString();
s = StringUtils.pad(s, cellSize); // left align this guy only
result.append(s);
}
// value
for (int j = 0; j < counts[i].length; j++) {
result.append(StringUtils.padLeft(nf.format(counts[i][j]), cellSize));
}
// the row total
if (printTotals) {
result.append(StringUtils.padLeft(nf.format(rowTotals[i]), cellSize));
}
result.append('\n');
}
// the col totals
if (printTotals) {
result.append(StringUtils.pad("Total", cellSize));
for (double colTotal : colTotals) {
result.append(StringUtils.padLeft(nf.format(colTotal), cellSize));
}
result.append(StringUtils.padLeft(nf.format(total), cellSize));
}
return result.toString();
}
/**
* For testing only.
* @param args Ignored
*/
public static void main(String[] args) {
Random random = new Random();
int length = 100;
double[] A = new double[length];
double[] B = new double[length];
double aAvg = 70.0;
double bAvg = 70.5;
for (int i = 0; i < length; i++) {
A[i] = aAvg + random.nextGaussian();
B[i] = bAvg + random.nextGaussian();
}
System.out.println("A has length " + A.length + " and mean " + mean(A));
System.out.println("B has length " + B.length + " and mean " + mean(B));
for (int t = 0; t < 10; t++) {
System.out.println("p-value: " + sigLevelByApproxRand(A, B));
}
}
public static int[][] deepCopy(int[][] counts) {
int[][] result = new int[counts.length][];
for (int i=0; i