summaryrefslogtreecommitdiff
path: root/src/main/java/org/apache/commons/math3/util/FastMathCalc.java
diff options
context:
space:
mode:
Diffstat (limited to 'src/main/java/org/apache/commons/math3/util/FastMathCalc.java')
-rw-r--r--src/main/java/org/apache/commons/math3/util/FastMathCalc.java678
1 files changed, 678 insertions, 0 deletions
diff --git a/src/main/java/org/apache/commons/math3/util/FastMathCalc.java b/src/main/java/org/apache/commons/math3/util/FastMathCalc.java
new file mode 100644
index 0000000..cf0f27c
--- /dev/null
+++ b/src/main/java/org/apache/commons/math3/util/FastMathCalc.java
@@ -0,0 +1,678 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.commons.math3.util;
+
+import org.apache.commons.math3.exception.DimensionMismatchException;
+
+import java.io.PrintStream;
+
+/**
+ * Class used to compute the classical functions tables.
+ *
+ * @since 3.0
+ */
+class FastMathCalc {
+
+ /**
+ * 0x40000000 - used to split a double into two parts, both with the low order bits cleared.
+ * Equivalent to 2^30.
+ */
+ private static final long HEX_40000000 = 0x40000000L; // 1073741824L
+
+ /** Factorial table, for Taylor series expansions. 0!, 1!, 2!, ... 19! */
+ private static final double FACT[] =
+ new double[] {
+ +1.0d, // 0
+ +1.0d, // 1
+ +2.0d, // 2
+ +6.0d, // 3
+ +24.0d, // 4
+ +120.0d, // 5
+ +720.0d, // 6
+ +5040.0d, // 7
+ +40320.0d, // 8
+ +362880.0d, // 9
+ +3628800.0d, // 10
+ +39916800.0d, // 11
+ +479001600.0d, // 12
+ +6227020800.0d, // 13
+ +87178291200.0d, // 14
+ +1307674368000.0d, // 15
+ +20922789888000.0d, // 16
+ +355687428096000.0d, // 17
+ +6402373705728000.0d, // 18
+ +121645100408832000.0d, // 19
+ };
+
+ /** Coefficients for slowLog. */
+ private static final double LN_SPLIT_COEF[][] = {
+ {2.0, 0.0},
+ {0.6666666269302368, 3.9736429850260626E-8},
+ {0.3999999761581421, 2.3841857910019882E-8},
+ {0.2857142686843872, 1.7029898543501842E-8},
+ {0.2222222089767456, 1.3245471311735498E-8},
+ {0.1818181574344635, 2.4384203044354907E-8},
+ {0.1538461446762085, 9.140260083262505E-9},
+ {0.13333332538604736, 9.220590270857665E-9},
+ {0.11764700710773468, 1.2393345855018391E-8},
+ {0.10526403784751892, 8.251545029714408E-9},
+ {0.0952233225107193, 1.2675934823758863E-8},
+ {0.08713622391223907, 1.1430250008909141E-8},
+ {0.07842259109020233, 2.404307984052299E-9},
+ {0.08371849358081818, 1.176342548272881E-8},
+ {0.030589580535888672, 1.2958646899018938E-9},
+ {0.14982303977012634, 1.225743062930824E-8},
+ };
+
+ /** Table start declaration. */
+ private static final String TABLE_START_DECL = " {";
+
+ /** Table end declaration. */
+ private static final String TABLE_END_DECL = " };";
+
+ /** Private Constructor. */
+ private FastMathCalc() {}
+
+ /**
+ * Build the sine and cosine tables.
+ *
+ * @param SINE_TABLE_A table of the most significant part of the sines
+ * @param SINE_TABLE_B table of the least significant part of the sines
+ * @param COSINE_TABLE_A table of the most significant part of the cosines
+ * @param COSINE_TABLE_B table of the most significant part of the cosines
+ * @param SINE_TABLE_LEN length of the tables
+ * @param TANGENT_TABLE_A table of the most significant part of the tangents
+ * @param TANGENT_TABLE_B table of the most significant part of the tangents
+ */
+ @SuppressWarnings("unused")
+ private static void buildSinCosTables(
+ double[] SINE_TABLE_A,
+ double[] SINE_TABLE_B,
+ double[] COSINE_TABLE_A,
+ double[] COSINE_TABLE_B,
+ int SINE_TABLE_LEN,
+ double[] TANGENT_TABLE_A,
+ double[] TANGENT_TABLE_B) {
+ final double result[] = new double[2];
+
+ /* Use taylor series for 0 <= x <= 6/8 */
+ for (int i = 0; i < 7; i++) {
+ double x = i / 8.0;
+
+ slowSin(x, result);
+ SINE_TABLE_A[i] = result[0];
+ SINE_TABLE_B[i] = result[1];
+
+ slowCos(x, result);
+ COSINE_TABLE_A[i] = result[0];
+ COSINE_TABLE_B[i] = result[1];
+ }
+
+ /* Use angle addition formula to complete table to 13/8, just beyond pi/2 */
+ for (int i = 7; i < SINE_TABLE_LEN; i++) {
+ double xs[] = new double[2];
+ double ys[] = new double[2];
+ double as[] = new double[2];
+ double bs[] = new double[2];
+ double temps[] = new double[2];
+
+ if ((i & 1) == 0) {
+ // Even, use double angle
+ xs[0] = SINE_TABLE_A[i / 2];
+ xs[1] = SINE_TABLE_B[i / 2];
+ ys[0] = COSINE_TABLE_A[i / 2];
+ ys[1] = COSINE_TABLE_B[i / 2];
+
+ /* compute sine */
+ splitMult(xs, ys, result);
+ SINE_TABLE_A[i] = result[0] * 2.0;
+ SINE_TABLE_B[i] = result[1] * 2.0;
+
+ /* Compute cosine */
+ splitMult(ys, ys, as);
+ splitMult(xs, xs, temps);
+ temps[0] = -temps[0];
+ temps[1] = -temps[1];
+ splitAdd(as, temps, result);
+ COSINE_TABLE_A[i] = result[0];
+ COSINE_TABLE_B[i] = result[1];
+ } else {
+ xs[0] = SINE_TABLE_A[i / 2];
+ xs[1] = SINE_TABLE_B[i / 2];
+ ys[0] = COSINE_TABLE_A[i / 2];
+ ys[1] = COSINE_TABLE_B[i / 2];
+ as[0] = SINE_TABLE_A[i / 2 + 1];
+ as[1] = SINE_TABLE_B[i / 2 + 1];
+ bs[0] = COSINE_TABLE_A[i / 2 + 1];
+ bs[1] = COSINE_TABLE_B[i / 2 + 1];
+
+ /* compute sine */
+ splitMult(xs, bs, temps);
+ splitMult(ys, as, result);
+ splitAdd(result, temps, result);
+ SINE_TABLE_A[i] = result[0];
+ SINE_TABLE_B[i] = result[1];
+
+ /* Compute cosine */
+ splitMult(ys, bs, result);
+ splitMult(xs, as, temps);
+ temps[0] = -temps[0];
+ temps[1] = -temps[1];
+ splitAdd(result, temps, result);
+ COSINE_TABLE_A[i] = result[0];
+ COSINE_TABLE_B[i] = result[1];
+ }
+ }
+
+ /* Compute tangent = sine/cosine */
+ for (int i = 0; i < SINE_TABLE_LEN; i++) {
+ double xs[] = new double[2];
+ double ys[] = new double[2];
+ double as[] = new double[2];
+
+ as[0] = COSINE_TABLE_A[i];
+ as[1] = COSINE_TABLE_B[i];
+
+ splitReciprocal(as, ys);
+
+ xs[0] = SINE_TABLE_A[i];
+ xs[1] = SINE_TABLE_B[i];
+
+ splitMult(xs, ys, as);
+
+ TANGENT_TABLE_A[i] = as[0];
+ TANGENT_TABLE_B[i] = as[1];
+ }
+ }
+
+ /**
+ * For x between 0 and pi/4 compute cosine using Talor series cos(x) = 1 - x^2/2! + x^4/4! ...
+ *
+ * @param x number from which cosine is requested
+ * @param result placeholder where to put the result in extended precision (may be null)
+ * @return cos(x)
+ */
+ static double slowCos(final double x, final double result[]) {
+
+ final double xs[] = new double[2];
+ final double ys[] = new double[2];
+ final double facts[] = new double[2];
+ final double as[] = new double[2];
+ split(x, xs);
+ ys[0] = ys[1] = 0.0;
+
+ for (int i = FACT.length - 1; i >= 0; i--) {
+ splitMult(xs, ys, as);
+ ys[0] = as[0];
+ ys[1] = as[1];
+
+ if ((i & 1) != 0) { // skip odd entries
+ continue;
+ }
+
+ split(FACT[i], as);
+ splitReciprocal(as, facts);
+
+ if ((i & 2) != 0) { // alternate terms are negative
+ facts[0] = -facts[0];
+ facts[1] = -facts[1];
+ }
+
+ splitAdd(ys, facts, as);
+ ys[0] = as[0];
+ ys[1] = as[1];
+ }
+
+ if (result != null) {
+ result[0] = ys[0];
+ result[1] = ys[1];
+ }
+
+ return ys[0] + ys[1];
+ }
+
+ /**
+ * For x between 0 and pi/4 compute sine using Taylor expansion: sin(x) = x - x^3/3! + x^5/5! -
+ * x^7/7! ...
+ *
+ * @param x number from which sine is requested
+ * @param result placeholder where to put the result in extended precision (may be null)
+ * @return sin(x)
+ */
+ static double slowSin(final double x, final double result[]) {
+ final double xs[] = new double[2];
+ final double ys[] = new double[2];
+ final double facts[] = new double[2];
+ final double as[] = new double[2];
+ split(x, xs);
+ ys[0] = ys[1] = 0.0;
+
+ for (int i = FACT.length - 1; i >= 0; i--) {
+ splitMult(xs, ys, as);
+ ys[0] = as[0];
+ ys[1] = as[1];
+
+ if ((i & 1) == 0) { // Ignore even numbers
+ continue;
+ }
+
+ split(FACT[i], as);
+ splitReciprocal(as, facts);
+
+ if ((i & 2) != 0) { // alternate terms are negative
+ facts[0] = -facts[0];
+ facts[1] = -facts[1];
+ }
+
+ splitAdd(ys, facts, as);
+ ys[0] = as[0];
+ ys[1] = as[1];
+ }
+
+ if (result != null) {
+ result[0] = ys[0];
+ result[1] = ys[1];
+ }
+
+ return ys[0] + ys[1];
+ }
+
+ /**
+ * For x between 0 and 1, returns exp(x), uses extended precision
+ *
+ * @param x argument of exponential
+ * @param result placeholder where to place exp(x) split in two terms for extra precision (i.e.
+ * exp(x) = result[0] + result[1]
+ * @return exp(x)
+ */
+ static double slowexp(final double x, final double result[]) {
+ final double xs[] = new double[2];
+ final double ys[] = new double[2];
+ final double facts[] = new double[2];
+ final double as[] = new double[2];
+ split(x, xs);
+ ys[0] = ys[1] = 0.0;
+
+ for (int i = FACT.length - 1; i >= 0; i--) {
+ splitMult(xs, ys, as);
+ ys[0] = as[0];
+ ys[1] = as[1];
+
+ split(FACT[i], as);
+ splitReciprocal(as, facts);
+
+ splitAdd(ys, facts, as);
+ ys[0] = as[0];
+ ys[1] = as[1];
+ }
+
+ if (result != null) {
+ result[0] = ys[0];
+ result[1] = ys[1];
+ }
+
+ return ys[0] + ys[1];
+ }
+
+ /**
+ * Compute split[0], split[1] such that their sum is equal to d, and split[0] has its 30 least
+ * significant bits as zero.
+ *
+ * @param d number to split
+ * @param split placeholder where to place the result
+ */
+ private static void split(final double d, final double split[]) {
+ if (d < 8e298 && d > -8e298) {
+ final double a = d * HEX_40000000;
+ split[0] = (d + a) - a;
+ split[1] = d - split[0];
+ } else {
+ final double a = d * 9.31322574615478515625E-10;
+ split[0] = (d + a - d) * HEX_40000000;
+ split[1] = d - split[0];
+ }
+ }
+
+ /**
+ * Recompute a split.
+ *
+ * @param a input/out array containing the split, changed on output
+ */
+ private static void resplit(final double a[]) {
+ final double c = a[0] + a[1];
+ final double d = -(c - a[0] - a[1]);
+
+ if (c < 8e298 && c > -8e298) { // MAGIC NUMBER
+ double z = c * HEX_40000000;
+ a[0] = (c + z) - z;
+ a[1] = c - a[0] + d;
+ } else {
+ double z = c * 9.31322574615478515625E-10;
+ a[0] = (c + z - c) * HEX_40000000;
+ a[1] = c - a[0] + d;
+ }
+ }
+
+ /**
+ * Multiply two numbers in split form.
+ *
+ * @param a first term of multiplication
+ * @param b second term of multiplication
+ * @param ans placeholder where to put the result
+ */
+ private static void splitMult(double a[], double b[], double ans[]) {
+ ans[0] = a[0] * b[0];
+ ans[1] = a[0] * b[1] + a[1] * b[0] + a[1] * b[1];
+
+ /* Resplit */
+ resplit(ans);
+ }
+
+ /**
+ * Add two numbers in split form.
+ *
+ * @param a first term of addition
+ * @param b second term of addition
+ * @param ans placeholder where to put the result
+ */
+ private static void splitAdd(final double a[], final double b[], final double ans[]) {
+ ans[0] = a[0] + b[0];
+ ans[1] = a[1] + b[1];
+
+ resplit(ans);
+ }
+
+ /**
+ * Compute the reciprocal of in. Use the following algorithm. in = c + d. want to find x + y
+ * such that x+y = 1/(c+d) and x is much larger than y and x has several zero bits on the right.
+ *
+ * <p>Set b = 1/(2^22), a = 1 - b. Thus (a+b) = 1. Use following identity to compute (a+b)/(c+d)
+ *
+ * <p>(a+b)/(c+d) = a/c + (bc - ad) / (c^2 + cd) set x = a/c and y = (bc - ad) / (c^2 + cd) This
+ * will be close to the right answer, but there will be some rounding in the calculation of X.
+ * So by carefully computing 1 - (c+d)(x+y) we can compute an error and add that back in. This
+ * is done carefully so that terms of similar size are subtracted first.
+ *
+ * @param in initial number, in split form
+ * @param result placeholder where to put the result
+ */
+ static void splitReciprocal(final double in[], final double result[]) {
+ final double b = 1.0 / 4194304.0;
+ final double a = 1.0 - b;
+
+ if (in[0] == 0.0) {
+ in[0] = in[1];
+ in[1] = 0.0;
+ }
+
+ result[0] = a / in[0];
+ result[1] = (b * in[0] - a * in[1]) / (in[0] * in[0] + in[0] * in[1]);
+
+ if (result[1] != result[1]) { // can happen if result[1] is NAN
+ result[1] = 0.0;
+ }
+
+ /* Resplit */
+ resplit(result);
+
+ for (int i = 0; i < 2; i++) {
+ /* this may be overkill, probably once is enough */
+ double err =
+ 1.0
+ - result[0] * in[0]
+ - result[0] * in[1]
+ - result[1] * in[0]
+ - result[1] * in[1];
+ /*err = 1.0 - err; */
+ err *= result[0] + result[1];
+ /*printf("err = %16e\n", err); */
+ result[1] += err;
+ }
+ }
+
+ /**
+ * Compute (a[0] + a[1]) * (b[0] + b[1]) in extended precision.
+ *
+ * @param a first term of the multiplication
+ * @param b second term of the multiplication
+ * @param result placeholder where to put the result
+ */
+ private static void quadMult(final double a[], final double b[], final double result[]) {
+ final double xs[] = new double[2];
+ final double ys[] = new double[2];
+ final double zs[] = new double[2];
+
+ /* a[0] * b[0] */
+ split(a[0], xs);
+ split(b[0], ys);
+ splitMult(xs, ys, zs);
+
+ result[0] = zs[0];
+ result[1] = zs[1];
+
+ /* a[0] * b[1] */
+ split(b[1], ys);
+ splitMult(xs, ys, zs);
+
+ double tmp = result[0] + zs[0];
+ result[1] -= tmp - result[0] - zs[0];
+ result[0] = tmp;
+ tmp = result[0] + zs[1];
+ result[1] -= tmp - result[0] - zs[1];
+ result[0] = tmp;
+
+ /* a[1] * b[0] */
+ split(a[1], xs);
+ split(b[0], ys);
+ splitMult(xs, ys, zs);
+
+ tmp = result[0] + zs[0];
+ result[1] -= tmp - result[0] - zs[0];
+ result[0] = tmp;
+ tmp = result[0] + zs[1];
+ result[1] -= tmp - result[0] - zs[1];
+ result[0] = tmp;
+
+ /* a[1] * b[0] */
+ split(a[1], xs);
+ split(b[1], ys);
+ splitMult(xs, ys, zs);
+
+ tmp = result[0] + zs[0];
+ result[1] -= tmp - result[0] - zs[0];
+ result[0] = tmp;
+ tmp = result[0] + zs[1];
+ result[1] -= tmp - result[0] - zs[1];
+ result[0] = tmp;
+ }
+
+ /**
+ * Compute exp(p) for a integer p in extended precision.
+ *
+ * @param p integer whose exponential is requested
+ * @param result placeholder where to put the result in extended precision
+ * @return exp(p) in standard precision (equal to result[0] + result[1])
+ */
+ static double expint(int p, final double result[]) {
+ // double x = M_E;
+ final double xs[] = new double[2];
+ final double as[] = new double[2];
+ final double ys[] = new double[2];
+ // split(x, xs);
+ // xs[1] = (double)(2.7182818284590452353602874713526625L - xs[0]);
+ // xs[0] = 2.71827697753906250000;
+ // xs[1] = 4.85091998273542816811e-06;
+ // xs[0] = Double.longBitsToDouble(0x4005bf0800000000L);
+ // xs[1] = Double.longBitsToDouble(0x3ed458a2bb4a9b00L);
+
+ /* E */
+ xs[0] = 2.718281828459045;
+ xs[1] = 1.4456468917292502E-16;
+
+ split(1.0, ys);
+
+ while (p > 0) {
+ if ((p & 1) != 0) {
+ quadMult(ys, xs, as);
+ ys[0] = as[0];
+ ys[1] = as[1];
+ }
+
+ quadMult(xs, xs, as);
+ xs[0] = as[0];
+ xs[1] = as[1];
+
+ p >>= 1;
+ }
+
+ if (result != null) {
+ result[0] = ys[0];
+ result[1] = ys[1];
+
+ resplit(result);
+ }
+
+ return ys[0] + ys[1];
+ }
+
+ /**
+ * xi in the range of [1, 2]. 3 5 7 x+1 / x x x \ ln ----- = 2 * | x + ---- + ---- + ---- + ...
+ * | 1-x \ 3 5 7 /
+ *
+ * <p>So, compute a Remez approximation of the following function
+ *
+ * <p>ln ((sqrt(x)+1)/(1-sqrt(x))) / x
+ *
+ * <p>This will be an even function with only positive coefficents. x is in the range [0 - 1/3].
+ *
+ * <p>Transform xi for input to the above function by setting x = (xi-1)/(xi+1). Input to the
+ * polynomial is x^2, then the result is multiplied by x.
+ *
+ * @param xi number from which log is requested
+ * @return log(xi)
+ */
+ static double[] slowLog(double xi) {
+ double x[] = new double[2];
+ double x2[] = new double[2];
+ double y[] = new double[2];
+ double a[] = new double[2];
+
+ split(xi, x);
+
+ /* Set X = (x-1)/(x+1) */
+ x[0] += 1.0;
+ resplit(x);
+ splitReciprocal(x, a);
+ x[0] -= 2.0;
+ resplit(x);
+ splitMult(x, a, y);
+ x[0] = y[0];
+ x[1] = y[1];
+
+ /* Square X -> X2*/
+ splitMult(x, x, x2);
+
+ // x[0] -= 1.0;
+ // resplit(x);
+
+ y[0] = LN_SPLIT_COEF[LN_SPLIT_COEF.length - 1][0];
+ y[1] = LN_SPLIT_COEF[LN_SPLIT_COEF.length - 1][1];
+
+ for (int i = LN_SPLIT_COEF.length - 2; i >= 0; i--) {
+ splitMult(y, x2, a);
+ y[0] = a[0];
+ y[1] = a[1];
+ splitAdd(y, LN_SPLIT_COEF[i], a);
+ y[0] = a[0];
+ y[1] = a[1];
+ }
+
+ splitMult(y, x, a);
+ y[0] = a[0];
+ y[1] = a[1];
+
+ return y;
+ }
+
+ /**
+ * Print an array.
+ *
+ * @param out text output stream where output should be printed
+ * @param name array name
+ * @param expectedLen expected length of the array
+ * @param array2d array data
+ */
+ static void printarray(PrintStream out, String name, int expectedLen, double[][] array2d) {
+ out.println(name);
+ checkLen(expectedLen, array2d.length);
+ out.println(TABLE_START_DECL + " ");
+ int i = 0;
+ for (double[] array : array2d) { // "double array[]" causes PMD parsing error
+ out.print(" {");
+ for (double d : array) { // assume inner array has very few entries
+ out.printf("%-25.25s", format(d)); // multiple entries per line
+ }
+ out.println("}, // " + i++);
+ }
+ out.println(TABLE_END_DECL);
+ }
+
+ /**
+ * Print an array.
+ *
+ * @param out text output stream where output should be printed
+ * @param name array name
+ * @param expectedLen expected length of the array
+ * @param array array data
+ */
+ static void printarray(PrintStream out, String name, int expectedLen, double[] array) {
+ out.println(name + "=");
+ checkLen(expectedLen, array.length);
+ out.println(TABLE_START_DECL);
+ for (double d : array) {
+ out.printf(" %s%n", format(d)); // one entry per line
+ }
+ out.println(TABLE_END_DECL);
+ }
+
+ /**
+ * Format a double.
+ *
+ * @param d double number to format
+ * @return formatted number
+ */
+ static String format(double d) {
+ if (d != d) {
+ return "Double.NaN,";
+ } else {
+ return ((d >= 0) ? "+" : "") + Double.toString(d) + "d,";
+ }
+ }
+
+ /**
+ * Check two lengths are equal.
+ *
+ * @param expectedLen expected length
+ * @param actual actual length
+ * @exception DimensionMismatchException if the two lengths are not equal
+ */
+ private static void checkLen(int expectedLen, int actual) throws DimensionMismatchException {
+ if (expectedLen != actual) {
+ throw new DimensionMismatchException(actual, expectedLen);
+ }
+ }
+}