MathUtils (h2o-core 3.42.0 API)

java.lang.Object
- water.util.MathUtils

public class MathUtils
extends java.lang.Object

Nested Class Summary

Nested Classes
Modifier and Type	Class and Description
`static class`	`MathUtils.BasicStats` Wrapper around weighted paralell basic stats computation (mean, variance)
`static class`	`MathUtils.ComputeAbsDiff`
`static class`	`MathUtils.DCT`
`static class`	`MathUtils.Norm`
`static class`	`MathUtils.SimpleStats`
`static class`	`MathUtils.SquareError`

Field Summary

Fields
Modifier and Type Field and Description

static double EULER_MASCHERONI_CONSTANT
Euler–Mascheroni constant (also called Euler's constant)

static long[] PRIMES

Fields
Modifier and Type	Field and Description
`static double`	`EULER_MASCHERONI_CONSTANT` Euler–Mascheroni constant (also called Euler's constant)
`static long[]`	`PRIMES`

Constructor Summary

Constructors
Constructor and Description

MathUtils()

Constructors
Constructor and Description
`MathUtils()`

Method Summary

All Methods Static Methods Concrete Methods
Modifier and Type	Method and Description
`static double`	`approxExp(double x)` Fast approximate exp
`static double`	`approxInvSqrt(double x)` Fast approximate 1./sqrt
`static float`	`approxInvSqrt(float x)` Fast approximate 1./sqrt
`static double`	`approxLog(double x)` Fast approximate log for values greater than 1, otherwise exact
`static double`	`approxSqrt(double x)` Fast approximate sqrt
`static float`	`approxSqrt(float x)` Fast approximate sqrt
`static float`	`approxSumSquares(float[] a, int from, int to)` Approximate sumSquares
`static double[][]`	`arrayTranspose(double[][] arr)`
`static int`	`combinatorial(int num, int d)`
`static boolean`	`compare(double a, double b, double absoluteTolerance, double relativeTolerance)` Compare 2 doubles within a tolerance.
`static int`	`compare(long x, long y)` Compare signed longs
`static int`	`compareUnsigned(long a, long b)` Copmarision of unsigned longs.
`static int`	`compareUnsigned(long hiA, long loA, long hiB, long loB)` Comparision of 128bit unsigned values represented by 2 longs
`static double`	`computeWeightedQuantile(Vec weight, Vec values, double alpha)`
`static java.math.BigInteger`	`convertDouble2BigInteger(double x)`
`static double[]`	`div(double[] nums, double n)`
`static float[]`	`div(float[] nums, float n)`
`static boolean`	`equalsWithinOneSmallUlp(double a, double b)`
`static boolean`	`equalsWithinOneSmallUlp(float a, float b)` Compare two numbers to see if they are within one ulp of the smaller decade.
`static boolean`	`equalsWithinRecSumErr(double actual, double expected, int n, double absum)`
`static double`	`innerProduct(double[] x, double[] y)`
`static double`	`l1norm(double[] x)`
`static double`	`l2norm(double[] x)`
`static double`	`l2norm2(double[] x)`
`static double`	`log2(double x)`
`static int`	`log2(int n)` Fast calculation of log base 2 for integers.
`static int`	`log2(long n)`
`static double`	`logFactorial(long y)`
`static double`	`logloss(double err)` Logloss
`static double[]`	`min_max_mean_stddev(long[] counts)`
`static double`	`roundToNDigits(double d, int n)`
`static double`	`sign(double d)`
`static double`	`sum(double[] from)`
`static float`	`sum(float[] from)`
`static float`	`sumSquares(float[] a)`
`static float`	`sumSquares(float[] a, int from, int to)`
`static double[]`	`wadd(double[] x, double[] y, double w)`
`static double`	`weightedSigma(long nobs, double wsum, double xSum, double xxSum)`
`static double`	`y_log_y(double y, double mu)`

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail
- EULER_MASCHERONI_CONSTANT
```
public static final double EULER_MASCHERONI_CONSTANT
```
  Euler–Mascheroni constant (also called Euler's constant)
  
  See Also:
  
  Constant Field Values
- PRIMES
```
public static final long[] PRIMES
```

Constructor Detail
- MathUtils
```
public MathUtils()
```

Method Detail

weightedSigma

public static double weightedSigma(long nobs,
                                   double wsum,
                                   double xSum,
                                   double xxSum)

logFactorial

public static double logFactorial(long y)

combinatorial

public static int combinatorial(int num,
                                int d)

computeWeightedQuantile

public static double computeWeightedQuantile(Vec weight,
                                             Vec values,
                                             double alpha)

approxSqrt
```
public static double approxSqrt(double x)
```
Fast approximate sqrt

Returns:

sqrt(x) with up to 5% relative error

convertDouble2BigInteger

public static java.math.BigInteger convertDouble2BigInteger(double x)

approxSqrt
```
public static float approxSqrt(float x)
```
Fast approximate sqrt

Returns:

sqrt(x) with up to 5% relative error

approxInvSqrt
```
public static double approxInvSqrt(double x)
```
Fast approximate 1./sqrt

Returns:

1./sqrt(x) with up to 2% relative error

approxInvSqrt
```
public static float approxInvSqrt(float x)
```
Fast approximate 1./sqrt

Returns:

1./sqrt(x) with up to 2% relative error

approxExp
```
public static double approxExp(double x)
```
Fast approximate exp

Returns:

exp(x) with up to 5% relative error

approxLog
```
public static double approxLog(double x)
```
Fast approximate log for values greater than 1, otherwise exact

Returns:

log(x) with up to 0.1% relative error

log2
```
public static int log2(int n)
```
Fast calculation of log base 2 for integers.

Returns:

log base 2 of n

log2
```
public static int log2(long n)
```

log2
```
public static double log2(double x)
```

div

public static float[] div(float[] nums,
                          float n)

div

public static double[] div(double[] nums,
                           double n)

sum

public static float sum(float[] from)

sum

public static double sum(double[] from)

sumSquares

public static float sumSquares(float[] a)

approxSumSquares
```
public static float approxSumSquares(float[] a,
                                     int from,
                                     int to)
```
Approximate sumSquares

Parameters:

a - Array with numbers

from - starting index (inclusive)

to - ending index (exclusive)

Returns:

approximate sum of squares based on a sample somewhere in the middle of the array (pos determined by bits of a[0])

sumSquares

public static float sumSquares(float[] a,
                               int from,
                               int to)

equalsWithinOneSmallUlp
```
public static boolean equalsWithinOneSmallUlp(float a,
                                              float b)
```
Compare two numbers to see if they are within one ulp of the smaller decade. Order of the arguments does not matter.

Parameters:

a - First number

b - Second number

Returns:

true if a and b are essentially equal, false otherwise.

equalsWithinOneSmallUlp

public static boolean equalsWithinOneSmallUlp(double a,
                                              double b)

equalsWithinRecSumErr

public static boolean equalsWithinRecSumErr(double actual,
                                            double expected,
                                            int n,
                                            double absum)

compare
```
public static boolean compare(double a,
                              double b,
                              double absoluteTolerance,
                              double relativeTolerance)
```
Compare 2 doubles within a tolerance. True iff the numbers difference within a absolute tolerance or within an relative tolerance.

Parameters:

a - double

b - double

absoluteTolerance - - Absolute allowed tolerance

relativeTolerance - - Relative allowed tolerance

Returns:

true if equal within tolerances

innerProduct

public static double innerProduct(double[] x,
                                  double[] y)

l2norm2

public static double l2norm2(double[] x)

l1norm

public static double l1norm(double[] x)

l2norm

public static double l2norm(double[] x)

wadd

public static double[] wadd(double[] x,
                            double[] y,
                            double w)

roundToNDigits

public static double roundToNDigits(double d,
                                    int n)

min_max_mean_stddev

public static double[] min_max_mean_stddev(long[] counts)

sign
```
public static double sign(double d)
```

y_log_y

public static double y_log_y(double y,
                             double mu)

compare

public static int compare(long x,
                          long y)

Compare signed longs

compareUnsigned

public static int compareUnsigned(long a,
                                  long b)

Copmarision of unsigned longs.

compareUnsigned

public static int compareUnsigned(long hiA,
                                  long loA,
                                  long hiB,
                                  long loB)

Comparision of 128bit unsigned values represented by 2 longs

logloss
```
public static double logloss(double err)
```
Logloss

Parameters:

err - prediction error (between 0 and 1)

Returns:

logloss

arrayTranspose

public static double[][] arrayTranspose(double[][] arr)

Class MathUtils

Nested Class Summary

Field Summary

Constructor Summary

Method Summary

Methods inherited from class java.lang.Object

Field Detail

EULER_MASCHERONI_CONSTANT

PRIMES

Constructor Detail

MathUtils

Method Detail

weightedSigma

logFactorial

combinatorial

computeWeightedQuantile

approxSqrt

convertDouble2BigInteger

approxSqrt

approxInvSqrt

approxInvSqrt

approxExp

approxLog

log2

log2

log2

div

div

sum

sum

sumSquares

approxSumSquares

sumSquares

equalsWithinOneSmallUlp

equalsWithinOneSmallUlp

equalsWithinRecSumErr

compare

innerProduct

l2norm2

l1norm

l2norm

wadd

roundToNDigits

min_max_mean_stddev

sign

y_log_y

compare

compareUnsigned

compareUnsigned

logloss

arrayTranspose