org.python.core.util

Class ExtraMath

java.lang.Object

org.python.core.util.ExtraMath

public class ExtraMath
extends Object

A static utility class with two additional math functions.

Field Summary

Fields

Modifier and Type	Field and Description
static double	CLOSE
static double	EPSILON

Constructor Summary

Constructors

Constructor and Description
ExtraMath()

Method Summary

Methods

Modifier and Type	Method and Description
static boolean	close(double v, double w)
static boolean	close(double v, double w, double tol) Are v and w "close" to each other?
static double	closeFloor(double v) Returns floor(v) except when v is very close to the next number, when it returns ceil(v);
static double	round(double x, int n) Round the argument x to n decimal places.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

EPSILON

public static double EPSILON

CLOSE

public static double CLOSE

Constructor Detail

ExtraMath

public ExtraMath()

Method Detail

close

public static boolean close(double v,
            double w,
            double tol)

Are v and w "close" to each other? Uses a scaled tolerance.

close

public static boolean close(double v,
            double w)

closeFloor

public static double closeFloor(double v)

Returns floor(v) except when v is very close to the next number, when it returns ceil(v);

round

public static double round(double x,
           int n)

Round the argument x to n decimal places. (Rounding is half-up in Python 2.) The method uses BigDecimal, to compute r(x*10ⁿ)*10^-n, where r() round to the nearest integer. It takes some short-cuts for extreme values.

For sufficiently small x*10ⁿ, the rounding is to zero, and the return value is a signed zero (same sign as x). Suppose x = a*2^b, where the significand we must have a<2. Sufficiently small means such that n log₂10 < -(b+2).

For sufficiently large x*10ⁿ, the adjustment of rounding is too small to affect the least significant bit. That is a*2^b represents an amount greater than one, and rounding no longer affects the value, and the return is x. Since the matissa has 52 fractional bits, sufficiently large means such that n log₂10 > 52-b.

Parameters:

x - to round

n - decimal places

Returns:

x rounded.