org.deri.iris.utils
Class StandardFloatingPointComparator

java.lang.Object
  org.deri.iris.utils.StandardFloatingPointComparator

All Implemented Interfaces:

IFloatingPointComparator

public class StandardFloatingPointComparator
extends Object
implements IFloatingPointComparator

Utility class for floating point comparison. This class allows for round-off errors by using only the specified most significant bits of the operands, after allowing for scale.

Field Summary

static int	BINARY_DIGITS_OF_PRECISION_DOUBLE ieee754 (double) uses 52 bits to represent the significand.
static int	BINARY_DIGITS_OF_PRECISION_FLOAT ieee754 (float) uses 23 bits to represent the significand.
static int	LEAST_SIGNIFICANT_BINARY_DIGITS_TO_IGNORE_DOUBLE The standard number of least significant bits to ignore for a double type.
static int	LEAST_SIGNIFICANT_BINARY_DIGITS_TO_IGNORE_FLOAT The standard number of least significant bits to ignore for a float type.
double	MAX_DIFFERENCE_FROM_ONE This number is used to find the maximum difference between two numbers that can be considered logically the same.
static IFloatingPointComparator	mStandardDoubleInstance The standard comparator for double type.
static IFloatingPointComparator	mStandardFloatInstance The standard comparator for float types.

Constructor Summary

StandardFloatingPointComparator(int numberOfSignificantBits)
Constructor.

Method Summary

int	compare(double a, double b) An error-safe comparison in the java style.
boolean	equals(double a, double b) Test two double values for equality.
static IFloatingPointComparator	getDouble() Singleton-like method to get the standard setup for dealing with double type.
static IFloatingPointComparator	getFloat() Singleton-like method to get the standard setup for dealing with float type.
boolean	greater(double a, double b) Floating-point error safe comparison.
boolean	greaterOrEquals(double a, double b) Floating-point error safe comparison.
boolean	isIntValue(double value) Indicates whether a double value contains an integer or a number very, very close to an integer.
boolean	less(double a, double b) Floating-point error safe comparison.
boolean	lessOrEquals(double a, double b) Floating-point error safe comparison.
boolean	notEquals(double a, double b) Floating-point error safe comparison.

Methods inherited from class java.lang.Object

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

Field Detail

MAX_DIFFERENCE_FROM_ONE

public final double MAX_DIFFERENCE_FROM_ONE

This number is used to find the maximum difference between two numbers that can be considered logically the same. Any difference greater than this is considered significant and must therefore indicate different numbers. This number must be scaled as it indicates the maximum allowed difference from 1.0

BINARY_DIGITS_OF_PRECISION_DOUBLE

public static final int BINARY_DIGITS_OF_PRECISION_DOUBLE

ieee754 (double) uses 52 bits to represent the significand.

See Also:

Constant Field Values

BINARY_DIGITS_OF_PRECISION_FLOAT

public static final int BINARY_DIGITS_OF_PRECISION_FLOAT

ieee754 (float) uses 23 bits to represent the significand.

See Also:

Constant Field Values

LEAST_SIGNIFICANT_BINARY_DIGITS_TO_IGNORE_DOUBLE

public static final int LEAST_SIGNIFICANT_BINARY_DIGITS_TO_IGNORE_DOUBLE

The standard number of least significant bits to ignore for a double type.

See Also:

Constant Field Values

LEAST_SIGNIFICANT_BINARY_DIGITS_TO_IGNORE_FLOAT

public static final int LEAST_SIGNIFICANT_BINARY_DIGITS_TO_IGNORE_FLOAT

The standard number of least significant bits to ignore for a float type.

See Also:

Constant Field Values

mStandardDoubleInstance

public static final IFloatingPointComparator mStandardDoubleInstance

The standard comparator for double type.

mStandardFloatInstance

public static final IFloatingPointComparator mStandardFloatInstance

The standard comparator for float types.

Constructor Detail

StandardFloatingPointComparator

public StandardFloatingPointComparator(int numberOfSignificantBits)

Constructor.

Parameters:

numberOfSignificantBits - The number of significant bits in the significand to use when comparing values. If a non-standard setup is used then it is up to the user to set this value correctly.

Method Detail

getDouble

public static IFloatingPointComparator getDouble()

Singleton-like method to get the standard setup for dealing with double type.

Returns:

The standard double-configured FloatingPoint object.

getFloat

public static IFloatingPointComparator getFloat()

Singleton-like method to get the standard setup for dealing with float type.

Returns:

The standard float-configured FloatingPoint object.

compare

public int compare(double a,
                   double b)

Description copied from interface: IFloatingPointComparator

An error-safe comparison in the java style.

Specified by:

compare in interface IFloatingPointComparator

Parameters:

a - A double value

b - A double value

Returns:

-1 if a is significantly less than b, +1 if a is significantly greater than b, 0 if a and b are close enough.

less

public boolean less(double a,
                    double b)

Description copied from interface: IFloatingPointComparator

Floating-point error safe comparison.

Specified by:

less in interface IFloatingPointComparator

Parameters:

a - A double value

b - A double value

Returns:

true If a is less than b.

greater

public boolean greater(double a,
                       double b)

Description copied from interface: IFloatingPointComparator

Floating-point error safe comparison.

Specified by:

greater in interface IFloatingPointComparator

Parameters:

a - A double value

b - A double value

Returns:

true If a is greater than b.

greaterOrEquals

public boolean greaterOrEquals(double a,
                               double b)

Description copied from interface: IFloatingPointComparator

Floating-point error safe comparison.

Specified by:

greaterOrEquals in interface IFloatingPointComparator

Parameters:

a - A double value

b - A double value

Returns:

true If a is greater than or close enough to be equal to b.

lessOrEquals

public boolean lessOrEquals(double a,
                            double b)

Description copied from interface: IFloatingPointComparator

Floating-point error safe comparison.

Specified by:

lessOrEquals in interface IFloatingPointComparator

Parameters:

a - A double value

b - A double value

Returns:

true If a is less than or close enough to be equal to b.

equals

public boolean equals(double a,
                      double b)

Test two double values for equality. The two values are determined to be equal if they are bit-wise identical or their values lie very close together. In order to discover whether any difference between the two numbers is significant, a maximum allowed difference is calculated.

Specified by:

equals in interface IFloatingPointComparator

Parameters:

a - A double value

b - A double value

Returns:

true If a and b are exactly the same or close enough to being equal.

notEquals

public boolean notEquals(double a,
                         double b)

Description copied from interface: IFloatingPointComparator

Floating-point error safe comparison.

Specified by:

notEquals in interface IFloatingPointComparator

Returns:

true If a and b are significantly different.

isIntValue

public boolean isIntValue(double value)

Description copied from interface: IFloatingPointComparator

Indicates whether a double value contains an integer or a number very, very close to an integer.

Specified by:

isIntValue in interface IFloatingPointComparator

Parameters:

value - The value to test

Returns:

true If value holds an integer.