org.simulator.math
Class N_Metric

java.lang.Object
  org.simulator.math.QualityMeasure
      org.simulator.math.N_Metric

All Implemented Interfaces:

Serializable

Direct Known Subclasses:

EuclideanDistance, ManhattanDistance, RelativeSquaredError

public class N_Metric
extends QualityMeasure

An implementation of an n-metric. An n-metric is basically the n^th root of the sum of the distances of every single element in two vectors (arrays), where this distance will always be exponentiated by the value of n.

Since:

1.0

Version:

$Rev: 392 $

Author:

Andreas Dräger, Roland Keller

See Also:

Serialized Form

Field Summary

protected double

root The n of the metric.

Fields inherited from class org.simulator.math.QualityMeasure

defaultValue, meanFunction

Constructor Summary

N_Metric()
Constructs a new NMetric with a default root of two.

N_Metric(double root)
Constructs a new NMetric with a customized root.

Method Summary

double	distance(MultiTable.Block.Column x, MultiTable.Block.Column y, double defaultValue) Returns the distance of the two vectors x and y with the given root.
double	distanceToZero(MultiTable.Block.Column x, double defaultValue)
double	getRoot() Returns the n of the metric.
void	setRoot(double root) Sets the root.

Methods inherited from class org.simulator.math.QualityMeasure

distance, distance, distance, getColumnDistances, getDefaultValue, getMeanFunction, setDefaultValue, setMeanFunction

Methods inherited from class java.lang.Object

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

Field Detail

root

protected double root

The n of the metric.

Constructor Detail

N_Metric

public N_Metric()

Constructs a new NMetric with a default root of two. This will result in the Euclidean distance. Other metrics can be used by either setting the root to another value or explicitly using the distance function where the root must be given as an argument.

N_Metric

public N_Metric(double root)

Constructs a new NMetric with a customized root. Depending on the values of root this results in different metrics. Some are especially important:

• one is the Manhattan norm or the city block metric.
• two is the Euclidean metric.
• Infinity is the maximum norm.

Parameters:

root -

Method Detail

distance

public double distance(MultiTable.Block.Column x,
                       MultiTable.Block.Column y,
                       double defaultValue)

Description copied from class: QualityMeasure

Returns the distance of the two vectors x and y with the given root. This may be the root in a formal way or a default value to be returned if the distance uses a non defined operation. If one array is longer than the other one additional values do not contribute to the distance. Double.NaN values are also ignored.

Specified by:

distance in class QualityMeasure

Parameters:

x - an array

y - another array

defaultValue - The value to be returned in cases in which no distance computation is possible.

Returns:

The distance between the two arrays x and y.

getRoot

public double getRoot()

Returns the n of the metric.

Returns:

n

setRoot

public void setRoot(double root)

Sets the root.

Parameters:

root -

distanceToZero

public double distanceToZero(MultiTable.Block.Column x,
                             double defaultValue)

Parameters:

x - expected

defaultValue -

Returns:

the distance of the column vector x to the point of origin.