fr.jussieu.gla.wasa.core
Class ErrorMixer
java.lang.Object
|
+--fr.jussieu.gla.wasa.core.ErrorMixer
- public class ErrorMixer
- extends java.lang.Object
Customizable strategy for comparating two sets of scores issued from
Constraint's calculations. Developers may subclass
ErrorMixer to provide their own proble-dependant error calculation
strategies.
- Version:
- $Revision: 1.4 $ $Date: 2002/04/05 16:31:45 $
Method Summary |
int |
compare(float[] first,
float[] second)
Manipulates Error scores. |
boolean |
isUsuable(float f)
Tests if the given parameter is recognized as a valid value for an Error
score. |
double |
sum(double sum,
float[] errors)
|
Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
ErrorMixer
public ErrorMixer()
isUsuable
public final boolean isUsuable(float f)
- Tests if the given parameter is recognized as a valid value for an Error
score.
- Parameters:
f
- The float to test.- Returns:
- true if it represents a valid Error score, false
otherwise.
compare
public int compare(float[] first,
float[] second)
- Manipulates Error scores. The default behavior is to sum everything.
The default behavior is that the highest sum
is the best. This method could be overriden in order to provide custom
evaluation (e.g. : consider peak values, or rescale logarithmically,
etc). The order of the errors in the array should be considered as
meaningless.
Warning : developers implementing their own ErrorMixer
should be aware that Error scores may contain "unusuable" value equals to
Float.NaN
constant. A check should be performed on any
float coming from input parameters with the
isUsuable( float )
instance method.
- Parameters:
first
- An array of scores. A value of null means that the
Constraint which calculated the score did not affect the Variable.second
- Same meaning as first.- Returns:
- 0 if two arrays express the same overall score, < 0
if second array is better, > 0 if first score array is better.
sum
public double sum(double sum,
float[] errors)