How to "quick search" in a sorted list ?

Question

Hi all,
I need that your advice to choose the best collection to modelize my data.
Let me describe you the context of my problem:
I work on an antenna pattern file which looks like this:
Elevation (deg);Gain (dB)
0;5
3;45
6;3.1
10;6.5
16;18.1
20;33.1
24;31.2
30;3.5 I need, for a given elevation angle, to get the related gain. if the exact value of the angle is not found, we will return the gain of the nearest angle.
For example:
If the elevation angle is 5 degrees, I must return 3.1 dB.This operation should be called a big number of time (about many hundred calls)
What are your advices to modelize my data ? What is the best collection to do it ? What is the best way to quick search in this data to get the gain realted to a given angle ?
I though to use the binary search tree. What do you think about this solution ? Do you known of the Java API contains an implementation of the binary search trees ?
I though too to use an hashset. The hashcode should be computed in order to access to the nearest angle in the list of values. But I did not find the alogrithm to compute a such hash code. Maybe it is a wrong way.
Please, help me.
Tell me your opinion and your advices to solve my problem.
Let me know if you need more information
Thanks

Answer 1

OK, it works.
I created a class for my data:public class ElGainPatternElement
  private double elevation;
  private double gain;
+getters/setters
}Then, I created a comparator to compare my elements on the elevation attribute.
Finally, I specialiazed the Collections.binarySearch method in order to return the value instead of the index.
I the object is not found, I return the inferior (or superior) object.
public static <T> T binarySearch(List<? extends T> list, T key,
      Comparator<? super T> c, boolean isMinStrategy)
    int index = Collections.binarySearch(list, key, c);
    // si objet non trouv�
    if (index < 0)
      // r�cup�rer l'index le plus proche de celui recherch�
      index = isMinStrategy ? (-index - 2) : (-index - 1);
      // si plus grand que tous les �lts, retourner le dernier de la liste
      if (index >= list.size())
        index = list.size() - 1;
      // si plus grand que tous les �lts, retourner le premier de la liste
      else if (index < 0)
        index = 0;
    // retourner l'objet le plus proche au regard de la strat�gie
    return list.get(index);
  }It works very well.
But now, I need to change the strategy to use if the wanted elevation is not in the list.
If you look at the example of my first post, the 5 degrees of elevation is not in my list. So, I have to interpolate the values of the gain in order to get a correct gain.
If possible, it should use the least squares interpolation (I'm not sure of my traduction of "moindres carr�s").
I have really no idea of the way to do it.
Can anyone help me with this interpolation ? I have never use algorithm of interpolation (except the mean of 2 values).
Maybe the Java API provides some utilities that can help me to interpolate my values ... any suggestions ?
Thanks