You need to order your collection on both x and y with x first (think about it as x and y forming an hypothetical number with x on the left side of the decimal point, and y on the right side: when you sort number in growing order, if the integral parts are equal, you sort on the decimal part).
Now, with an equal predicate, it will be difficult to sort values (you can only tell if they are equal, not if one is before another). Instead, you need to implement the comparable interface, with the following method:
public int compare(Object obj) {
if (obj == null || !(obj instanceof MyObject)) {
// raise an Exception
}
MyObject other = (MyObject)obj;
if (x < other.x) return -1;
if (this.x > other.x) return 1;
if (this.y < other.y) return -1;
if (this.y > other.y) return 1;
return 0;
}
If your array is sorted according to this comparison, you just need to keep the last entry with the same x in your array to get what you want. This means that you remove entries unless its successor has a different x.
This method is interesting of you don't want to keep your original data, but only keep the result: it will update the existing array in place, for a complexity of O(n) in time (not counting the sorting, which should happen anyway if I understand your question correctly).
Alternatively, the whole filtering can be achieved by applying a fold on your collection, where folding here is simply keeping the highest y for a same x (as you stated it precisely in your question). Because your collection is already sorted on x, it means that it is naturally partitioned on values of x, so you can build a new collection by accumulating the correct x for each partition in another array. For each element of the array, you compare it with the last inserted entry in your new array, if the x are the same, you replace it with the object with the highest y. If the x are different, you add it to the new array.
The advantage of this approach is that you don't need to change the sorting algorithm, but on the other hand you need a new array. This algorithm should therefore work in O(n) space and time.
Finally, this algorithm may be adapted to an in place update of the original array. It is slightly more complicated, but would let you avoid the extra allocation (crucial on embedded systems).
Pseudo code:
int idx = 0;
while (idx + 1 < Objarray.size()){
MyObj oi = Objarray.get(idx), on = Objarray.get(idx+1);
if (oi.x == on.x) {
if (on.y < oi.y)
Objarray.remove(idx++);
else
Objarray.remove(idx+1);
} else
idx++;
}
Note that while working in constant space, it might be slightly less efficient than the allocating algorithm, because of the way ArrayList works internally (though it should be better than using other usual container types).
x == x+1is never true. You probably meantx == x1and mixed it up with the fact that in your list, entries with the samexvalue are stored consecutively.equals(Object other), create aComparator.