integer compression method

Quote:

have a large array with a range of integers that are mostly not continuous, eg 1-100,1-1000 etc.

This is not clear to me.

Quote:

All numbers are unique and different increasing.

If the numbers belongs to a range (e.g. 1-10000) then you may use just the necessary bits to store them (14 bits are enough for storing 0-16383).
For number increasing monotonically you may store the first number and then the differences between the next and the previous. Storing such differences usually requires fewer bits than storing the whole numbers.
Of course you may also try an available, general compression facility like the one suggested by Richard.

If your numbers are increasing, compute the differences, as CPallini said, and observe the histogram of these deltas.

If the deltas are repeating often enough, you can use Huffman encoding on them (most frequent values encoded with fewer bits).

If you compute the "entropy" of the histogram, that can tell you how many bits of information will remain after compression.