Instead of increasing number of probes when retrieving candidates, increasing both the number of candidates to linearly look through and looking through the same points multiple times, instead generate a random point and hope to find a point that contains the correct filters and that it might be correct. This will be a way of accepting a might-be-correct result faster than just increasing number of probes.
Instead of increasing number of probes when retrieving candidates, increasing both the number of candidates to linearly look through and looking through the same points multiple times, instead generate a random point and hope to find a point that contains the correct filters and that it might be correct. This will be a way of accepting a might-be-correct result faster than just increasing number of probes.