Optimize bloomfilter issue 7346 #7494
Conversation
Refactored optimalNumOfBits and optimalNumOfHashFunctions to improve efficiency and clarity.
| static int optimalNumOfHashFunctions(long n, long m) { | ||
| // (m / n) * log(2), but avoid truncation due to division! | ||
| return max(1, (int) Math.round((double) m / n * Math.log(2))); | ||
| static int optimalNumOfHashFunctions(double p) { |
There was a problem hiding this comment.
The change here replaces the optimalNumOfHashFunctions method, which originally took parameters n (expected insertions) and m (total number of bits), with a new version that instead uses p (desired false positive probability). While this new approach aligns with cases where p is known, it reduces flexibility for use cases where m and n are more readily available or preferred for direct configuration.
To maintain backward compatibility and allow both methods of calculation, it would be more effective to overload optimalNumOfHashFunctions rather than replacing it. Overloading the method would allow users to specify either n and m or p as needed, based on their use case, without introducing breaking changes.
Context
This PR optimizes the calculation of the number of hash functions and bits used in BloomFilter, reducing redundant computations and improving readability.
Changes
Implemented a more efficient algorithm for
optimalNumOfBitsReplaced hash calculations in
optimalNumOfHashFunctionsUpdated tests to reflect these changes
Impact
Improves performance when using
BloomFilter.Tests affected
Tests have been modified to validate the new calculation methods.