iverase commented on a change in pull request #905: LUCENE-8990: Add
estimateDocCount(visitor) method to PointValues
URL: https://github.com/apache/lucene-solr/pull/905#discussion_r331353384
##########
File path: lucene/core/src/java/org/apache/lucene/index/PointValues.java
##########
@@ -241,9 +241,28 @@ default void visit(DocIdSetIterator iterator, byte[]
packedValue) throws IOExcep
* than {@link #intersect(IntersectVisitor)}.
* @see DocIdSetIterator#cost */
public long estimateDocCount(IntersectVisitor visitor) {
- return (long) Math.ceil(estimatePointCount(visitor) / ((double) size() /
getDocCount()));
+ long estimatedPointCount = estimatePointCount(visitor);
+ int docCount = getDocCount();
+ double size = size();
+ if (estimatedPointCount >= size) {
+ // math all docs
+ return docCount;
+ } else if (size == docCount || estimatedPointCount == 0L ) {
+ // if the point count estimate is 0 or we have only single values
+ // return this estimate
+ return estimatedPointCount;
+ } else {
+ // in case of multi values estimate the number of docs using the
solution provided in
+ //
https://math.stackexchange.com/questions/1175295/urn-problem-probability-of-drawing-balls-of-k-unique-colors
+ // then approximate the solution for points per doc << size() which
results in the expression
+ // D * (1 - ((N - n) / N)^(N/D))
Review comment:
It is clarify in the following line.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
