How does the cypher query optimizer estimate number of rows if I have a predicate like "40 > n.age > 30" without any index?

[Sun Yuhan]

If such a query is given " match (n:PEOPLE) where 40 > n.age > 30 return n", how does the query optimizer estimate the number of rows returned by the query? I know for relational database, they have histogram information about distribution of values for each attribute. But for a graph database, how does this estimation is made?

Or a point to the corresponding page of source code is really helpful. Thx!

[Michael Hunger]

We currently only have the selectivity of the index, histograms on value distributions are planned for some point in the future.

So I think it mostly multiplies the total number by the selectivity, but I'm not the expert there.

Perhaps Petra can point you to the right place in the code?

Cheers, Michael 

On Wed, Jul 19, 2017 at 9:53 PM, Sun Yuhan <wdmz...@gmail.com> wrote:

If such a query is given " match (n:PEOPLE) where 40 > n.age > 30 return n", how does the query optimizer estimate the number of rows returned by the query? I know for relational database, they have histogram information about distribution of values for each attribute. But for a graph database, how does this estimation is made?

Or a point to the corresponding page of source code is really helpful. Thx!

--
You received this message because you are subscribed to the Google Groups "Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neo4j+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Petra Selmer]

Hi Sun

This is a good place to start: https://github.com/neo4j/neo4j/blob/3.2/community/cypher/cypher-compiler-3.2/src/main/scala/org/neo4j/cypher/internal/compiler/v3_2/planner/logical/CardinalityCostModel.scala

Cheers

Petra

Petra Selmer

Engineer and member of the Cypher Language Group

Neo Technology

8th Floor, Friars Bridge Court

41-45 Blackfriars Road

London SE1 8NZ