Thank you so much Michael, About point 1) yes that will improve the exution time,thanks 2) well... I noticed that it will increase the execution time.. if I write MATCH path = shortestPath((S)-[:CONNECTED*1..2]->(E)) I saw time increase, Am I doing a mistake? 3)yeah, my relation's type is CONNECTED
APOC: wonderfull plugin, but... wich algo can I use to increase my performance? Every algo work with a starting node and a final node.. I have only one of the two.. I tried apoc.path.expand, it works but quite the same as my previous query. SPATIAL: well that is an idea that I was evaluating.. first of all: if I reduce the number of nodes still the number of relations and the paths to calculate won't change will this be enought in order to reduce ex. time? Doesn't it depends from the number of relations that my query finds starting from the node specified? THANK YOU! Il giorno giovedì 7 luglio 2016 17:48:15 UTC+2, Michael Hunger ha scritto: > > Hi, > > 1. you should use a label + index or constraint to find your starting node > 2. I'd change it to use shortest path perhaps, at least rewrite it a bit > 3. do you have only one relationship-type ? > > // lookup via label + osm_id in index/constraint > > MATCH (S:Waypoint) WHERE S.osm_id="1683208894" > MATCH path = (S)-[rels *0..75]->(E) > WHERE REDUCE(cost=0.0, r IN rels | cost + toFloat(r.minutes)) < 15 > WITH E, REDUCE(cost=0.0, r IN rels | cost + toFloat(r.minutes)) as cost > RETURN cost, collect(distinct E) as isochrones > ORDER BY cost > > I'm not sure if the cost expression is pulled into the expansion, please > compare the query above with the one below > > explain MATCH (S:Waypoint) WHERE S.osm_id="1683208894" > MATCH path = (S)-[rels *0..75]->(E) > WITH E, REDUCE(cost=0.0, r IN rels | cost + toFloat(r.minutes)) as cost > WHERE cost < 15 > RETURN cost, collect(distinct E) as isochrones > ORDER BY cost > > > In reality you want to use an path expansion algorithm. If you use Neo4j > 3.0.x you can check out the graph algorithms of the APOC library (like > dikjstra, A*, allSimplePaths etc) > > > https://github.com/neo4j-contrib/neo4j-apoc-procedures#graph-algorithms-work-in-progress > > Or perhaps check out Neo4j Spatial which got an uplift for 3.0 as well, > you could use a boundary-circle for a first estimate, e.g. 30km and then do > the post-filtering with the actual cost > > http://github.com/neo4j-contrib/spatial > > Craig can say more to that. > > > On Thu, Jul 7, 2016 at 3:36 PM, BalzaelRisingDark <gallega...@gmail.com > <javascript:>> wrote: > >> Hi everyone, >> I'm new in Neo4j world, I'm trying to elaborate isochrones in very large >> DB created from osm datas and loaded into Neo4j. >> Isochrones should be intended as: "All nodes reachable from a starting >> node within a certain time", in my db the nodes are for example the >> crossroads of a city and the relations are the streets with the time needed >> in minutes to go from a crossroad to another. >> I wrote something like this in cypher >> >> MATCH (S) WHERE S.osm_id="1683208894" //to find the starting node >> MATCH path = (S)-[*0..75]->(E) //to find all nodes connected with a max >> of hops of 75 relations, if I do not put a limit it will calculate every >> possible path for every possible node in a 10GB db... not the best.. >> WITH E,MIN( REDUCE(cost=0.0, r IN RELATIONSHIPS(path) | cost + >> toFloat(r.minutes)) ) AS cost //to calculate cost >> WHERE cost<15 //if i want only the isochrones of 15 min >> RETURN cost,collect(E) as isochrones >> ORDER BY cost >> >> The problem is that execution's time degenerate after 70 relations and 70 >> is enough for isochrones of 12 minutes, I want to find at least 30 minutes >> isochrones. >> Is that possible in Cypher without waiting a year? >> Will be better if go for it by Java? Am I forced to write an apposite >> algorithm for this? >> >> THANKS! >> M.G. >> >> -- >> 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+un...@googlegroups.com <javascript:>. >> For more options, visit https://groups.google.com/d/optout. >> > > -- 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+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
