On 12/23/2014 3:55 AM, Hans Skov-Petersen wrote:
Hi List,
Being rather new at PgRouting (but not to network analysis in general),
I am looking for means by which I efficiently can construct all pair,
shortest path OD matrices for quite large networks (i.e. a graph of the
road network of Denmark, 2.107.981 edges, 860.380 nodes).
I'm not familiar with the term OD, but I assume you mean Distance matrices of some kind.
I am looking at the All Pairs Shortest Path, Johnson’s or Floyd-Warshall
Algorithms.
If you are generating a distance matrix for 860,380 nodes that will be 740,253,744,400 cells which is really HUGE amount of data!
What are you planning on doing with all this data?
Questions:
1)Can anyone guide me to examples of the SQL syntax (or other code)
involved in applying the algorithms
As Manikanta mentioned look at the tutorial, you can also look at the test files for trivial examples:
https://github.com/pgRouting/pgrouting/blob/master/src/apsp_johnson/test/apsp_johnson-any-00.test
https://github.com/pgRouting/pgrouting/blob/master/src/apsp_warshall/test/apsp_warshall-any-00.test
2)Is there a way that the search (and output) can be simplified by
limiting the search to only take OD within a given max-search-radius
into account?
select seq, id1, id2, round(cost::numeric, 2) as cost from pgr_apspWarshall('select id, source, target, cost from apspw', false, false);
The first argument to the function is a sql query for the edges that you want to be used to build your graph from. So you can restriction what edges you pass it by change that query to something like:
'select id, source, target, cost from apspw where st_dwithin(geom, st_setsrid(st_makepoint(<longitude>,<latitude>),4326), <radius>)'
Where <radius> is in degrees and 1 degrees is approximately equal to 127788 meters.
-Steve
Cheers and Thanks
Hans
_______________________________________________
Pgrouting-users mailing list
Pgrouting-users@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/pgrouting-users
