# pgRouting: calculate many-to-one shortest paths

In a previous question on how to calculate many shortest paths, @dkastl pointed to the pgr_kdijkstra function, which allows to calculate each time the total cost of one source-to-many targets shortest paths, by giving an array of target nodes.

What I'd like to do is, to calculate the total cost of many sources-to-one target shortest paths. A solution could be to switch source and target, but then I'm actually calculating the total cost of a different chain of paths, given the fact I'm using a directed topology network.

@UnderDark already provided a solution for this problem, but, out of curiosity, my question is:

Is there a way of modifying the pgr_kdijkstra function such that you're able to give an 'array of sources' and 'one target' as input?

Something like (given the example of pgRouting manual):

``````SELECT seq, id1 AS source, id2 AS target, cost FROM pgr_kdijkstraCost(
'SELECT id, source, target, cost FROM edge_table WHERE cost >= 0',
array[4,12], 10, false, false);

seq | source | target | cost
-----+--------+--------+------
0 |      4 |     10 |    4
1 |     12 |     10 |    2
``````

Thanks!

## 1 Answer

If you swap all oneway streets, then you can still get the same routes as a many-to-one calculation. Much faster than doing many one-to-one routes. If you don't care about speed, but just want to do it as simple as possible, use the query from underdark which I guess is something about doing all-to-all and selecting the needed routes from that.

• Thanks for your response @Uffe. Swapping the one-way streets is a nice trick which I didn't think of; I'll consider that. The query given by Underdark is not really an all-to-all, at least if you create a table with the sources and targets defining the routes to be calculated on beforehand. – kadéem Jul 8 '14 at 11:09