We have a network dataset with around 300k lines.

We have a car that has to do inspections on some elements along a route. It needs to visit 500-700 elements (most of them are close to each other).

The points create one simple route but are not ordered.

The way we solve the problem is to load the points as steps and to ask for best order route (usually the first point is really the start).

It creates a good route that visits every element.

The only problem is the time it takes. About 7 minutes to 700 points.

When the points is ordered it takes a few seconds.

Is there any way to speed the best route algorithm?

Thanks

That is a lot of combination to figure out. Have you tried a

lexicographic sort on the coordinates? Or is that what you mean by sorted? You really need to limit you search radius for the next possible candidate.