Opened 13 years ago
Closed 11 years ago
#382 closed defect (fixed)
GriddedGeoDataSet performance
| Reported by: | Ned Field | Owned by: | Kevin Milner |
|---|---|---|---|
| Priority: | major | Milestone: | OpenSHA 1.3 |
| Component: | commons | Version: | |
| Keywords: | Cc: |
Description
The following methods of GriddedGeoDataSet? are really slow:
contains(Location loc)
get(Location loc)
getLocation(int index)
due to the use of nodeList (a LocationList? that comes from nodeList = region.getNodeList(), where region is the GriddedRegion? passed in the constructor).
The reason is nodeList.indexOf(loc) is much slower than region.indexForLocation(loc).
For my problem (ERF_Calculator.getNucleationRatesInRegion((*)) it was at least two orders of magnitude, maybe more.
I don't think nodeList is even needed in this class, as methods from the region object can provide the info in the above methods, and getLocationLIst() can just return region.getNodeList().
I also don't think the HashMap? is needed (a double[] array for z-axis values should do), although I haven't thought through the use of the latitudeX boolean in this context.
I haven't made any changes in case revision to junit tests also need to be made.
This was done long ago, now uses fast methods from the GriddedRegion? object