Ticket #4262 (closed: fixed)

Opened 9 years ago

Last modified 5 years ago

SmoothNeighbours: incorrect errors?

Reported by: Janik Zikovsky Owned by: Janik Zikovsky
Priority: minor Milestone: Release 2.0
Component: Mantid Keywords:
Cc: owen.arnold@… Blocked By:
Blocking: Tester: Wenduo Zhou

Description

Hi Janik,

I've noticed that in the processing of errors in SmoothNeighbours, for MatrixWorkspaces, you square the error, then multiply by the weight, then take the sqrt of the error. Pascal is complaining that the weights aren't coming out right on SmoothNeighbours, but the only dubious thing that I see done is this multiplication by the error2. Why are we applying the weights to the error2?

Thanks, Owen.

Change History

comment:1 Changed 9 years ago by Janik Zikovsky

Refs #4262: Fixing error calculation for SmoothNeighbours

As per proper error propagation formulae, when multiplying by an exactly known value (the weight), the error squared is scaled by the weight squared. Test modified to reflect this: the output workspace has lower errors since each pixel has improved statistics.

Changeset: 8f69d6b3aea7692c518529024652965819be9910

comment:2 Changed 9 years ago by Janik Zikovsky

I'm not sure what Pascal sees incorrect in the weights, but the errors were indeed incorrect as I describe above.

comment:3 Changed 9 years ago by Janik Zikovsky

  • Status changed from new to accepted

comment:4 Changed 9 years ago by Owen Arnold

I tested this out and it seems to work. Code changes look sensible too.

comment:5 Changed 9 years ago by Janik Zikovsky

  • Status changed from accepted to verify
  • Resolution set to fixed

comment:6 Changed 9 years ago by Wenduo Zhou

  • Status changed from verify to verifying
  • Tester set to Wenduo Zhou

comment:7 Changed 9 years ago by Wenduo Zhou

  • Status changed from verifying to closed

comment:8 Changed 5 years ago by Stuart Campbell

This ticket has been transferred to github issue ​5109

Note: See TracTickets for help on using tickets.