Project

General

Profile

Task #20303

Switch protodune to tool-based dataprep

Added by David Adams about 1 year ago. Updated 12 months ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Start date:
07/10/2018
Due date:
% Done:

0%

Estimated time:
Duration:

Description

As shown in my talk at the last collaboration meeting, a simple configuration of the tool-based dataprep gives the same results as the default service-based configuration for protoDUNE simulation.

We should make the tool-based dataprep the default.

History

#1 Updated by David Adams about 1 year ago

  • Status changed from New to Feedback

I modified dune/DataPrep/Tool/fcl/dataprep_dune.fcl to rename the block adcprep_sim to adcprep_with_services_sim. This is the default block for configuring service-based dataprep. I updated dune/Utilities/services_dune.fcl to use the new name and confirmed protodune dataprep still runs.

I then added the blocks adcprep_with_tools_sim and adcprep_with_tools_data to the former to configure tool-based dataprep and updated protodune_reco_services.RawDigitPrepService in the latter to point to the first block. I confirmed that protoDUNE simulation reco now uses the tool-based data and, for one event by eye, gives the same results as the old reco.

The change is committed to dunetpc.

Tingjun, could you check that this has not destroyed downstream protodune reco?

#2 Updated by Tingjun Yang about 1 year ago

Hi David,

According to the CI test:
http://dbweb5.fnal.gov:8080/LarCI/app/ns:dune/storage/docs/2018/07/10/stdout_bVQNQWq.log
there do seem to be some changes in data product sizes:

764: Check for differences in the size of data products
765: difference(s)
766: 
767: 38c38
768: < Reco | caldata |  | art::Assns<raw::RawDigit,recob::Wire,void> | 715
769: ---
770: > Reco | caldata |  | art::Assns<raw::RawDigit,recob::Wire,void> | 896
771: 57c57
772: < Reco | pmtrack |  | art::Assns<recob::Track,recob::Hit,recob::TrackHitMeta> | 1391
773: ---
774: > Reco | pmtrack |  | art::Assns<recob::Track,recob::Hit,recob::TrackHitMeta> | 1409
775: 59c59
776: < Reco | pmtrack |  | art::Assns<recob::Track,recob::Hit,void> | 1391
777: ---
778: > Reco | pmtrack |  | art::Assns<recob::Track,recob::Hit,void> | 1409
779: 79,80c79,80
780: < Reco | pmtrack |  | art::Assns<recob::Track,recob::SpacePoint,void> | 1391
781: < Reco | caldata |  | std::vector<recob::Wire> | 715
782: ---
783: > Reco | pmtrack |  | art::Assns<recob::Track,recob::SpacePoint,void> | 1408
784: > Reco | caldata |  | std::vector<recob::Wire> | 896
785: 88c88
786: < Reco | pmtrack |  | art::Assns<recob::SpacePoint,recob::Hit,void> | 1391
787: ---
788: > Reco | pmtrack |  | art::Assns<recob::SpacePoint,recob::Hit,void> | 1409
789: 98c98
790: < Reco | pmtrack |  | std::vector<recob::SpacePoint> | 1391
791: ---
792: > Reco | pmtrack |  | std::vector<recob::SpacePoint> | 1408

Are they expected?

Tingjun

#3 Updated by David Adams about 1 year ago

Tingjun:

We do expect some difference as the new reco uses a different algorithm to find ROIs. This is discussed in the collaboration meeting talk. The new reco uses a simple threshold algorithm which (as configured) has a higher signal efficiency but lets through more noise. Do you have any downstream metrics such as track-finding efficiency and purity?

Thanks.

da

#4 Updated by Christoph Alt about 1 year ago

Hi,

Is there any update concerning the changes in the data product sizes?

Christoph

#5 Updated by David Adams 12 months ago

Tingjun has accepted the change for protodune.

#6 Updated by David Adams 12 months ago

  • Status changed from Feedback to Closed


Also available in: Atom PDF