Log of known changes in uboonecode and effect on validation plots¶
The table below contains a log of known changes in uboonecode, with notes about how we expect them to affect the validation plots. To see what the "comparison to base MC" validation plots should look like for the current version of develop compared to MCC 8, see Example plots - Comparison to base MC.
Note that for older releases, CI validation becomes sporadic since this was triggered manually several months later. Results from these select uboonecode releases can be found in docdb 15357.
|Date||ubcode release notes||LarSoft release notes||Comments|
|2018-08-13||v06_84_00||v06_84_00||Fix to Gaushit thresholds bug introduced in v06_82_00, so plots have reverted back to being correct. Also note that in this release the calorimetry plots appear to be incorrect. This seems to be due to some issue at the reco2 stage resulting in no pandoraNu data products being produced (or being produced but having a size of 0). Work is being undertaken to understand this.|
|2018-07-10||v06_83_00||v06_83_00||Addition of the number of trajectory points per track variable to CI validation pass. Last weeks didn't have this variable so the histogram should be empty.|
|2018-06-28||v06_82_01||v06_82_00||First split release but fails to build due to missing dependencies (ubsim)|
|2018-06-28||v06_82_00||v06_82_00||There's a change here in the number of reconstructed tracks and showers. Suspect commit affects Gaushit but currently under investigation.|
|2018-04-13||v06_74_00||v06_74_00||Shifters noted a change in the calorimetry plots. This is due to a new database tag (v2r0), which changes the status of a number of channels.|
|2018-02-16||v06_68_00||v06_68_00||Switched to generating without space charge for BNB MC, as is done with production. Expect some weirdness in the plots due to this. The reduction in track completeness is accompanied by a significant increase in very low completeness tracks (between completeness of 0% and 5%).|
|2018-02-09||v06_67_01||v06_67_01|| I see very little difference between these releases. There is some change in the calorimetry variables, which I believe probably comes from a commit by Tracy: 2018-02-06 Usher, Tracy L : Revert handling of long pulse trains to previous method of breaking into n pulses of at least m ticks width
(controlled by fhicl).
|2018-02-09||v06_66_00||v06_66_00|| First entry in log. Known differences between production (v06_26_01_11) and develop (v06_66_00) can be found here, but mainly come down to:
* Database tags update in uboonecode v06_29_00
* Pandora version update to v03_07_00 in uboonecode v06_60_00
|2017-12-15||v06_60_00||v06_60_00||There is a small change in the number of reconstructed showers here, however no change in the number of reconstructed tracks. There is a small reduction in the track start and end resolution and a significant change in the track completeness. This is thought to come from the introduction of pandora v03_07_00|
|2017-09-01||v06_48_00||v06_48_00||The differences in the hit distributions are not currently understood. There were calwireroi updates with this release which seem like a likely cause but it seems there is no concrete evidence of this.|
|2017-08-10||v06_46_00||v06_46_00||ROI finder changes induce hit multiplicity changes, and other minor hit-related changes.|
|2017-07-14||v06_43_00||v06_43_00||Bug fix in determining correct response for use on V plane un U and Y shorted region results in a reduced number of reconstructed tracks.|
|2017-06-22||v06_41_00||v06_41_00||CalWireROI updates affect hit distributions on the V plane, which is thought to be expected.|
|2017-06-09||v06_39_00||v06_39_00||Minor reconstruction differences. Introduction of the track direction bug (later fixed).|
|2017-05-11||v06_35_00||v06_35_00||CORSIKA timing bug is fixed, small difference in calorimetry plots due to a bug fix in the calibrations (XvY bug).|
|2017-04-29||v06_34_00||v06_34_00||Updates to hit distributions, assumed to be due to gaushit changes, but not confirmed.|
|2017-03-25||v06_28_01||v06_28_01||database tags update and hit thresholds are updated cause significant change in the hit plots due to a bug in the treatment of the shorted region. This was fixed in a subsequent version.|