Difference between revisions of "Staff Priorities - Masha"
From SCECpedia
Jump to navigationJump to search| Line 21: | Line 21: | ||
** Max suggested an optimization to N test (to avoid look up of forecast bin for each observed event): Trac ticket #353: Optimize RELM N-test by avoiding event to bin look up when masking bit is set for the whole forecast | ** Max suggested an optimization to N test (to avoid look up of forecast bin for each observed event): Trac ticket #353: Optimize RELM N-test by avoiding event to bin look up when masking bit is set for the whole forecast | ||
** Trac ticket #352: Remove RELM R evaluation test from all forecasts groups at SCEC testing center | ** Trac ticket #352: Remove RELM R evaluation test from all forecasts groups at SCEC testing center | ||
| + | ** Per Max's request, run N-test for 100 events with HDF5 forecast for 0.1 degree Global region, disable forecast's masking bit | ||
** Work on migration to upgraded OS and Python 3 | ** Work on migration to upgraded OS and Python 3 | ||
Revision as of 00:51, 11 June 2015
Current Activities
Yesterday
- CSEP:
- Peter Bird's SHIFT_GSRM2f model for 0.1 degree Global region:
- ifort 13 was picking up 64-bit libraries at compile time, but was configured to run in 32-bit
- Re-built ifort dependent models, ran acceptance tests OK
- Re-generated new forecasts for new group with SHIFT_GSRM2f model
- Evaluated new forecasts with N, T and W tests, getting error at clean up for HDF5 format of the forecast file
- Added acceptance test for SHIFT_GSRM2f model, runs OK on csep-cert
- Alvaro verified the differences, OK to stitch to ifort 13
- Work on migration to upgraded OS and Python 3: building CSEP code base on new image with John's help
- Peter Bird's SHIFT_GSRM2f model for 0.1 degree Global region:
Today
- TD: reprocess due to maximum number of users for CMT ftp
- CSEP:
- Status call at noon
- csep-cert: new SHIFT_GSRM2f model for 0.1 degree Global region
- Investigate an error at clean up for HDF5 format of the forecast file for new forecast group
- Update Alvaro's models with new reference data as generated by ifort 13
- Max suggested an optimization to N test (to avoid look up of forecast bin for each observed event): Trac ticket #353: Optimize RELM N-test by avoiding event to bin look up when masking bit is set for the whole forecast
- Trac ticket #352: Remove RELM R evaluation test from all forecasts groups at SCEC testing center
- Per Max's request, run N-test for 100 events with HDF5 forecast for 0.1 degree Global region, disable forecast's masking bit
- Work on migration to upgraded OS and Python 3
Blocked: None
Followups: None
TODO
- Trac ticket #346: Evaluate Oceanic Transform Faults forecasts since 2012/01/01
