Skip to content

Statsout#275

Open
madDipanjan wants to merge 66 commits into
masterfrom
statsout
Open

Statsout#275
madDipanjan wants to merge 66 commits into
masterfrom
statsout

Conversation

@madDipanjan

Copy link
Copy Markdown
Contributor

Modification to statistics output routines.

Moved the contents of modstatsdump.f90 and writestat_nc.f90 to out_stats.f90, out_instants.f90 and out_write_nc.f90 in an organized way. Cleaned up the unnecessary large memory allocation.

- out_stats.f90: rename all stats NetCDF output files by dropping the
  '_out' suffix (e.g. stats_t_out -> stats_t, stats_xyt_out -> stats_xyt,
  etc.) and update the corresponding Fortran character index positions.
- gather_outputs.sh: update globs and string comparisons to match the new
  filenames without '_out'; use base-name extraction (%%.*) in the y-merge
  loop to correctly distinguish stats_t from stats_tree; add a dot-guard
  in the x-merge loop to skip raw per-process files accidentally matched
  by the broader glob; remove the now-unneeded end-of-script rename block
  for stats_xy/stats_xyt.
@mvreeuwijk mvreeuwijk added this to the 3.0 milestone Apr 3, 2026
madDipanjan and others added 27 commits April 8, 2026 00:42
* added parallel output, unit test_io, nkslice

* added ni/jslice, but every processor writes output

* comment old ij slicedump

* Add files via upload

* Add files via upload

* Add files via upload

* Add files via upload

* Add files via upload

* Add files via upload

* Update Synthetic Inflow Guide with new content

* Add files via upload

* Comment out OMP_NUM_THREADS export line

Comment out the export of OMP_NUM_THREADS for clarity.

* Add files via upload

* Add files via upload

* Add files via upload

* Add files via upload

* Add files via upload

* Add files via upload

* Delete tools/job_gather_out1d

* Delete tools/gather_out1d.py

* Delete src directory

* Add files via upload

* Refactor data gathering scripts

- Fix output filename format in stats module (stats_t_out -> stats_t)
- Improve gather_outputs.sh: merges 1D stats_t outputs along x
- Reorganize gathering scripts: rename gather_outputs_1D.sh to gather_outputs_v0.sh
- Kept gather_outputs.sh back up: copied to gather_outputs_v1.sh

* Bug fix. Properly checks if the ibm reconstruction point ids are within the bound, if not skipped.

* Add files via upload

* Variable naming modification.

* Refactor upsvpt and vpsvpt computation; and clean up.

* Refactor gather outputs script.

* Clean up and code refactoring.

* Some more code refactoring and clean up.

* Added upthlp and vpthlp to output in tavg stats.

* Clean up.

---------

Co-authored-by: jh6016 <56837101+jh6016@users.noreply.github.com>
tfielddump was removed from the OUTPUT namelist group in this branch
(renamed to tinstantdump). Update all integration test namoptions files
to use the new variable name so the solver does not abort on an unknown
namelist variable.
…put from out_stats.f90

- NO_TREE_FIELDS: ('ut','vt','wt') -> ('u','v','w') matching new NC var names
- PREFIX: 'tdump' -> 'stats_t', 'treedump' -> 'stats_tree'
- _parse_tile_indices: 4-part -> 3-part filename (no py index; myidy==0 writes
  full-y assembled file per x-rank via writeoffset ring communication)
- _load_global_fields: x-only assembly; each file covers full jtot y extent
- _run_case stale cleanup: updated glob patterns to match new filenames
…ts dump fires

out_stats.f90 initialises tsamplep=0 and tstatsdumpp=0, so at least two time
steps are needed before either accumulator reaches tstatsdump=0.005.  With
dtmax=0.01 and runtime=0.01 only one step ran and files had 0 records.
Setting runtime=0.02 gives two steps; on the second step both accumulators
reach 0.01>=0.005 and the stats_t/stats_tree files get one record each.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants