Quickstart¶
This page shows a minimal end-to-end run that produces residual plots and diagnostic tables. It assumes you already have a tempo2-compatible data set and a working tempo2 installation or container image.
Config Layout Quick Reference¶
Use this map when choosing where to put files:
configs/runs/: runnable mode profiles passed to--config.configs/workflows/: multi-step orchestration files passed toworkflow --file.configs/workflows/steps/: reusable workflow step configs.configs/catalogs/: reusable mapping/lookup data.configs/rules/: declarative behavior policies (relabel/overlap/PQC profiles).configs/state/lockfiles/: generated ingest lock snapshots/validation.configs/settings/: legacy location; do not add new files.
For full layout guidance, see Configuration Layout.
Minimal configuration¶
Create a pipeline.toml file:
home_dir = "/data/epta"
singularity_image = "/images/tempo2.sif"
dataset_name = "EPTA"
branches = ["main", "EPTA"]
reference_branch = "main"
results_dir = "results"
jobs = 4
Run the pipeline¶
pleb --config pipeline.toml
CLI overrides¶
Any key from the TOML config can be supplied directly as a CLI flag and will override the file value. These are equivalent:
pleb --config pipeline.toml --jobs 8 --run_fix_dataset
pleb --config pipeline.toml --set jobs=8 --set run_fix_dataset=true
Boolean keys accept --key (true) and --no-key (false).
The pipeline prints a run tag pointing to the output directory. Inside that directory you will find:
Residual plots grouped by backend and pulsar.
Change reports comparing the configured branches.
Summary tables and QC artifacts (if enabled).
Optional QC stage¶
If you have pqc and libstempo installed, you can enable outlier and
transient detection:
pleb --config pipeline.toml --qc
The QC stage is designed to highlight residual outliers, transient events, and structured behavior in residuals versus time, frequency, and orbital phase. [Coles2011] [Keith2013]
Declarative relabel rules¶
For post-QC backend splitting (for example, dividing one -sys into epoch
sub-groups), point FixDataset at a TOML rules file:
pleb --config pipeline.toml \
--set run_fix_dataset=true \
--set fix_relabel_rules_path="configs/rules/relabel/relabel_rules.example.toml"
Declarative overlap rules¶
For duplicate-preference policies (for example, prefer dual-channel WSRT P2 rows over paired single-channel rows), use overlap rules:
pleb --config pipeline.toml \
--set run_fix_dataset=true \
--set fix_overlap_rules_path="configs/rules/overlap/overlap_rules.example.toml"
Per-backend PQC profiles¶
To override PQC thresholds for selected backends while keeping global defaults as fallback:
pleb --config pipeline.toml \
--set run_pqc=true \
--set pqc_backend_col="sys" \
--set pqc_backend_profiles_path="configs/rules/pqc/backend_profiles.example.toml"
TOML-configurable outlier strategies¶
You can now choose the outlier decision strategy directly in TOML by listing which QC columns should be treated as outlier evidence.
pleb --config pipeline.toml \
--set run_fix_dataset=true \
--set fix_qc_remove_outliers=true \
--set fix_qc_outlier_cols='["bad_point","robust_outlier","robust_global_outlier","bad_mad"]' \
--set qc_report=true \
--set qc_report_compact_pdf=true \
--set qc_report_compact_outlier_cols='["bad_point","robust_outlier","robust_global_outlier","bad_mad"]'
Compact PDF report¶
To generate a compact final PDF summary of likely bad TOAs (with plots):
pleb --config pipeline.toml \
--set qc_report=true \
--set qc_report_compact_pdf=true \
--set qc_report_compact_pdf_name="qc_compact_report.pdf"
The compact report also writes per-backend action CSVs under
<qc_report_dir>/action_lists/ with exact flagged TOAs and reason columns.
Exact-overlap keep/drop catalog¶
The exact duplicate overlap map is loaded from
configs/catalogs/system_tables/overlapped_timfiles.toml by default. To override:
pleb --config pipeline.toml \
--set run_fix_dataset=true \
--set fix_overlap_exact_catalog_path="configs/catalogs/system_tables/overlapped_timfiles.toml"
Next steps¶
For details on the CLI, see Command-Line Interface.
For practical tuning of TOML settings and detection strategies, see Configuration Reference.
For statistical context (chi-square, red noise, DM variations), see Concepts.