pyphi.campaign

Distribute PyPhi computations across an HTCondor pool as batch campaigns.

A campaign is a self-contained directory of serialized task files that independent condor jobs execute via python -m pyphi.campaign run; results are collected from per-task output files. prepare() writes the directory, the user submits the generated submit file with condor_submit, and status() / collect() operate purely on the directory’s files — a task is done exactly when its output file exists and loads.

Functions

collect(directory[, partial, sia, ...])	Reassemble a campaign's outputs into its result.
prepare(substrates, *, states[, subsets, ...])	Materialize a sweep as a self-contained HTCondor campaign directory.
prepare_ces(substrates, *, states[, ...])	Materialize a scoped CES sweep as a campaign.
scope_report(directory)	Read the scope report(s) a CES campaign's collection wrote.
status(directory)	Classify every task from the output files and refresh remaining.txt.

Classes

CESShardTask(task_id, kind, substrate_label, ...)	One shard of a scoped cause-effect computation for one system.
CampaignStatus(directory, n_tasks, n_cells, ...)	A campaign's task ledger: which tasks are done, failed, or pending.
CampaignTask(task_id, kind, compute, ...)	One condor job's worth of work: the cells it owns and their context.
CampaignTaskOutput(task_id, pyphi_version, ...)	A task's per-cell outcomes, aligned 1:1 with the task's cells.
CellOutput(status, result, traceback[, aux])	One cell's outcome: ok (with the result), skipped, or error.
SIAShardTask(task_id, kind, substrate_label, ...)	One stride of the system-partition sweep for one system.
ScopeReport(mechanisms_computed, ...)	What a scoped campaign computed, excluded, and certifies.

Submodules

merge	Tie-preserving merges of shard outputs.
runner	Execute one campaign task file and write its output document.
scope	Declarative feasibility surfaces for scoped cause-effect analyses.
shards	Shard planning for scoped cause-effect campaigns.