Skip to content

/neuroflow:brain-run

Run a computational brain model simulation — configure run parameters, launch, and collect outputs.

/brain-run handles everything after the model is built: configuring a run, launching the simulation, inspecting outputs, and saving results.

When to use it

  • You want to run a simulation of an assembled model
  • You need to configure simulation duration, time step, and recording targets
  • You want to submit a run to HPC or a cluster
  • You need to inspect and summarise simulation outputs

What it does

Claude asks:

  1. Which model to run — path to model code, or picks it up from .neuroflow/brain-build/
  2. What simulation duration and time step
  3. What inputs / stimuli to apply during this run
  4. What outputs to record (spike trains, membrane voltage, LFP, population rate, BOLD signal)
  5. Run locally or submit to HPC/cluster

Steps

  1. Write run-config.md — model path, duration, dt, inputs, recording targets, output directory
  2. Confirm or generate a run script (run_sim.py or equivalent) that loads the model and applies the run config
  3. Run the simulation (or prepare the HPC submission script)
  4. After the run: load and inspect outputs — check for obvious errors (no spikes, unbounded activity, NaN values), produce a brief summary plot or statistics
  5. Write run-summary.md — duration, time step, key output statistics (mean firing rate, dominant frequency), path to output files

Configs and summaries go to .neuroflow/brain-run/. Simulation output files (spike data, voltage traces, figures) go to models/results/.

Files read and written

Direction Files
Reads .neuroflow/project_config.md, .neuroflow/flow.md, .neuroflow/brain-run/flow.md, .neuroflow/brain-build/flow.md
Writes .neuroflow/brain-run/, .neuroflow/brain-run/flow.md, .neuroflow/sessions/YYYY-MM-DD.md, models/results/ (simulation outputs)