Skip to content

Frequently Asked Questions

Common questions about idmtools.

Installation

Q: Which Python versions are supported?

A: Python 3.10+ (64-bit only).

Q: Do I need to install all platforms?

A: No! Use pip install idmtools[platform] to install only what you need. See Installation Guide.

Q: Do I need login to Docker Hub or Github Container Registry?

A: No, all docker images are public unless you want to push your docker image to Github Container Registry which is idmtools default docker registry.

Usage

Q: How do I choose between platforms?

A: See Platform Comparison for guidance.

Q: Can I run simulations locally without a cluster?

A: Yes! Use the Container platform. See Platform Overview.

Q: How do I debug failed simulations?

A: Check simulation logs, stderr.txt file, use docker container platform for local testing, and enable verbose logging.

Performance

Q: How many simulations can I run at once?

A: Depends on platform:

  • COMPS: Thousands or more
  • Slurm: Cluster-dependent
  • Container: Limited by local resources

Q: How do I speed up parameter sweeps?

A: Use batch execution, optimize simulation code, and choose the right platform. See Parameter Sweeps.

Errors

Q: "Module not found" error

A: Ensure idmtools is installed: pip install idmtools[full]. Or pip install complained model

Q: Platform connection failed

A: Check your configuration and network connectivity. See Configuration.

Q: Out of memory errors

A: Reduce batch size, use fewer workers, or increase system resources.

Common Workflow Questions

Q: How do I check if my experiment is still running?

Use experiment.status after calling experiment.run(), or check the platform's web interface (e.g., COMPS portal). Or use idmtools cli command for Container and Slurm platforms.

Q: Why is my config.json empty?

Make sure you pass a JSON string to Asset(content=..., filename=config.json), not a Python dict. Use json.dumps(config) to convert.

Q: How do I add files my model needs?

Use experiment.assets for files shared across all simulations, and simulation.assets for per-simulation files.

Q: Can I run experiments locally for testing?

Yes, use Platform("Container") to run experiments locally using Docker containers.

Q: How do I rerun a failed experiment?

You can retrieve an existing experiment by ID and resubmit:

Python
from idmtools.core.platform_factory import Platform

platform = Platform("COMPS")
experiment = platform.get_item(item_id="<experiment_id>", item_type="experiment")
experiment.run(wait_until_done=True, platform)

Q: How do I pass parameters to my model?

Use JSONConfiguredPythonTask which automatically writes a config.json file that your model reads. Alternatively, use CommandLine to pass arguments directly. See Parameter Sweeps.

Getting Help

Can't find your answer?