Frequently asked questions
Common questions about Fire Opal
Does Fire Opal offer a free tier?
Yes! You can sign up for Fire Opal for free and choose to upgrade to our paid tier at a later time. Fire Opal's free tier allows you to make up to 10 calls to each Fire Opal function per day. To upgrade to the paid tier, please contact us.
How can I retrieve Fire Opal job results?
Each call to a Fire Opal is assigned a unique action ID. You can view and retrieve the results from previously run jobs by using the activity_monitor to identify the action ID, and then using get_result to retrieve the results.
This is the recommended method, since retrieving results directly from the hardware provider will not include Fire Opal's post-processing, thus the performance improvement will be limited compared to results obtained from the direct execution response.
Can I execute Fire Opal calls asynchronously so that I can use my kernel to run other functions?
Yes! Fire Opal's execute, iterate, and solve_qaoa return a FireOpalJob object which can be used to check the status and results. FireOpalJob's status method will not block your notebook while the result method will block until the job is done executing.
What devices do you support?
Currently, you can use Fire Opal with all cloud-accessible IBM Quantum Platform devices available through your usage plan. Some dedicated hardware devices might require white-glove onboarding.
Support for other backend providers and cloud platforms is on the way! Learn more about supported hardware providers.
Get in touch via the email above to learn about our device roadmap or to request support for a specific device or provider.
How many concurrent jobs can I run using Fire Opal?
Fire Opal allows you to run up to 20 concurrent jobs at a time. If additional jobs are submitted while 20 jobs are already running, you will receive an error message. Once jobs complete, further jobs may be submitted up to the concurrency limit of 20. We recommend batching circuits together in one submission in order to enable running as many circuits as possible given the concurrent job limit.
How large can a batch of circuits be?
Fire Opal has two restrictions on batches of circuits:
- 300 circuits or fewer per batch
- 50 MB or less of data per batch
When running very large circuits, it is possible that fewer than 300 circuits may begin to exceed the 50 MB data maximum. If this threshold is exceeded, an error message will be returned and the Fire Opal job will not be queued or end up running. If this happens, please try a smaller batch size.
Can I use simulators with Fire Opal?
Simulators are not supported by Fire Opal. The benefits of Fire Opal's advanced error suppression and mitigation techniques are tailored for real quantum computers. Our goal is to enable users to perform research and development on real devices as this is the only way to get a true indication of algorithmic performance. Even noisy simulators have biases that can skew results to make them less informative.
Why am I getting a localhost refused to connect. error when Fire Opal tries to authenticate?
This error may occur if you are using a web-based development environment, such as Google Colab or JupyterLab, which are incompatible with Fire Opal's browser-based authentication. To avoid this error, authenticate using an API key.
Why am I getting an error message that I have reached the maximum number of API limits per day?
The Free Plan has a limit of 10 calls per user per day. To increase your limit, please contact Q-CTRL.
Why do I see an additional job tagged with "fire-opal-mitigation"?
Fire Opal runs a lightweight form of error mitigation called "measurement error mitigation". This involves collecting data from a few test circuits, and it is similar to QiskitRuntime's default resilience_level=1. The entire process takes about 10 seconds or less and will be run at most once every 6 hours. You are more likely to see this job if you are using a private device.
Where should I reach out if I have additional questions?
If your question is not addressed in this FAQ, you can contact us.
Additionally, you may join the Q-CTRL Discord server to get in contact directly with the team regarding questions or product feedback!
