As a learner, I would like to run code on IBM Quantum hardware without leaving the textbook, so that I can understand the impacts of noise in my experiments.

[lerongil]

Background

Learners in the textbook should be able to run code on hardware within the textbook to see that it exists and understand the concept of noise. The final implementation should seem like a magical experience where the user needs to merely learn and write code to run on hardware without any other sign-up flows or inputing their own API token. In an effort to remain agile, this user story has been broken up into 2 phases: Exploratory and Execution. We should take as little time as possible in both phases to release this user story so please comment on suggestions for scoping down or clarifying.

Goal

Learners should be able to run qiskit code on IBM Quantum hardware directly from the scratchpad and code cells within the qiskit textbook.

Exploratory Phase:

• [x] Research the feasibility of the below acceptance criteria

Acceptance Criteria:

• [x] User can copy their IBM Quantum API token to Qiskit account
• [x] User is able to run code on hardware when relevant without any account code inputted
• [x] User is able to interact with hardware in a chapter with error mitigation material
• [x] IBMer is able to measure textbook execution contribution metrics

Design file:

• Figma design

Tasks to be done to accomplish the criterias:

• 1197

• 1198

• 1199

• 1200

[pandasa123]

Some guardrails / thoughts:

• Adding Users: The user should have an existing IBM Quantum account to maintain a consistent compliance policy? We could do a specific textbook hub but we might be limited by a propagation issue?
• Systems: I think this should be restricted to Open systems if we're doing a consistent experience across all users OR systems that a specific user has access to across all the hubs?

Current Experience in Textbook:

• Most users use the simulators in the Textbook. This is by design
• If users wanted to run on hardware, they would have to create an IBM Quantum (or any other hardware provider) account, add provider authentication code (IBM.enable_account(<API-TOKEN>)), and finally proceed to pick their backend

Open Questions:

• Which parts of the experience should we simplify?
• How do we balance magic and explainability?

[Tansito]

From previous conversations about the interaction between our users and IQX users:

https://github.com/Qiskit/qiskit.org/discussions/2268#discussioncomment-1577608

This time if we want to increase the priority on this issue maybe we can contact with @delapuente directly and maybe Jorge or Ismael Rozas (Salva can tell us who can be good to join the conversation).

[pandasa123]

Summary

We're going to use a server side cookie instead of API token, for security purpose

From a user POV: they have to create accounts on Qiskit and IQX to start. If they have never added their API token on Qiskit, we can ask them to add it and store that server side

• It's on our team to implement, but @francabrera can provide support
• There's a few existing examples with the Swagger API etc
• Since we're going to continue to use the qiskit.org domain rather than quantum-computing.ibm.com

Next Steps

• @francabrera will pull up an example that utilises Segment as well 🎉
• @Tansito, @eddybrando, and @francabrera will sync about the implementation details 🚀

[JRussellHuffman]

Design file here: https://www.figma.com/file/RWKSDXw64Pm6WG4ObDAl6H/2475-Run-on-hardware?node-id=37%3A721

Also adding that link to the description

[JRussellHuffman]

Removing the design label because the designs are mostly done.

[lerongil]

One slight update to the acceptance criteria: we want to make sure that this is something that can be measured. The last bullet in the acceptance criteria ensure we have attribution to this feature development.

[Tansito]

One slight update to the acceptance criteria: we want to make sure that this is something that can be measured. The last bullet in the acceptance criteria ensure we have attribution to this feature development.

Do you mean to have an score (like I don't know, from 0 to 10) and not only a check that says if the user passes or not?

[JRussellHuffman]

I think it's purely for analytics. We want to know if our users are actually using this feature.

[pandasa123]

@frankharkins, I've created an issue to add a hardware example