Analyst(s): Dr. Bob Sutor
Publication Date: September 17, 2024
IBM’s introduction of Qiskit Functions marks an attempt by the company to produce a catalog of cloud services for quantum computing that depends on the company’s own Qiskit Runtime. Will it be populated by a rich selection of quantum application, circuit, and control code that will attract and keep more developers on the platform?
What Is Covered in This Article:
- The Qiskit Open-Source Quantum Development Toolkit and IBM’s Qiskit Runtime
- How developers extend their own work via code written by others
- Qiskit Functions and the initial services available
The News: On September 16, IBM announced the availability of the first wave of Qiskit Functions, managed abstracted quantum services provided by the company and third-party partners for accelerating development workflows. The initial round of services includes functions for Chemistry and Optimization applications and several low-level functions for error suppression and mitigation. For additional details, see the IBM Quantum Research blog post “Announcing the Qiskit Functions Catalog.”
Quantum in Context: IBM’s New Qiskit Functions Accelerate Development
Analyst Take: With technology as new and quickly evolving as quantum computing, we cannot expect developers to write and rewrite the non-core code they need to support their applications. While open-source libraries in languages such as Python can help the devs, users require high-performance and tested runtime functionality created by experts. Qiskit Functions is a solid attempt by IBM to provide such a collection of runtime services. The initial offering is small but should increase healthily, especially after the IBM Quantum Developer Conference 2024 in November.
Qiskit
Qiskit originally stood for “Quantum Information Science Kit” and is a quantum software development kit (SDK). IBM put the initial source code on GitHub in 2017 under the Apache License 2.0. For most of its life, it was a set of libraries primarily in Python, though IBM has rewritten much of the underlying code in Rust since the 1.0 release earlier this year for faster performance and to handle larger circuits with more qubits.
Every metric I have ever seen has put Qiskit as the number one quantum SDK, with the most downloads and users. Its main competitors are Google Cirq, Xanadu Pennylane, and Microsoft Q#. IBM has built a global ecosystem of tens of thousands of developers through an online textbook, challenges, videos, and courses. There are at least a dozen books about the SDK, with translations into many languages. My own book Dancing with Python: Learn to Code with Python and Quantum Computing teaches classical and quantum coding and uses Qiskit examples.
The Qiskit software architecture allows quantum computing hardware vendors to build custom backends via Python libraries, allowing developers to use the SDK to write quantum code that runs on multiple platforms. The Python Package Index includes many of these libraries, including those for IonQ, Rigetti Computing, Amazon Braket, Alpine Quantum Technologies, and Quantinuum.
Writing Quantum Applications and the Qiskit Runtime
For many years, the sequence of coding and running quantum circuits on IBM hardware looked like this:
- Learn Python
- Install the Qiskit Python libraries
- Learn quantum computing and how to do it via Qiskit
- Write your Python code in an editor like Visual Studio Code; I have been coding for decades, and this is the best programmer’s editor I have ever used
- Within your code, create quantum circuits via Qiskit methods and functions
- Via Qiskit, choose a quantum simulator or computer and run your code
- Debug, going back to step 5, if necessary
- Do something with the results of the quantum computation.
Other than step 6, almost everything happened on your laptop or desktop computer. The services for executing the code were almost all about controlling the job and where it ran. Suppose your overall local application is repeatedly called a quantum computer on the cloud. In that case, each iteration had a significant overhead as each circuit moved across the Internet, waited in a queue, ran, and had the server return the result.
As with many kinds of computing, the code you write is not what runs directly on the hardware. Qiskit translates your circuit into an intermediate form, optimizes it to a degree you can specify, and then creates the sequence of instructions that runs on the hardware. The final form is similar to what we would classically call machine code, and you can program that in close-to-the-hardware classical assembly language if you wish. While the high-level Qiskit libraries are in Python and often wrap around Rust code, you can also program using OpenQASM (Open Quantum ASseMbly language). OpenQASM and compiled Qiskit give developers very fine control over quantum computing hardware and allow partial remediation of errors.
In 2022, IBM introduced the Qiskit Runtime. It allows you to push the execution of much more of your quantum application onto the cloud. Qiskit Runtime executes iterative code close to the quantum computer, removing much of the overhead. It also allows you to invoke quantum services already resident in the cloud. This simplifies your code and lets you use third-party features to enhance or improve what your code does or how well it executes.
While Qiskit itself is open source, Qiskit Runtime is not. Qiskit Runtime is an infrastructure service implemented to execute the primitives introduced in the Qiskit SDK efficiently on IBM Quantum hardware. Base Primitives are made available in the open-source Qiskit SDK and can be used to implement services for other hardware. Alpine Quantum Technology and Amazon Bracket each use these primitives to create their own services.
IBM encourages developers to make Qiskit Runtime the standard way they execute their quantum applications.
Qiskit Functions
Qiskit Functions uses the Qiskit Runtime model to deliver quantum functionality at runtime. IBM divides Qiskit Functions into two categories: application functions and circuit functions.
In this first release, the application functions are
- Hardware-efficient Quantum Approximate Optimization Algorithm (QAOA) for graph problems, by Q-CTRL, and
- Quantum simulation for chemistry, by QunaSys.
The circuit functions are
- Error suppression, by Q-CTRL,
- Error mitigation, by Algorithmiq,
- Error suppression and mitigation, by Qedma, and
- Error suppression, mitigation, and AI circuit optimization, by IBM.
This first release is weighted toward preventing errors or reducing their effects. This makes sense since error correction and fault tolerance are major research areas for quantum computing. You need not correct errors that don’t exist. I consider many of the circuit functions to be control functions, and I think they would have been better described that way by IBM.
The subsequent additions to Qiskit Functions should lean more toward applications than low-level circuit support. While IBM understands that its users are often physicists or quantum computational scientists, it must demonstrate that Qiskit Functions can accelerate the development of applications that users need. Perhaps the name for such developers is “quantum data scientists,” as IBM suggests, though I think “quantum application engineers” is a better label.
Repositories and catalogs must be curated. While the Qiskit Functions collection has six entries now, there will be many more if the program is successful, which I think it will. IBM or a contracted third party must ensure all entries are versioned, updated with Qiskit and IBM Quantum hardware, or removed.
For additional information, see the follow blog post and press releases:
- IBM Quantum Research blog post: “Announcing the Qiskit Functions Catalog”
- Algorithmiq press release: “Algorithmiq Launches High-Performing Error Mitigation Solution in IBM’s Qiskit Functions Catalog”
- Q-CTRL press release: “Q-CTRL Delivering New Services to IBM Quantum Network”
- Qedma press release: “Qedma’s Next-Generation Quantum Error-Mitigation Software Available as IBM Qiskit Function”
- QunaSys press release: “QunaSys’ QURI Chemistry Launched within IBM’s Qiskit Functions Catalog, Aims to Simplify Utility-Scale Quantum Computing for Materials Development”
What to Watch:
- Can IBM attract additional third parties to the Qiskit Functions platform?
- Will authors of Qiskit Functions feel they are adequately compensated for being part of the program?
- Will the Qiskit Functions get promptly updated to match the development of Qiskit itself?
- Will third-party developers feel locked into the IBM Quantum offering, given the increasing number of systems offered by other vendors?
Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.
Other insights from The Futurum Group:
Quantum in Context: A Qubit Primer
Quantum in Context: IBM Extends Qiskit 1.0 via AI Tools
Quantum in Context: Quantum Software Development Kit Qiskit Turns 1.0
Author Information
Dr. Bob Sutor is an expert in quantum technologies with 40+ years of experience. He is the accomplished author of the quantum computing book Dancing with Qubits, Second Edition. Bob is dedicated to evolving quantum to help solve society's critical computational problems.
