IBM Qiskit
IBM Corp. today announced it’s expanding its open-source quantum software engineering toolkit Qiskit to cover the entire software development stack and better equip developers with the ability to build practical solutions for quantum processors for research and enterprise applications.
Launched in 2017, the Qiskit software development kit started as a way to assist developers to go quickly from code to quantum circuit design. The update pulls together all of the tools that IBM has built for quantum circuits including the 1.0 software development kit, an artificial intelligence circuit optimization service, a simplified runtime service, an AI-powered Qiskit Assistant and an open-source tool to run supercomputing workloads on quantum hardware.
Circuit design can be an extremely complex task because quantum circuits are not built the same way classical computers are, which operate with stable state “1” and “0.” Qubits within a quantum computer can exist in a possible state between one and zero or be “entangled” with another quantum bit. By taking advantage of these properties, it’s possible to design quantum circuits that can execute faster than a classical computer for certain complex equations.
However, mapping a potential problem into a solution that can be written out into a quantum circuit is not an easy task and that’s where Qiskit provides an advantage. IBM notes that since its debut, Qiskit has become the most popular quantum software kit across the industry with more than 600,000 users who have executed more than 3 billion quantum circuits to date.
“The global adoption of quantum computing — and the discovery of quantum advantage — will require a combination of leading quantum hardware alongside a robust and performant software stack to run workloads,” said Jay Gambetta, an IBM fellow and vice president at IBM Quantum. “These two pillars are fundamental to the algorithm discovery that has begun on utility-scale quantum hardware.”
With this update, Qiskit is now a “mature and comprehensive software stack for the enterprise, government organizations, research institutions and universities,” IBM said. The Qiskit SDK now allows developers to optimize circuits for quantum hardware more than 16 times faster than previous versions with a threefold memory reduction.
Alongside the SDK, users also have access to the Qiskit Transpiler Service, which is a code compiler that takes computer code and creates quantum circuits. The transpiler now has an AI-powered optimization system embedded in it that maps out the best possible design for circuits, including optimal placement for qubits, routing and physical placement.
IBM has combined this with a new Qiskit Assistant, powered by Watson X, that can quickly assist developers in generating the code they need to produce a quantum circuit. Its a Visual Studio extension currently in alpha mode. To use it, a developer can write out an English description of the problem they want to solve in code comments and the Assistant will produce code that the transpiler service can realize into a quantum circuit.
The company also launched Qiskit Serverless, an open-source runtime tool that allows developers to easily run quantum-centric supercomputing workloads in hybrid environments. This means that after going through the motions to create optimized quantum circuits they can execute workloads across quantum hardware and classical clusters. Combined with the inclusion of Qiskit Runtime Service, with simplified execution modes, users can tailor their workloads for performance on more than eight types of the most popular quantum hardware platforms.
The company said this is the next step in bringing “utility-scale quantum computing” into fruition. IBM first demonstrated utility-scale capabilities in its hardware in 2023 and pushed forward with the launch of its first modular quantum supercomputer, the Quantum System Two, powered by the company’s 133-qubit Heron processor.
The advent of utility scale capabilities, an era in which quantum processors exceed the processing speed and power of classical computing simulating quantum computers. That means that for the first time, quantum computers can be used to explore practical problems rather than only experimenting on more quantum processes.
“At Brookhaven, we have used Qiskit to execute circuits on IBM’s quantum hardware, which has resulted in almost 20 published papers to date, including exploring the frontiers of physics, dynamic systems, condensed matter systems and more,” said James Misewich, associate laboratory director for energy and photon sciences at Brookhaven National Laboratory. “Qiskit has also enabled our teams to develop extensions that push forward our exploration of bosonic and hybrid qubit-bosonic circuits, and how they could advance fundamental quantum algorithm development and error correction.”
Quotes
Quantum computing will one day unlock a new era of accelerated business performance. With IBM Quantum, we are making rapid progress in the hardware, software, and applications needed to enable this. — Arvind Krishna, Chairman and CEO of IBM
With Qiskit, we are not just building a quantum computer, we are building an entire quantum ecosystem that will benefit industries from finance to healthcare. — Bob Sutor, Vice President of IBM Q Strategy and Ecosystem
The development of Qiskit represents a significant step towards democratizing access to quantum computing. It allows businesses to experiment and innovate with quantum algorithms that can eventually solve real-world problems. — Dario Gil, Senior Vice President and Director of IBM Research
Qiskit has enabled our partners and clients to explore quantum computing in ways that were previously unimaginable. By making quantum programming more accessible, we are accelerating the discovery of quantum advantage. — IBM Quantum Blog
HamLib: A library of Hamiltonians for benchmarking quantum algorithms and hardware
We recently benchmarked Qiskit's default transpilation performance on Hamiltonians collected in Hamlib (https://arxiv.org/abs/2306.13126) and compared it to TKET's default transpilation (Qiskit 1.1, TKET 1.26, optimization level 2). Hamlib is a large collection of Hamiltonians that represent binary and discrete variable optimization problems as well as condensed matter physics models and electronic/vibrational structures of molecules. Before a Hamiltonian can be solved on a quantum computer, a quantum circuit transpiler, such as Qiskit, must transform the Hamiltonian into a concrete quantum circuit that is compatible with the computational primitives of a target quantum computer.
The results of this benchmark are clear - Qiskit outperforms TKET on 86.3% of the 863562 evaluated Hamlitonians and performs equally on 7.1% of the Hamiltonians. On average, Qiskit reduces the size of a generated quantum circuit by 18% compared to TKET while demonstrating an average runtime speedup of 4.4x. Moreover, Qiskit scales better, yielding larger circuit size reductions and runtime speedups on larger Hamiltonians. These results cement Qiskit as the leader in quantum circuit transpilation for utility-scale problems and beyond.
It is a hard challenge to develop a transpiler that efficiently finds high-quality solutions on a general quantum circuit defined by a user. The Qiskit Developers need to navigate nuanced tradeoffs between transpilation quality, runtime and user expectations for the trillions of quantum circuits submitted by IBM Quantum users.
I'm happy with the achieved results and look forward to the continuing leadership of Qiskit for quantum circuit transpilation and quantum computing software. In the future, the Qiskit will continue to push for excellence in transpilation performance and runtime by developing novel methods for quantum circuit transpilation and finding efficient realizations.
