This week we highlight a paper jointly written by IBM Research in New York and Q-CTRL in Sydney that sets out to standardize quantum software user-interfaces with back-end data structures for quantum programming.
With commercial-scale quantum computing still in its infancy, it can seem a bit like the Wild West out there in terms of the language and protocols used to describe and develop functionality for this emerging technology.
Research teams, hardware-platform experimentalists, user-interface developers and technology companies have at times produced different idioms, like isolated islands of expertise inventing their own languages.
But the age of isolation in quantum technology is finally ending. Soon we may all say "Shibboleth."
Q-CTRL Contributes to the Qiskit Programming Handbook
Q-CTRL is pleased to have contributed to an IBM Research-led paper that seeks to standardize the application programming interface for quantum computing.
CEO and founder of Q-CTRL, Professor Michael Biercuk, said: "This document is one of the first handbooks for programmers seeking to give visibility into the machine language controlling the actual quantum hardware. Given Q-CTRL's role building drop-in replacements for standard quantum logic operations at the physical level, we were thrilled to contribute to this guide."
The paper, whose lead author is IBM's David McKay, outlines the need for standardized protocols and programming frameworks as user access to quantum technology becomes more common. It has developed these standards using IBM's open-source Qiskit, which is one the most widely used quantum programming framework given the success of the open-access IBM Q Experience.
The paper notes that "there is a pressing need for standardized APIs so that algorithm designers, circuit designers, and physicists can be provided a common reference frame for designing, executing, and optimizing experiments".
Published on the Physics arXiv, the paper is jointly authored by a stellar list of some of IBM's top quantum researchers, including Jerry Chow, Jay Gambetta and James Wooton. Joining them is Q-CTRL Founder Professor Biercuk, Q-CTRL's Lead Quantum Engineer Michael Hush, and Junior Quantum Control Engineer Jiayan Chen.
In April Q-CTRL became one of just eight startups worldwide to become a preferred collaborator for the IBM Q Network. This has allowed Q-CTRL to help define and communicate key techniques in programming something quite new in these systems: hardware-level controls.
Dr Hush explained one of the central advances of this paper; "Everyone is making moves to create open-source languages to program a quantum computer. What's different here is IBM is making a positive move to give users access at a lower level."
"Q-CTRL aims to be the trusted source for quantum control solutions," he added. "This paper shows we are a central collaborator with IBM, leading the field to define high-impact control solutions for quantum computing hardware."
happy to get OpenPulse and our future quantum API specifications out to the community https://t.co/4Z1LuPdVzu. Let us know what you think.— Qiskit (@qiskit) September 13, 2018
Opening access to the back-end specifications will allow programmers to go beyond just choosing gates in quantum algorithms - it sets the stage to give control over qubit drive operations. For superconducting qubits like those used in the IBM Q Experience, this means control over the microwave fields that physically manipulate the quantum coherent devices that enact the quantum computation.
Quantum algorithms are made up of "logic gates." Applying a gate defines an overall operation to be performed on the qubit, like a quantum bit flip. What the OpenPulse standard allows is precise definition of how that bit flip is physically enacted - how the qubit state travels from A to B across what is known as the Bloch sphere. (The Qiskit logo uses a stylized Block sphere).
Dr Hush said: "The OpenPulse standard that we contributed to allows the user to define the shape of the microwave pulse over time."
Changing the shape of the pulses allows the qubit to take a different path on the Bloch sphere - changing that path can build-in robustness against errors.
Vitally, this simple extension in functionality afforded by the OpenPulse framework now allows users to directly import Q-CTRL's solutions as drop-in replacements for the conventional microwave pulses used in today's hardware, but now with better error performance. As part of the manuscript the Q-CTRL team contributed example control solutions which provide robustness against hardware error, written in the OpenPulse standard.
Dr Hush said Q-CTRL's involvement gives a twofold advantage. First, Q-CTRL will benefit by being able to develop even better control solutions that sit just above the hardware level, tailored specifically for the IBM system.
And secondly he said that while a quantum algorithm programmer might not understand or even care about those details, they can leverage Q-CTRL's solutions and expertise to extract better performance from the hardware. Ultimately, that means they can achieve better results with today's NISQ quantum computers.
The real strength of this collaboration, Professor Biercuk said, is that by defining such control standards, they can be adapted and used outside the IBM environment to describe common controls.
"Q-CTRL's Quantum Firmware works across all qubit technologies. Building a common language for programming hardware devices can help ensure that all teams are able to leverage the huge advantages we offer with minimum friction," he said.