Quantum Orchestration Platform: A virtual machine for quantum computing

Startup Quantum Machines unveils what could be a key to accelerating quantum computing adoption, already used in production
Written by George Anadiotis, Contributor

We would not blame you if you've never heard of Quantum Machines before -- we had not either. Quantum Machines is an Israeli startup, funded with $5.5 million by Battery Ventures and TLV Partners, and founded by three physics PhDs: Itamar Sivan, Yonatan Cohen, and Nissim Ofek. They all spent many years in top universities working on the cutting-edge of quantum computing. 

All three of them began their PhDs with the expectation of continuing along the academic path. During their doctoral studies, however, Cohen and Sivan were part of the founding team of the Weizmann Institute Entrepreneurship Program. Once they were exposed to entrepreneurship, there was no turning back.

When the three of them finished their doctorates, it was clear that the field in which they had been active for so many years was finally 'happening.' They understood that they needed to address a challenge that was holding the industry back. Taking on the layers of computing, they found an unmet need, and a bottleneck whose solution would drive the industry forward today and in the future.

Hardware and software layers

Sivan, co-founder and CEO of Quantum Machines (QM), explained that while a classical computer has two fundamental layers being hardware and software, the quantum computer in all mainstream realizations has three layers, being quantum hardware, classical hardware, and software.

The quantum processor is, above all, where the potential for immense computational power lies. However, to operate a quantum processor, dedicated classical hardware is required. This classical hardware is responsible for performing mathematical operations on the quantum bits by sending electromagnetic pulses to the qubits.

Quantum Machines developed the Quantum Orchestration Platform (QOP): A full hardware and software solution, which they claim has the most advanced classical hardware worldwide for the operation of quantum processors. In addition, QOP offers a convenient software interface for the seamless programming of even the most complex algorithms. 


Quantum Machine's founders have deep expertise in quantum computing

Early adopters include many of the leading players in the quantum computing race, including multinational corporations, startups, government labs, and academic institutions. Although names were not disclosed, Sivan said QM's customers are among the biggest names leading the race towards building scaled-up useful quantum computers, and include teams in six countries working on multiple different qubit platforms.

Sivan emphasized that QOP can be directly integrated with any quantum processor: "Basically, any company or institution developing quantum processors may now buy the Quantum Orchestration Platform, and right away be able to run the most complex algorithms possible."

What this means, he went on to add, is that introducing the Quantum Orchestration Platform is more than an accomplishment for Quantum Machines alone: It's a cross-industry platform that can benefit all the players and drive the field as a whole forward through its powerful capabilities.

These capabilities include ultra-low feedback latency for application ranging from ultra-rapid calibrations to quantum-error-correction, general multi-qubit control-flow including real-time branching based on data acquired and processed from multiple qubits, and fully parametric programming of gates, pulses, and even real-time classical processing.

A quantum computing programming language

The fundamental software interface of QOP is its quantum assembly -- QUA. Using QUA, QM's quantum computing programming language, QOP translates classical code into a quantum assembly language that can then run on any quantum processor. Sivan said QUA is a higher-level language that can be used intuitively to program quantum algorithms.

As this was not entirely clear to us, we asked Sivan whether this quantum assembly language can be applied to any underlying hardware. Sivan clarified that QM's assembly code is unique to QOP and can only run on its hardware. However, he went on to add, QOP can be integrated into any quantum computer.

Other APIs and programming languages can easily interface with QM's assembly, Sivan said. This is something QM is doing with multiple players that have developed their own programming languages. There is common ground to all quantum computers, and that is, how quantum algorithms are orchestrated on them: 

"QM has built its hardware and software stack in the most general way, which applies to all quantum computers, and there was no need to work with specific companies developing quantum processors to make it happen. Deep knowledge and expertise of the quantum control stack among QM's team is what has allowed us to do this successfully."


Quantum Machines has built the equivalent of a virtual machine intermediate layer to make programming quantum computers more accessible

When asked if QM thinks this could work as a kind of standard to accelerate quantum computing progress and adoption, Sivan concurred. 

"Quantum processors hold great promise for immense computational power. The main reason why we do not have full-scale quantum computers yet is that to scale-up the technology, the quality of qubits will have to substantially improve alongside the increasing number of qubits in a quantum processor. 

However, it is a fact that even today, most teams cannot realize the potential of the quantum processors they currently have. It is for exactly that reason that QOP is being introduced and offered to all the industry players -- for them to realize the potential of their quantum processors today and be prepared for the quantum breakthroughs of tomorrow.

From near-term applications of quantum computers to monumental challenges like quantum-error-correction, the most complex algorithms can be programmed and run. Furthermore, beyond useful algorithmics, even development and optimization-focused algorithms such as the quantum processors' calibrations can be run ultra-rapidly with QOP.

Given that more and more players are beginning to use QOP, we surely anticipate an acceleration of the field as a whole and its faster progress towards useful applications."

Write once, deploy anywhere

What about interfacing with other programming languages? Sivan said that once integrated with a quantum processor. The system can be used as-is, through QM's programming languages, or any other programming language thanks to compilers or transpilers, which can be used to translate from one language to another.

Sivan added, QOP is universal in the sense that it can both support all quantum computing programming languages, and is also qubit-agnostic and can run algorithms on any quantum processor.

Last but not least, noting that there are vacancies for machine learning experts in QM piqued our curiosity. What does QM use machine learning for? Internal apps only, or is translating machine learning models in the roadmap too? If yes, at what stage is this now?

Sivan noted this relates to something very fundamental and offered an example to illustrate it. To merely describe the information contained in a quantum processor with 300 quantum bits, he said, one will need a classical processor with more transistors than the number of atoms in the universe. 


In order to describe the information contained in a quantum processor with 300 quantum bits, one will need a classical processor with more transistors than the number of atoms in the universe.

Recursion Pharmaceuticals

This is not an anecdote, but rather, it stems from something extremely fundamental: The complexity of quantum systems grows exponentially with the number of units (being the quantum bits). This is not the case for classical systems, which are the systems used to operate the quantum processors.

This discrepancy is the underlying reason why the development of the Quantum Orchestration Platform is extremely challenging, for which it is only natural that machine learning and AI algorithmics is leveraged. For example, QM utilized machine learning for the optimization of pulses sent to qubit and aimed and reading their quantum states.

QM also uses neural-nets in real-time to infer whether there are errors in the quantum bits or not, as well as for the algorithmics of the compiler. Sivan believes it is clear today that ML will play a significant role in the development of quantum computers, and that we are now only seeing the tip of the iceberg of this exciting and challenging domain.

Instead of going after the development and optimization of quantum processors per se, Quantum Machines chose to focus on the access layer. The Quantum Orchestration Platform looks like a much-needed middleware that can make programming quantum computers easier. A virtual machine of sorts.

Much like the Java Virtual Machine enabled the "write once, deploy anywhere" motto to become a reality, and paved the way for Java's dominance as an enterprise software programming language, the QOP looks like it has the potential to be equally transformational for quantum computing.

Editorial standards