Quantum Machines is offering a combination of classical hardware and software for the control and operation of quantum processors; the company calls it the Quantum Orchestration Platform. QOP has a software interface for programming.
Quantum Machines is an Israeli startup, funded with a total of $22.5 million, led by groups such Battery Ventures, TLV Partners, Harel Insurance Group and Israeli investor Avigdor Willenz. It was founded by three PhDs in physics: Itamar Sivan, Yonatan Cohen, and Nissim Ofek. Facing the challenges of computing, they have directed their research toward quantum technologies.
Quantum computers could solve very complex tasks that are far beyond the capabilities of conventional supercomputers but, unfortunately, quantum states are extremely sensitive to interference from the external environment. Reducing this interference is one of two big challenges for quantum computing. The other is the construction of efficient and scalable hardware. Many complex error correction strategies have been devised to solve these problems.
The interest in quantum computing comes from the considerable amount of computing potential in quantum bits (qubits) which are exceedingly difficult to manage, both in terms of quantity and quality. Quantum Machines is developing new systems that aim to optimize the control of quantum systems.
“We always like the analogy of the quantum processor as a muscle, an extremely strong muscle that can perform extremely heavy lifting in terms of competitive power, but this power is useless alone and needs a brain to perform the tasks brilliantly. And that’s exactly what we have developed for quantum computers. We develop systems that make quantum processors work to realize their potential,” said Itamar Sivan, co-founder and CEO at Quantum Machines.
The capabilities of the new QOP platform include ultra-low feedback latency for applications ranging from quantum error correction to general multi-qubit control flow, while also offering comprehensive programming and classic real-time processing.
Qubit and quantum error correction
The fundamental element for quantum computation is the qubit (quantum bit). Unlike the conventional bit, a qubit can exist not only in the zero or one state, but it can also overlap both possibilities. Moreover, in a quantum processor, there can be more qubits in a state of superposition connected among them, to the point that they express a group behavior, called entanglement. This state of entanglement is the basis for the incredible computing power of quantum computers, and the source of their potential to solve complex tasks beyond the capabilities of traditional supercomputers.
The bad news is that quantum information is highly sensitive to environmental disturbances. This and other quantum peculiarities make error correction necessary to obtain useful results from the computation. The necessary operations for error correction are not only very complex, but they must also keep the quantum information unchanged.
The weaker magnetic field determines an inversion (“bit-flip”) on the qubits, which switch their possibilities to be |0⟩ and |1⟩ compared to the other qubits, or phase inversions (“phase-flip”) which switch the mathematical relationship between their two states.
For quantum computers, it is advisable to find schemes to protect information even when individual qubits are corrupted. In addition, these schemes must detect and correct errors without directly measuring the qubits, because the measurements collapse the coexisting possibilities of the qubits.
The correction of a quantum error is fundamental to most quantum computer projects, as it helps preserve the fragile quantum states on which quantum calculation depends.
Once a measurement has been made on the qubits, however, the overlapping condition collapses, and the qubits take on defined values. The key to the design of the quantum algorithm is to manipulate the quantum state of the qubits so that, when the overlapping condition collapses, the result is (with high probability) the solution to a problem.
Speaking of qubit control, we also refer to the problem of overheating. To avoid this, quantum machines in operation are typically placed in refrigerated environments around absolute zero. Researchers so far have focused on the construction of individual quantum “demonstration” systems, able to highlight the potential of these computers, and had to do so using the tools (although high performance) of traditional computer science and then using cryogenic refrigerators to keep the temperature of computers low.
Usually, hundreds or thousands of cables are needed to connect the computer to the chiller and control one qubit at a time: a tangle that prevents the creation of larger quantum systems, consisting of hundreds or thousands of qubits, systems that could perform much more complex calculations.
A classical computer consists of hardware and software whereas a quantum computers are hybrid machines that combine quantum features with a classical computer that, essentially, manages the quantum apparatus. The computing potential is located in the quantum processor. However, to run a quantum processor, you need dedicated classical hardware that is responsible for performing mathematical operations on quantum bits by sending electromagnetic pulses to the qubits.
Much of the industry attention has been focused on actual quantum processors, but as these machines become more powerful, it is the classical part — the digital command conversion system for use in the analog world of quantum computing — that is becoming a bottleneck.
Quantum Machines has built its own custom pulse system that can handle multi-qubit manipulation while being independent of the quantum processor with which it interacts.
“We always like to say that if you have a quantum processor with 300 quantum bits, it can store an unbelievable amount of information. If you wanted to store the same amount of information in a classical processor, it would take more classical transistors than the number of atoms in the universe. Now, this comes from something fundamental, and that is the fact that the complexity of quantum systems scales exponentially with the number of qubits. This is not the case with classical systems,” Sivan said.
“On the other hand, it is the classical system that we use to make quantum systems work in quantum processors,” he continued. “So, this is the fundamental reason why, if today you have a fantastic quantum processor with hundreds of qubits, it does not mean that you can run complex algorithms on it, it just means that you have hundreds of qubits with maybe good features, like coherence. And the main reason is this discrepancy between the quantum system you control and the one you want to run algorithms on in the classical system that controls it.”
Inside the classical processor, you see a lot of transistors that are not randomly positioned. They are connected to each other in very specific ways to incorporate the processor logic. “I could make a multiplier, for example, or if I connect more and more transistors, I could make a neural network,” said Sivan.
Classical processors have both the data and all the logic within it. Quantum processors work very differently. You may think oa quantum processor as a huge memory. If you want to apply a logical operation to it, you need to send impulses. Every impulse you send to the quantum processor will be a logical operation. This can be thought of as many lines on which you have trains of pulses hitting the qubits and how these pulses hit the qubits then run the algorithm on the quantum processor.
“Currently, we are working to scale the correction of quantum errors more and more, up to thousands of qubits and beyond. So, running the most complex quantum algorithms on next-generation quantum processors, presents a number of electrical engineering challenges,” said Sivan.
Sivan said that the largest multinational corporations and research organizations working in quantum computing are collaborating with Quantum Machine to adopt its platform. Any company or institution developing quantum processors can now purchase the Quantum Orchestration Platform to run the most complex algorithms possible.
Quantum Computing Programming Language
The fundamental software interface is its quantum assembler — QUA. Using QUA, QM’s pulse-level quantum computing programming language, QOP translates classical code into a quantum assembler language that can then be run on any quantum processor.
“There are many programming languages like C that are currently being used to program quantum computers. We believe that the thing that will make the most sense in the coming years is actually using a low-level language specifically designed for quantum computers, and that is what we created with QUA. The QUA language is not as an assembler, but instead is a low-level language in the sense that it is actually programming the most fundamental logical operations of quantum processors, similar to the most fundamental logical operations of classical processors. It is basically the lowest language for quantum computers. And of course, our goal is to standardize it, and it is one of the main candidates to become a standard for quantum computers,” said Sivan.
Sivan has made it clear that Quantum Machines assembly code is unique to QOP and can only work on its hardware. However, he added that QOP can be integrated into any quantum computer and used through QM programming languages, or any other programming language thanks to compilers or transpilers.
The Quantum Orchestration Platform appears to be an indispensable middleware which might make the programming of quantum computers easier.
Quantum encryption and artificial intelligence (AI)
Quantum computing will be able to solve a number of currently used cryptographies, such as RSA. “RSA is a hackable protocol even with classical computers but would take an immense amount of time. A quantum computer is believed to be capable of performing the exact same task within minutes, thus destroying the standard RSA encryption protocols we have. Of course, this will require full-scale quantum computers, which have not been built yet built,” said Sivan.
The use of quantum algorithms in artificial intelligence techniques will increase the learning capabilities of machines. In the field of computer security, as computers become smarter and faster, codes become easier to decode: that’s why a more advanced cryptography mechanism, like the one promised by quantum cryptography, is urgently needed.
AI can also contribute with the creation of new algorithms that are able to correct quantum and logical errors, the discovery of aspects not yet understood in quantum mechanics, the creation of new, more efficient hardware structures.
Artificial intelligence could also correct quantum errors. To correct them, the researchers of the Max Planck Institute have conducted studies using the same strategy that AlphaGo, the artificial intelligence program of Google DeepMind, had previously used to learn how to play Go to beat the world champion. In this case, the game, called Quantum Go, is aimed at preserving the qubit state: the artificial intelligence, called Alice, challenges an imaginary opponent who does everything possible to induce the quantum computer to error and destroy the qubit state. Alice must find the right moves to avoid it.