A blueprint for making quantum computers easier to program

When Peter Shor, a professor at MIT and now a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL), first demonstrated the potential of quantum computers to solve problems faster than conventional computers, he inspired scientists to imagine countless possibilities for emerging technologies. Yet thirty years later, the quantum edge is still an unreached peak.

Unfortunately, quantum computing technology is not yet fully operational. A major challenge is translating quantum algorithms from abstract mathematical concepts into concrete codes that can run on quantum computers. Programmers of ordinary computers can use languages ​​such as Python and C++ with structures consistent with standard classical computing abstractions, but quantum programmers do not have this luxury. Quantum programming languages ​​rarely exist today, and they are relatively difficult to use because quantum computing abstractions are still changing. In their recent work, the MIT researchers emphasized that this difference exists because quantum computers do not follow the same rules for each step of a program, providing all computers with a basic structure called control flow. process, and provides a new abstract model for quantum computers that may be easier to program.

In a paper to be presented at the ACM Conference on Object-Oriented Programming, Systems, Languages, and Applications, the group outlines a new conceptual model of a quantum computer, called a quantum control machine, that could bring us closer to writing programs As easy to write as a regular classical computer. Such an achievement would help power tasks that ordinary computers cannot efficiently complete, such as breaking down large numbers, retrieving information from libraries, and simulating how molecules interact to discover drugs.

Lead author Charles Yuan SM 22, a doctoral student at CSAIL, said our work presents principles that guide how to properly program quantum computers. Ordinary classical computers will eventually turn quantum computers into classical computers and lose their performance advantages. These laws explain why quantum programming languages ​​are tricky to design and point us to a way to improve them.

Old School Computing vs. New School Computing

One of the reasons why today’s classic computers are relatively easy to program is that their control flows are quite simple. The basic building blocks of a classical computer are simple: binary digits, or bits, which are simple collections of zeros and ones. These components assemble into the instructions and components of a computer architecture. An important component is the program counter, which locates the next instruction in the program by recalling the next direction from memory, much like a chef following a recipe. As the algorithm sequentially moves through a program, control flow instructions called conditional jumps update program counters to cause the computer to advance to the next instruction or deviate from the current step.

In contrast, the basic building blocks of quantum computers are qubits, which are quantum versions of bits. This kind of quantum data exists as zero and one at the same time, which is called superposition. Based on this idea, quantum algorithms can choose to execute a superposition of two instructions simultaneously, a concept called quantum control flow.

The problem is that existing quantum computer designs do not include program counters or the equivalent of conditional jumps. In practice, this means that programmers typically achieve control flow by manually arranging logic gates that describe computer hardware, a tedious and error-prone process. To provide these capabilities and close the gap with classical computers, Yuan Zheng and his co-authors created the quantum control machine, an instruction set for quantum computers that works similar to the classical idea of ​​a virtual machine. In their paper, the researchers envisioned how programmers could use this instruction set to implement quantum algorithms to solve problems such as digital factorization and simulating chemical interactions.

As a technical key to this work, the researchers demonstrated that a quantum computer cannot support the same conditional jump instructions as a classical computer and showed how to modify it to operate correctly on a quantum computer. Specifically, the instructions of a quantum control machine are reversible, meaning they can run forward and backward in time. Quantum algorithms require all instructions (including control flow instructions) to be reversible so that it can process quantum information without accidentally destroying its superposition and producing wrong answers.

The hidden simplicity of quantum computers

Yuan said you don’t need to be a physicist or mathematician to understand how this futuristic technology works. Quantum computers don’t have to be mysterious machines that require scary equations to understand, he said. With the Quantum Control Machine, the CSAIL team aims to lower the barrier to entry for people to interact with quantum computers by raising unfamiliar quantum control flow concepts to a level that reflects familiar control flow concepts in classical computers. By highlighting the considerations of building and programming a quantum computer, they hope to educate people outside the field about the power of quantum technology and its ultimate limits.

Still, the researchers caution that, like many other designs, it’s not yet possible to directly translate their work into a practical hardware quantum computer due to the limitations of today’s qubit technology. Their goal is to develop ways to implement a wider variety of quantum algorithms as programs that efficiently utilize a limited number of qubits and logic gates. Doing so will bring us closer to running these algorithms on quantum computers that come online in the near future.

Patrick Rall, a researcher at the MIT-IBM Watson Artificial Intelligence Laboratory (who was not involved in the paper), said that the basic functions of quantum computing models have been a central discussion in quantum computing theory since its inception. The earliest of these models were quantum Turing machines capable of quantum control flow. However, the field has largely moved toward simpler, more convenient circuit models, and quantum lacks control flow. Yuan, Villanyi, and Carbin successfully capture the underlying reasons for this shift from a programming language perspective. While control flow is central to our understanding of classical computing, quantum is completely different! I expect this observation to be critical to the design of modern quantum software frameworks as hardware platforms become more mature.

The paper lists two other CSAIL members as authors: doctoral student Gi Villnyi 21 and associate professor Michael Carbin. Their work was supported in part by the National Science Foundation and the Sloan Foundation.