Many decades have passed since Gordon Moore forecasted that the number of transistors on a microchip would double every two years, and with it, the computer’s speed and performance. Almost every technology we use daily, such as laptops, and smartphones, are displays of Moore’s observation.
However, lately, density, operations speed, and energy consumption of transistors on microchips have been challenging Moore’s tendency, which is limiting the ability to develop better classical computers. Solving complex industrial problems in a reasonable time and accurately is, in many cases, beyond the reach of state-of-the-art classical technologies. For this reason, experts have started to worry upon anticipating Moore’s law termination.
Quantum computing is a promising technology that has been showing its superiority over classical computers for solving specific tasks. This success has positioned the quantum approach as a top candidate to bring an advantage over classical computers to solve industrial problems in the near future.
A smart way to propel advantage of quantum technologies is to integrate quantum processing units (QPUs) into high performance computing centers. The idea of this classical-quantum hybrid approach is to delegate specific routines to the QPU while executing large computational tasks. However, the readily access to quantum computers is scarce, widening the gap between companies that need cutting edge technologies to develop their applications and technology. To give a fair chance to more users access to quantum computing, accelerate the integration process of QPUs in High Performance Computing centers, and tune the QPU role in hybrid, complex workflows, it is crucial to simulate quantum devices.
The first emulator to recreate neutral atoms’ hardware
Emulators are classical computing programs that simulate QPUs dynamics for specific architectures. At PASQAL, we have created the first emulator for neutral atoms' quantum processors called EMU-TN.
In neutral atoms processors, we shine fine-tuned lasers onto neutral atoms to create qubits, the quantum information units. Neutral atoms QPUs are one of the most scalable quantum technologies available, allowing us to manipulate hundreds of qubits to tackle complex problems. This is a great feature when it comes to calculating with quantum computers; however, it is a huge problem for classical computers when it comes to simulating QPUs.
Where is the problem? In quantum computing, each qubit is represented by a two-level quantum state system, and the number of quantum states we need to handle to compute the dynamics of a QPU grows exponentially with the number of qubits. This exponential growth burden classical computers when it comes to simulate their dynamics, motivating researchers to find techniques to simplify the calculations for a large number of qubits.
Our novel emulator, EMU-TN, is based on Tensor Networks, an approach that improves how the quantum dynamics escalate with the number of qubits, while reproducing accurate results.
Tensor networks are powerful structures we can use to represent complex quantum systems in an efficient manner. They are based on the concept of tensors—multi-dimensional arrays of numbers—which are used to compress the information that represents the quantum system. Even as the system size grows, useful properties of the ‘true’ quantum system can be efficiently computed using a series of specialized algorithms.
Implementing tensor networks is crucial, since one outstanding feature of EMU-TN is that it allows the full dynamics of the quantum system.
Recreating hybrid classical-quantum workflows
Our new emulator, which is designed to run integrated on high performance computing clusters, mimics the functioning of our QPUs recreating its programming interface. It will help the users assess their hybrid classical-quantum workflow and estimate the quantum resources needed to execute their algorithms. Furthermore, with EMU-TN, we not only simulate perfect, noiseless computers but the noise that may appear during a calculation, allowing the users to diagnose the QPU's performance.
In our paper, we show how we can use our platform to simulate the execution of different real-life applications suitable for hybrid architectures that use neutral atoms quantum devices using as much as 100 qubits. For example, in material sciences, we can efficiently simulate different phases of matter in 1D and 2D crystals. We can also study optimization problems that can be tackled using graphs.
EMU-TN is the first on-demand emulator specialized in quantum dynamics for neutral atoms that will help our users develop their quantum algorithms and complex hybrid classical-quantum workflows. This platform will significantly improve the research in the developing quantum industry, as well as allow a broad scope of engineers, scientists, and students to explore the potential of our quantum devices to solve the most pressing problems in science and technology.
Bidzhiev, K., Wennersteen, A., Beji, M., Dagrada, M., D'Arcangelo, M., Grijalva, M., Le Henaff, A-C., Quelle, A., Sashala Naik, A. Cloud on-demand emulation of quantum dynamics with tensor networks. Preprint available: 2302.05253.pdf (arxiv.org).