Exploring the Potential of Google's 53-Qubit Sycamore Processor

 

1. Introduction to Quantum Computing

Quantum computing is an emerging technology that harnesses the laws of quantum mechanics to process information. The foundation of quantum computing lies in quantum bits or qubits, which constitute a quantum processor. Qubits exhibit unique characteristics, crucial for the power of quantum computing. Unlike classical bits that can represent either 0 or 1, qubits can exist in multiple states simultaneously. This phenomenon, known as superposition, enables quantum processors to perform numerous calculations concurrently. Another vital property of qubits is entanglement, allowing qubits to be linked together, where the state of one qubit affects the state of another, regardless of the distance separating them. Entangled qubits can outperform classical bits in terms of computational power. When a qubit operation is conducted on entangled qubits, the qubit is considered an n-bit operation, affecting all qubits in the system, leading to an exponential growth of the system’s Hilbert space (2n). For quantum computers, harnessing qubits' superposition and entanglement is paramount for achieving a quantum advantage over classical computers (AbuGhanem & Eleuch, 2023). A widely recognized benchmark for assessing the capabilities of quantum processors is the implementation of quantum algorithms on random quantum states and their subsequent characterization. While a smooth characterization becomes intractable in the large limit, measuring local observables offers a feasible approach that complements exogenous probing.

The Quantum Supremacy Experiment executed on the 53-qubit Sycamore Processor exemplified this approach. Quantum computing is a rapidly evolving field with significant investments from major tech companies, research institutions, and governments. Progress in quantum hardware advancements, quantum software development, and quantum algorithms has fostered research in various industries, including finance, automotive, pharmaceuticals, energy, and cybersecurity. These companies aim to develop practical quantum applications to gain a competitive edge in the market. Quantum computing employs quantum-mechanical phenomena to manipulate information, offering capabilities beyond classical computers. Classical computing relies on bits as information carriers. A bit can exist in one of two states, 0 or 1, representing the off or on state of a transistor in a computer chip. In contrast, quantum computing employs qubits as information carriers. A qubit can exist in a state of 0, 1, or any quantum superposition of these two states, enabling quantum systems to process information differently than classical systems.

1.1. Basic Principles of Quantum Computing

To facilitate an understanding of the following sections describing quantum technologies in detail, key principles underlying quantum computing are delineated here. In particular, quantum circuit representations of algorithms are discussed, as these are most closely aligned with the goals of the experimental demonstrations. A quantum computer operates over quantum states that encode information using the principles of quantum mechanics. The basic unit of quantum information is a qubit, which is a two-level quantum system, such as the polarization of a single photon, the energy levels of an ion, or the ground and excited states of a superconducting junction. A qubit is analogous to a classical bit, but unlike a classical bit that may be in one of two states, 0 or 1, a qubit can be in a linear superposition of both states. A qubit in state ∣ψ⟩ can be expressed as a linear combination of the computational basis states ∣0⟩ and ∣1⟩:

∣ψ⟩ = α∣0⟩ + β∣1⟩,

where α and β are complex probability amplitudes that satisfy the normalization condition |α|^2 + |β|^2 = 1. A quantum system with multiple qubits can represent a large amount of information, as the system’s state space grows exponentially with the number of qubits N; specifically, it scales as 2^N. In a state ∣Ψ⟩ of N qubits, all possible basis states ∣x⟩ = |x_0, x_1, …, x_(N−1)⟩, with x_j ∈ {0, 1} for j = 0, 1, …, N − 1, exist such that the corresponding projections yield the probability p(x) of measuring the basis state ∣x⟩:

p(x) = ⟨Ψ∣x⟩⟨x∣Ψ⟩.

Similarly to classical computation, a quantum computation proceeds by a sequence of operations or gates acting on qubits. Each gate manipulates qubits and is described by a unitary operation that transforms the quantum state according to a unitary matrix U. The gate operation can be represented in matrix form, where the subscript denotes the number of qubits over which it acts. A quantum circuit is a schematic representation of a quantum algorithm, consisting of qubit lines and gate symbols, similar to classical circuit diagrams. Measurements are represented by boxes at the end of the circuit that project the quantum state onto basis states. Quantum gates can create superpositions and entangle qubits; thus, the power of a quantum computer originates from qubit entanglement, which permits interconnections that are impossible for classical bits. A common choice of basis states is the computational basis, but other measurement bases can be chosen to yield different probabilities. Rapid advancements in quantum computing hardware have been fueled by the development of quantum algorithms that exploit unique quantum properties. Among the many proposed algorithms, two are especially well-known: the integer factoring algorithm by Shor and the unstructured search algorithm by Grover. Shor’s algorithm has great significance for cryptography, as it can factor large integers in polynomial time, whereas the best-known classical algorithms run in sub-exponential time (S. Humble et al., 2018). Grover’s algorithm offers a quadratic speedup compared to the best classical algorithm for search optimization in unsorted databases or constraint satisfaction problems. Implementing quantum computing at scale presents great challenges, including isolating qubits from the environment to preserve coherence time while allowing control and readout. Further challenges include precise state preparation and calibration of gate fidelities below some error threshold.

2. Google's Quantum Supremacy Claim

In 2019, Google publicly claimed to have achieved a quantum supremacy, a landmark technological and scientific achievement, by a 53-qubit quantum processor, Sycamore. This triggered a wide discussion in the scientific community on the validity and the meaning of the claim, pointing to both challenges and new avenues of research (Kalai et al., 2022). A precise analysis of the quantum supremacy experiment, along with a review of results relevant to understanding the significance of this experiment historically, scientifically, and technologically are provided.

In a simple language, quantum supremacy is the point at which a quantum computer can perform calculations impossible to the most powerful classical computers. Attaining quantum supremacy is important, because if true, it implies that there is a family of computational problems that are irreplaceably quantum and cannot be efficiently simulated by classical computers (AbuGhanem & Eleuch, 2023). Google’s Sycamore quantum computer of 53 qubits performed a certain computation exponentially harder for classical computers, as evidenced by demonstrating the ability to sample from the output distribution of the computation, a random circuit sampling (RCS), which would take a classical supercomputer an infeasibly long time.

2.1. What is Quantum Supremacy?

What is Quantum Supremacy?

The term “quantum supremacy” designates the benchmark at which quantum devices outperform classical computers on a well-defined task ( (Kalai et al., 2022) ). As a milestone in the quantum computing race, quantum supremacy is a common goal pursued by academia and tech giants alike. However, there are important distinctions to be made. Quantum supremacy is a computational viewpoint focused exclusively on the class of problems solved, while quantum advantage considers also the speed-up angle and cross-over point with respect to the best classical solution. Furthermore, quantum supremacy is a general notion that can be claimed on an abstract device with no prior knowledge on its inner workings. On the other hand, statements of quantum advantage are often device-specific.

Broadly speaking, a quantum device can claim supremacy if three conditions are met: (i) it has been certified as a quantum device, (ii) a classically hard problem has been experimentally solved on it and (iii) the results have been empirically validated. As the third condition holds trivial in the quantum domain, much effort goes in devising witness/benchmarking strategies to ensure the fidelity of the quantum operations during the computation. While discussions about the feasibility of achieving quantum supremacy date back to the early days of quantum computing, attempts at experimental demonstrations have been more recent. Google’s announcement of quantum supremacy with its 53-qubit Sycamore processor in 2019 has been the most publicized demonstration. However, it was not the first such attempt, as some early NISQ devices from other researchers had run quantum algorithms on them with large classical simulation overheads.

2.2. Sycamore Processor: Technical Overview

Google's Sycamore processor is designed to conduct quantum computational experiments with a 53-qubit superconducting chip. The Sycamore processor consists of 53 superconducting qubits in a planar configuration, connected by 74 fixed-frequency transmon couplers. Qubits are implemented with active on-chip controls to perform one-qubit gates, frequency multiplexed off-chip readout for state measurement and energy relaxation time estimation, and parametric couplings for entangling two-qubit gates. In addition to gate-based quantum computations, each qubit’s energy spectrum can be dynamically tuned to ensure desired interactions between a specified pair of qubits while keeping other qubits uninteracted, thus allowing for a Variable-Qubit Mode (VQM) architecture (Harris et al., 2010). This experiment for near-term quantum processors demonstrated two types of fixed-frequency transmon qubits—the high coherence qubits and the low-frequency control qubits. A careful balance was established between maximizing operational bandwidth and ensuring sufficient qubit-qubit coupling strength. Flexibility in control design is expected to allow for further improvement of individual qubit performance.

Remarkable features of Sycamore’s architecture include a focused approach to scalable quantum designs, where the ability to independently tune qubit frequency is incorporated into a coaxially-designed VQM and parallel classical control method to significantly increase the number of connectable qubits. This design has achieved a qubit layout with several thousand micrometers qubit-to-qubit connectivity distance without decreasing coherence times. After compilation with an efficient noise-aware two-qubit scheduling algorithm, Sycamore’s superconducting system performs a diverse series of 2-, 3-, and 4-qubit gate operations with consistency better than 99.8%. Despite the classical control chip’s thermal complexity, this system has demonstrated a measurement fidelity better than 99% across a 27-qubit chip. Sycamore provides a high-quality platform with exhaustive characterizations for submitting research proposals on novel algorithms, designs, and applications with near-term quantum devices.

Skeptics argue that the 53-qubit Sycamore processor’s independence cannot be verified due to various complexities. Employing a qubit variability approach, Sycamore characterizes operator fidelity at the process level with fixed feedback. The fidelity errors model a sum of systematic and random errors. Experimental data results and simulations suggest that parasitic cross-talk dominates coherent errors, so periodic on-chip calibrations are essential. Sycamore experiments suggest that multiqubit gates may be necessary to achieve strategic speedup in quantum systems over classical ones. Scalability challenges persist in all quantum processors, especially in tailoring gate size and qubit coupling. Ideal coupling limits any speedup over purely classical simulation. Yielding fault-tolerance and overhead qubit requirements are significant challenges, particularly in superconducting approaches where 2-qubit gate fidelity demands exceed 99%. Processes and results explored here add to a developing landscape of quantum technologies and complement complementary theoretical and experimental investigations.

3. Applications and Implications of the Sycamore Processor

Google’s 53-qubit Sycamore processor is discussed, summarizing the key innovations in its architecture and control systems, as well as its results from a demonstration of quantum supremacy. While Sycamore is engineered to showcase quantum means of computation, the underlining technologies and concepts can support a prodigious range of computational applications across music, art, science and economics. This includes the hosting of quantum processors on the internet, enabling new work practices and modes of collaboration. Opportunities to explore quantum advantages over classical means of perception and computation are highlighted.

With the Sycamore processor, quantum hardware is now available to explore applications beyond the technology’s initial mission of demonstrating quantum advantage. Some focus on the processor’s impact on specific professions and sectors, while others consider the broader societal implications. Quantum cryptography is examined first, addressing fundamental limits to security offered by the laws of quantum physics and how they can be employed to create secure systems. Symmetric key cryptography in classical computing is described, outlining how quantum key distribution exploits the entanglement of quantum states transmitted over a network to enable perfectly secure shared symmetric keys.

Sectors such as economics, finance, logistics and resource allocation rely on complex systems that underappreciated classical means of simulation. The potential of using the Sycamore processor to implement Ising spin Hamiltonians is detailed, in which physical qubits simulate a cost function defined by a network’s connectivity. Ground states of these Hamiltonians encode optimal solutions, and quench dynamics can be adopted to explore the energy landscape. Since machine learning has been key to interpreting classical simulations of quench dynamics, the complementary use of quantum machine learning to address artificial intelligence applications is noted. The transformative impact of such applications is conjectured, especially as neural network architectures deepen. These themes are explored through prose and images, and discussions with quantum computing experts, researchers and industry leaders.

As with previous computing revolutions, the widespread adoption of quantum computing raises ethical considerations. Systemic questions are asked regarding who will benefit most from quantum advancements, how to ensure equitable access, and what will be done to safeguard sensitive data for industries such as defense and health. Technology and society have always gone hand-in-hand, and as a new chapter unfolds, the precursors of quantum computing strive to connect the dots between advancements in technology and the desired outcomes for society as a whole.

A panoply of quantum systems are now at a maturity level that encourages exploration of computational applications. Some are fundamentally different from Sycamore in that they seek to exploit transitions between well-characterized many-body states, with thermalization controlling precision or time-stamp computation modes. Others share Sycamore’s architecture or use variants on small scale superconducting processors. Addressing the need for standardized performance benchmarking that go beyond fidelity and gate-error rates, figures of merit are proposed for characterizing implementation of quantum algorithms, focusing on Hamiltonian simulation with quench dynamics. Benchmarks are exampled for several physical systems, including Sycamore, and guidelines on how to experimentally realize them are provided.

To date, all demonstrations of quantum advantage employ either random circuit sampling or variational Hamiltonian simulation. Driving a quantum system with unreduced complexity ensures that bringing classical means of emulation within the rules of the game becomes intractable, scouting for some reliable figure of merit becomes paramount. Classical emulation can be dramatically accelerated by running computations on high-performance computing clusters. Reconciling classical emulation with exponential alignment is forbidding because it implicitly assumes an ensemble of quantum states is being handled, all demonstrations to date rely on a fixed preprocessing of initial pure states which don’t thermalize.

3.1. Quantum Cryptography

Quantum cryptography is one of the most studied and promising applications of quantum computing technology. At the heart of quantum cryptography is the concept of quantum key distribution (QKD). Using the laws of quantum mechanics, it is theoretically possible to offer security that cannot be broken, regardless of the computational power available to an attacker (Jobair Hossain Faruk et al., 2022). This is a significant advantage over classical cryptography methods, where security relies on the intractability of certain mathematical problems. So far, all attempts to break the cryptographic security of classical systems have focused on exploiting flaws in their implementation. QKD systems are resilient to eavesdropping; however, like any practical security system, they must be implemented correctly.

A number of protocols and methodologies are utilized in quantum cryptography, one of the seminal efforts in the field was BB84, developed in 1984 (Shenoy H. et al., 2018). To date, the most mature quantum cryptography systems are QKD systems. Although quantum cryptography has been extensively studied, scaling it to larger networks has proven challenging. One of the hardest problems is how to implement quantum cryptography on a larger scale and integrate it into existing infrastructures. There is a need for systems that are simple to understand, operate, and maintain. Additionally, there is a clear desire for systems that use as much commercially available technology as possible. If quantum cryptography systems become practical, it would render much of today’s cybersecurity obsolete. Assurance in confidentiality would not only protect communications from today’s adversaries but from all conceivable future threats as well.

Research will focus on making quantum cryptography as simple, practical, and accessible as possible. During this process, questions will be posed and considerations made to implement quantum cryptography systems that are workable and meaningful beyond proof-of-principle demonstrations. Ultimately, the goal is to develop quantum cryptography systems that, as cybersecurity systems, could be used to protect classical networks against future quantum threats. In broad strokes, an overview of the state-of-the-art concerning quantum cryptography as affected by the Sycamore processor follows.

3.2. Optimization and Machine Learning

The Sycamore processor is investigated for further applications beyond its proof-of-principle demonstration of computational supremacy experiments. Its transformative potential is outlined for optimization and machine learning applications and how they can benefit from quantum computing approaches. Optimization problems present in several fields, from logistics and telecommunications to finance and drug discovery, are NP-hard. Finding solutions that are as close as possible to optimal is crucial, and quantum approaches might yield faster solutions than state-of-the-art classical methods (Abbas et al., 2023). D-Wave Systems have developed quantum annealers for similar optimization problems using a completely different architecture from Sycamore with a much larger number of qubits, albeit at an inferior quantum volume.

The focus is given to implementing a quantum algorithm on a gate-model machine with low hardware budgets rather than the more mature quantum firing ansatz model. In addition and perhaps equally important for the Sycamore processor, the potential use of a quantum advantage machine for enhancing classical machine learning models is considered. It has been demonstrated that machine learning models can exploit quantum speedup for some tasks, particularly those related to pattern recognition and data-processing operations (Salehi et al., 2022). In general, the hope is that for large datasets, a suitable quantum algorithm would dramatically reduce the processing time compared to the best classical algorithm. For some quantum approaches, a polynomial speedup is expected, and for others, classically intractable problems can be exponentially faster with quantum computation than the best-known classical methods. In addition to expected speedup, quantum computing can have an entirely new computational basis, and the combination of classical and quantum systems could yield better and more efficient models for artificial intelligence. That quantum computing is still in its infancy is no secret, and all proposed experiments come with plenty of caveats. Most notably, insight into the implementation of quantum algorithms for real-world applications beyond toy problems is still lacking, as formulating robust quantum algorithms on larger problem scales is complex and difficult.

However, it has been thought-provoking to consider potential future directions in research and development for these two paths linked to the same Sycamore technology and for which the expertise would be most relevant. It must be noted that exploring original problems or their simpler versions would often be computationally intractable for classical computers. Despite all quantum technology challenges, the “moon-shot” approach makes sense.

4. Challenges and Future Developments

Currently, there exist many challenges that are both novel and deeply entrenched. Among the many questions to be resolved going forward are how to best benchmark quantum hardware and algorithms, and how to best and most efficiently scale those systems up in complexity. Even among noiseless quantum benchmarking protocols, the many viable figures of merit and what they signify in terms of quantum state preparation, evolution, measurement, and implementation complexity must be diligently thought through. Near-term quantum processors will come with the promise of opportunities, yet they will be hampered by poor connectivity, limited coherences, and high gate-error rates, all of which dictate trade-offs in algorithmic choice and hardware benchmarking strategies (AbuGhanem & Eleuch, 2023).

Quantum processors that implement two-qubit gates through a combination of microwave control pulses and engineered resonant couplings to local bus resonators come with many independent tunable parameters. The calibration of these quantum processors is a delicate undertaking that requires balancing competing protocols to achieve the desired implementation fidelity of two-qubit gates. Future experimental implementations of multi-qubit gates will likely expose further experimental challenges. Sophisticated control techniques must be accompanied by a theory of quarum states that quantifiably isolate relevant state manifolds and address issues of robustness to experimentally unavoidable imperfections. The ideal theory of quarum states will likely be geometry driven, unifying considerations drawn from optimal control, quantum information theory, information geometry, and statistical mechanics considerations of entropy and temperature. To realize practical quantum systems, a panoply of considerations must be addressed, almost all of which are inter-related.

References:

AbuGhanem, M. & Eleuch, H. (2023). NISQ Computers: A Path to Quantum Supremacy. [PDF]

S. Humble, T., Thapliyal, H., Munoz-Coreas, E., A. Mohiyaddin, F., & S. Bennink, R. (2018). Quantum Computing Circuits and Devices. [PDF]

Kalai, G., Rinott, Y., & Shoham, T. (2022). Google's Quantum Supremacy Claim: Data, Documentation, and Discussion. [PDF]

Harris, R., W. Johnson, M., Lanting, T., J. Berkley, A., Johansson, J., Bunyk, P., Tolkacheva, E., Ladizinsky, E., Ladizinsky, N., Oh, T., Cioata, F., Perminov, I., Spear, P., Enderud, C., Rich, C., Uchaikin, S., C. Thom, M., M. Chapple, E., Wang, J., Wilson, B., H. S. Amin, M., Dickson, N., Karimi, K., Macready, B., J. S. Truncik, C., & Rose, G. (2010). Experimental Investigation of an Eight Qubit Unit Cell in a Superconducting Optimization Processor. [PDF]

Jobair Hossain Faruk, M., Tahora, S., Tasnim, M., Shahriar, H., & Sakib, N. (2022). A Review of Quantum Cybersecurity: Threats, Risks and Opportunities. [PDF]

Shenoy H., A., Pathak, A., & Srikanth, R. (2018). Quantum cryptography: key distribution and beyond. [PDF]

Abbas, A., Ambainis, A., Augustino, B., Bärtschi, A., Buhrman, H., Coffrin, C., Cortiana, G., Dunjko, V., J. Egger, D., G. Elmegreen, B., Franco, N., Fratini, F., Fuller, B., Gacon, J., Gonciulea, C., Gribling, S., Gupta, S., Hadfield, S., Heese, R., Kircher, G., Kleinert, T., Koch, T., Korpas, G., Lenk, S., Marecek, J., Markov, V., Mazzola, G., Mensa, S., Mohseni, N., Nannicini, G., O'Meara, C., Peña Tapia, E., Pokutta, S., Proissl, M., Rebentrost, P., Sahin, E., C. B. Symons, B., Tornow, S., Valls, V., Woerner, S., L. Wolf-Bauwens, M., Yard, J., Yarkoni, S., Zechiel, D., Zhuk, S., & Zoufal, C. (2023). Quantum Optimization: Potential, Challenges, and the Path Forward. [PDF]

Salehi, T., Zomorodi, M., Plawiak, P., Abbaszade, M., & Salari, V. (2022). An optimizing method for performance and resource utilization in quantum machine learning circuits. ncbi.nlm.nih.gov