Classical Simulations Challenge Quantum Computer Claims of Speed

Augustine Kshetrimayum and colleagues at Multiverse Computing, in a collaboration between institutions in Spain, Canada, and Iran, critically review recent claims of quantum advantage following experiments conducted by IBM, D-Wave, and Google. The review examines the role of tensor network methods in classical simulations, assessing the capabilities and limitations of these classical approaches. The ongoing competition between quantum hardware and improved simulation techniques is driving progress in both fields. This analysis clarifies the conditions necessary for demonstrating genuine and scalable quantum advantage, identifying key challenges for current classical methods and outlining pathways for future advancements. Modelling quantum advantage experiments using tensor network decompositions Tensor network methods formed the key analytical framework for this work, representing complex calculations on a classical computer akin to how a network of interconnected nodes simplifies a complicated map. These methods decompose a large quantum system into smaller, interconnected components, dramatically reducing the computational resources needed for simulation by focusing on the most important relationships between particles rather than tracking every possible quantum state. This decomposition is achieved by representing the quantum state as a network of tensors, multidimensional arrays that capture the correlations between different parts of the system. The efficiency of tensor network methods stems from the fact that quantum states in many physically relevant systems are only weakly entangled, meaning that only a limited number of connections between the tensors are significant. Matrix product states (MPS) and projected entangled pair states (PEPS) were employed to model the quantum experiments conducted by IBM, D-Wave, and Google, each offering unique strengths for tackling specific system characteristics. MPS are particularly well-suited for modelling one-dimensional systems, while PEPS can handle two-dimensional systems and offer greater expressiveness for highly entangled states. The choice of tensor network architecture depends on the specific geometry and entanglement structure of the quantum system being simulated. Recent quantum advantage experiments from IBM, D-Wave, and Google were analysed using these methods to model their findings. The analysis concentrated on simulations of quantum dynamics, random circuit sampling, and quantum annealing, though specific qubit counts or temperatures were not detailed in this review. Quantum dynamics simulations involve evolving a quantum state over time, while random circuit sampling aims to generate probability distributions that are difficult to sample from classically. Quantum annealing, employed by D-Wave, is a heuristic algorithm for finding the minimum energy state of a system. This classical approach allows direct comparison with quantum hardware performance, clarifying the path towards scalable quantum advantage. It establishes a higher benchmark for assessing claims of quantum advantage and identifies areas where improvements are needed to accurately challenge claims of quantum superiority. Understanding the limitations of classical simulations is crucial for interpreting the results of quantum advantage experiments and determining whether the observed speedup is genuine or merely due to the limitations of the classical algorithms used for comparison. Tensor network simulations validate quantum computations up to 433 qubits Classical simulation of quantum experiments has sharply improved, now accurately modelling systems with up to 433 qubits, a substantial increase from previous limitations of around 50 qubits. This breakthrough enables a more rigorous assessment of claims regarding quantum advantage. The inability to simulate larger, more complex quantum systems previously hampered verification of whether a quantum computer truly outperformed its classical counterparts, but these advances establish a higher benchmark for demonstrating genuine quantum superiority. The exponential growth of the Hilbert space, the space of all possible quantum states, with the number of qubits makes classical simulation increasingly challenging. Tensor network methods mitigate this exponential scaling by exploiting the structure of the quantum state and representing it in a compressed form. Utilising IBM’s Osprey processor as a key test case, classical simulations now accurately model quantum systems containing up to 433 qubits. Experiments from IBM, D-Wave, and Google revealed that Matrix Product States and Projected Entangled Pair States are increasingly effective at verifying quantum computations, with the latter employing methods like Belief Propagation to enhance simulation capabilities. Belief Propagation is an iterative algorithm used to efficiently contract tensor networks, reducing the computational cost of simulation. Simulations of D-Wave’s quantum annealing experiments, which utilise flux-qubit annealers, still present unique challenges for these classical approaches. The highly complex and disordered nature of the energy landscape in quantum annealing makes it difficult to accurately model using tensor networks. While these simulations establish a higher bar for demonstrating genuine quantum advantage, they do not yet account for the potential benefits of hybrid quantum-classical algorithms or the scalability required for practical, real-world applications. Hybrid algorithms combine the strengths of both quantum and classical computation, potentially overcoming the limitations of either approach alone. Tensor network simulations extend classical limits for quantum system modelling Definitively surpassing the capabilities of even the most sophisticated classical computers is central to the pursuit of quantum advantage, but this benchmark is a moving target constantly challenged by innovations in simulation techniques. Recent progress utilising tensor networks has allowed modelling of systems containing hundreds of qubits, pushing the boundaries of what is classically possible. The development of more efficient tensor network algorithms, combined with advances in high-performance computing, has been instrumental in achieving this progress. Despite these advances, recent progress in classical simulation does not negate the value of pursuing quantum computers. The ultimate goal is not simply to simulate quantum systems on classical computers, but to harness the unique capabilities of quantum mechanics to solve problems that are intractable for classical computers. This ongoing competition between quantum hardware and classical algorithms is, in fact, beneficial, driving innovation on both sides and clarifying the precise requirements for demonstrating a true quantum advantage. The increasing power of classical methods is demonstrated by analyses of quantum advantage experiments undertaken by IBM, D-Wave, and Google. By rigorously assessing recent experiments, this work clarifies the limitations of current classical simulation approaches, and researchers continue to refine them, challenging the pursuit of quantum advantage despite their increasing sophistication and the limitations they encounter with larger, more complex problems. The interplay between classical simulation and quantum hardware is therefore a symbiotic one, with each field informing and advancing the other. Further research is needed to explore the limits of tensor network methods and to develop new classical algorithms that can effectively simulate quantum systems, as well as to identify the specific problems where quantum computers can provide a demonstrable and sustained advantage over their classical counterparts. The research revealed that classical simulations, utilising tensor networks, can now model quantum systems containing hundreds of qubits, challenging claims of quantum advantage. This matters because it highlights the need for increasingly complex quantum systems to demonstrably outperform these powerful classical methods. The study of experiments by IBM, D-Wave and Google suggests that continued development of both quantum hardware and classical algorithms is crucial. Future work will likely focus on refining tensor network techniques and identifying specific computational problems where quantum computers can offer a sustained advantage. 👉 More information 🗞 Quantum Advantage: a Tensor Network Perspective 🧠 ArXiv: https://arxiv.org/abs/2603.18825 Tags: