Four extra qubits enough for some quantum simulations

Researchers at Mitsubishi UFJ Financial Group, Inc., alongside colleagues at Keio University and Toyota Motor Corporation, have developed quantum algorithms achieving exponentially accurate open quantum simulations with a surprisingly limited resource requirement. The work details methods for simulating complex physical phenomena using either four plus the base-2 logarithm of a parameter ‘M’, representing the maximum number of Pauli strings in a single jump operator, or seven qubits total, a substantial reduction compared to previous approaches.

These new randomized algorithms achieve logarithmically short circuit depth, scaling accuracy without being limited by the parameters of the simulated system, though with specific conditions for parameter independence as detailed in Theorems 2 and 3. This contribution promises to make open quantum system simulations more feasible on early fault-tolerant quantum computing devices.

Randomized Dissipation for Accurate Open Quantum Simulation

Open quantum system simulations can now be achieved with a remarkably small increase in qubit count; specifically, four plus the base-2 logarithm of a parameter ‘M’, representing the maximum number of Pauli strings in a single jump operator, or seven qubits total are needed. The team’s approach minimizes gate counts while preserving a circuit depth that scales logarithmically with desired accuracy, an improvement over previous methods where computational time scaled as O(T^2/ε).

These algorithms address a longstanding challenge in simulating open quantum systems, where accurately representing the loss of quantum coherence, known as dissipation, typically demands a substantial number of ancillary qubits. Previous methods often prioritized efficient circuit depth, demanding a larger number of ancilla qubits for comparable accuracy, hindering progress toward simulations of complex physical phenomena. This is accomplished through a novel random circuit compilation method that uses dissipative processes with a single jump operator.

The researchers further refined their approach by concatenating random compilation with an alternative encoding of jump operators, effectively removing dependence on some parameters specifying the Lindbladian. Instead of directly encoding these operators, they utilized randomized Hamiltonian simulation, treating the real and imaginary parts as Hamiltonians. This technique achieves high accuracy due to the inherent randomness, and when combined with quantum singular value transformation, further optimizes the circuit complexity.

“We can concatenate this alternative encoding and the random compilation of Lindblad dynamics,” the paper explains, detailing how this process minimizes the remaining parameters influencing the simulation. Notably, the team draws a connection between their algorithm and the stochastic Schrödinger equation (SSE), a widely used method for simulating open quantum systems.

While both approaches employ stochastic applications of unitary and jump operations, the new method’s dissipative process differs from SSE as it relies on a composition of minimal dissipation and potentially non-physical superoperators. This distinction highlights a unique characteristic of the randomized dissipation technique, offering a potentially advantageous pathway for future quantum simulations and contributing to the feasibility of open quantum system simulations on early fault-tolerant quantum computing devices.

Lindblad Dynamics and Open Quantum System Analysis

The work demonstrates that these algorithms can achieve a desired accuracy with minimal ancilla qubits, 4 + ⌈ log_2M⌉ in the first case and 7 in the other, offering flexibility in modeling complex scenarios. The researchers numerically confirmed the practical advantages of their method through detailed analysis of gate and ancilla counts, revealing a clear benefit over existing approaches. This analysis indicates a potential for significant speedups and resource savings in simulating complex quantum phenomena, such as those found in many-body quantum systems and reservoir engineering.

“We presented a randomized quantum algorithm for estimating the physical properties of the general Lindblad dynamics,” the paper states, highlighting the algorithm’s exponentially short circuit depth and near-constant ancilla qubit requirements. Further investigation will focus on improving the gate complexity and potentially achieving even higher-order scaling while maintaining minimal ancilla usage, though constructing such efficient higher-order schemes for Lindblad dynamics presents a conceptual hurdle, unlike the established product formula used in unitary dynamics.

Logarithmic Scaling of Accuracy in Quantum Simulation

This represents a departure from approaches demanding a large number of ancilla qubits to coherently encode all possible jump operators, a hurdle for scaling simulations. These algorithms achieve a circuit depth that scales logarithmically with desired accuracy, and either partial or complete independence from the parameters specifying the Lindbladian. While recent randomization techniques for simulating Lindblad dynamics share some characteristics, they have not yet achieved this logarithmic gate complexity concerning accuracy, or ‘ε’.

The work details a decomposition of the dissipative process using a single jump operator, enabling simulation of the resulting superoperators with quantum circuits. This approach uses a transfer matrix representation of superoperators, building on Taylor expansion of the dynamical map, and allows for a more efficient use of quantum resources. Numerical analysis confirms the improvements in practical parameter regions, demonstrating the impact of reducing both gate and ancilla counts.

The algorithms rely on a process of repeated applications, achieving a 1 + O(t/r) scaling, where ‘r’ represents the number of repetitions, providing an upper bound on simulation error. Although the remaining dependence on ‘M’ can be removed, it requires a polynomial increase in gate count and potentially affects the scaling of simulation time and desired accuracy. “We may use near-term quantum algorithms that work with fewer ancilla qubits, but they typically require a large sampling overhead to achieve an accurate simulation, leading to longer end-to-end runtime,” the researchers note.

Minimal Ancilla Qubit Requirements: 4+⌈log₂M⌉ Algorithm

This contrasts with fully fault-tolerant algorithms demanding a larger number of ancilla qubits for comparable accuracy. The work details a trade-off, accepting a potentially increased circuit depth in exchange for minimizing qubit overhead, a strategy justified by numerical analysis detailed in the paper. A formalism introduced by the researchers enables simulation of superoperators, maps describing the evolution of quantum states, with minimal ancilla qubit usage; adding a single ancilla qubit allows for the simulation of these superoperators on a system of n+1 qubits.

This approach utilizes an anti-linear map, denoted ‘J’, which transforms a quantum state ‘A’ into its adjoint, ‘A^†’, facilitating efficient representation within the quantum circuit. Parameter values for simulations of the transverse-field Ising model (TFIM) and the Fermi-Hubbard model (FHM) are provided, including hopping amplitudes and two-body loss rates, allowing for direct comparison with existing methods.

Detailed analysis reveals the performance of the new algorithm, designated Theorem 2, relative to channel LCU and first-order HS-based methods, specifically regarding gate counts and ancilla qubit requirements for both TFIM and FHM simulations. Figures included in the research illustrate the dependence of these requirements on both epsilon (ε), representing desired accuracy, and tau (τ), a parameter related to simulation time. The data demonstrates that for early fault-tolerant quantum computing devices, expected to be limited in both gate count and logical qubit capacity, this algorithm presents a practical and viable option.

Gate Complexity Comparison: Existing vs. New Methods

A reduction to four plus the base-2 logarithm of a parameter ‘M’, representing the maximum number of Pauli strings in a single jump operator, or seven qubits suffices for exponentially accurate open quantum simulations, a departure from prior methods demanding significantly more resources. This minimized ancilla count distinguishes the new algorithms detailed in the work, alongside a focus on parameter independence, meaning performance isn’t heavily reliant on the specifics of the simulated system.

Detailed analysis reveals a trade-off in circuit depth, where the algorithms accept a τ⁴ scaling factor stemming from concatenating randomized techniques, but justify this through numerical analysis, suggesting it’s a worthwhile compromise for the gains in qubit efficiency.

While some fully fault-tolerant algorithms demand a larger number of ancilla qubits for comparable accuracy, the researchers demonstrate that their method, designated Theorem 2, can achieve gate complexities of O(τ⁴ log³(τ/ε)), without dependence on K or m, parameters defining the system, offering an advantage over previous approaches. This combination of features, logarithmic scaling in accuracy, low ancilla count, and independence from parameters m and K, is notable, as many existing methods struggle to optimize all these aspects simultaneously. Further investigation will compare gate counts for highly complicated dissipative systems, but the current results demonstrate a practical advantage over existing methods, particularly for systems with a large M and bounded norm.

Theorem 2: K & m Independent Gate Complexity

Gate complexity for simulating open quantum systems can now be achieved without scaling with key system parameters, a result detailed in newly published work. Theorem 2 demonstrates that the number of gates required for accurate simulation remains independent of both K, representing the number of jump operators, and m, a parameter defining the system, a departure from many existing methods where ancilla requirements increase alongside these values.

This independence represents a refinement in algorithm design, potentially unlocking more efficient simulations. “Gate complexity in Theorem 2 is independent of K and m,” the researchers state, highlighting a core benefit of their approach. The ability to decouple gate complexity from K and m is particularly notable given the challenges of scaling quantum simulations.

Existing proposals often see ancilla requirements grow with multiple parameters, including M, K, ε, or n, creating bottlenecks for complex systems. By achieving this independence, the new algorithms offer a pathway toward simulating larger and more intricate open quantum systems with existing and near-term quantum hardware. Theorem 3 expands on this, achieving independence from K, m, and M, though with a gate complexity of O(τ⁴ log³(τ/ε)).

The algorithms utilize a time slicing method, effectively simulating the time evolution of the system with a sampling overhead and a minimal number of ancilla qubits, 4 + ⌈ log_2M⌉ in the first case and 7 in the other, independent of the number of terms M in the system’s Hamiltonian. This combination of features positions these algorithms as a viable option for early fault-tolerant quantum computing devices, where qubit resources are still limited.

Theorem 3: Reduced Ancilla with O(τ⁴ log³(τ/ε)) Complexity

Achieving open quantum system simulations with just seven ancilla qubits is now possible, according to newly detailed algorithms that minimize resource demands alongside computational accuracy. The work demonstrates a circuit depth of O(τ⁴ log³(τ/ε)), offering a specific complexity bound for these simulations. This reduction in qubit overhead stems from a design that achieves independence from parameters m, K, and M, factors defining the system Hamiltonian and jump operator count, a feat not commonly found in current proposals.

While some algorithms offer logarithmically short circuit depth for accuracy, they often require encoding all possible jump operators, leading to substantial ancilla consumption. Numerical analysis justifies this exchange, suggesting a practical benefit for early fault-tolerant quantum computers.

The algorithms utilize a randomized technique, incorporating additional sampling to maintain accuracy. “The same problem as Theorem 2 can be solved by using O(|O|^2 log(1/δ)/ε^2) samples from a set of quantum circuits that use 7 ancilla qubits,” the researchers state, highlighting the efficiency of this approach. These techniques are not limited to open system simulations, potentially extending to a wider range of quantum computing challenges. The team presents a τ⁴ scaling factor stemming from concatenating randomized techniques, but justifies this through numerical analysis, suggesting it’s a worthwhile compromise for the gains in qubit efficiency.

Observable Estimation & Advantage Over State Preparation

Observable estimation offers a distinct path to extracting physical information from quantum systems, bypassing the need for full state reconstruction through tomography. Several recent studies prioritize this approach, as properties like state populations can often be determined by observing measurable quantities without requiring a complete description of the quantum state itself. These methods typically demand a circuit depth scaling of O(T²/ε) to achieve a specified accuracy, ε, and often require substantial sampling efforts to obtain reliable results.

However, the algorithms detailed in this work demonstrate a potential for reduced computational cost. By focusing on expectation value estimation, specifically, calculating Tr[Oρ(t)], the researchers achieved a circuit depth that scales logarithmically with desired accuracy, and either partial or complete independence from the parameters specifying the Lindbladian.

This improvement stems from a novel quantum algorithm that randomly generates circuits incorporating mid-circuit measurements and qubit resets, important for maintaining the characteristics of a completely positive trace-preserving (CPTP) map. The sampled circuits, derived from a set Wᵥ according to a probability distribution cᵥ/C, allow for efficient estimation of the desired physical properties.

The practical implications of this approach are further highlighted by the minimized ancilla requirements. Algorithm 1, detailed in the paper, requires setting the number of time segments and calculating a constant, C, using a method outlined in Appendix C. With these parameters defined, the number of samples needed, N, scales as O(C²|O|²/ε²), where |O| represents the spectral norm of the observable.

This scaling suggests an advantage over methods requiring larger sampling overheads, and the team notes that the simulation is not limited to observable estimation, extending to the estimation of non-linear functions of the final state as well. The algorithm’s efficiency is further supported by Theorem 2, which, combined with a truncated decomposition, completes the proof of its effectiveness.

More like this

Technology NewsResearchers compute atomic properties on quantum computer

Quantum AlgorithmsAlgorithm Reduces Quantum Circuit Complexity by Nearly Seventy Percent

Analog ComputingQuantum Algorithm Simulates Open Quantum Systems

Quantum SimulationQuantum Simulation: CPUs, GPUs & Frameworks