⚛️ quantum physics

Variational Real-Time Dynamics on Reduced Operator Manifolds

The paper introduces operator-projected variational quantum real-time evolution (OVQRTE), a resource-efficient algorithm that enforces Ehrenfest equations on a selected operator set to accurately simulate correlated quantum systems on near-term hardware, as demonstrated through benchmarks on the Anderson impurity model and integration with quantum-selected configuration interaction.

Original authors: Aeishah Ameera Anuar, PV Sriluckshmy, Riccardo Rossi, Fedor Simkovic

Published 2026-09-15

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Aeishah Ameera Anuar, PV Sriluckshmy, Riccardo Rossi, Fedor Simkovic

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the microscopic world of atoms and electrons, matter does not behave like the solid objects we touch every day. Instead, particles exist in a state of constant, interconnected fluctuation, where the behavior of one particle is instantly linked to the behavior of others, no matter how far apart they are. Scientists call this "correlation," and it is the key to understanding how materials conduct electricity, become superconductors, or change color. To predict how these systems change over time—how they evolve from one state to another—researchers need to simulate their real-time dynamics. This is a notoriously difficult task. Classical computers, which power our smartphones and supercomputers, struggle immensely with these calculations because the amount of information required to describe a correlated system grows so fast that it quickly overwhelms even the most powerful machines. Quantum computers, which use the same strange laws of physics to process information, offer a potential solution, but current devices are still small and prone to errors, making them unreliable for long, complex simulations.

A team of researchers has developed a new method to make these simulations possible on today's imperfect quantum hardware. They call their approach operator-projected variational quantum real-time evolution. Rather than trying to track the entire, overwhelming state of a system at once, which is too much for current machines to handle, this method focuses only on a specific, manageable set of observable properties. Imagine trying to follow the movement of a massive, swirling crowd by tracking only the positions of a few key people and the general flow of the group around them, rather than trying to record every single person's step. By concentrating on these selected properties, the researchers can update the quantum computer's settings step-by-step, ensuring that the simulation follows the correct physical laws for those specific properties without needing to calculate the impossible details of the whole system. This strategy drastically reduces the number of measurements required, making the process fast enough to run on existing devices while still capturing the essential physics.

The researchers tested this new method first on a theoretical model of a magnetic chain, known as the Heisenberg model, using a computer simulator. They found that their approach required significantly fewer measurements than previous methods to achieve the same level of accuracy. In fact, for larger systems, the new method needed up to one hundred times fewer measurement steps. This efficiency is crucial because every measurement on a quantum computer takes time and introduces a small amount of noise; reducing the number of steps means the simulation stays accurate for longer. The team then moved from simulation to reality, running their algorithm on a 24-qubit superconducting quantum processor called IQM Emerald. They used it to simulate a complex model of an impurity atom interacting with a surrounding sea of electrons, a scenario known as the Anderson impurity model. Even with the noise inherent in current hardware, the method successfully tracked the system's evolution, demonstrating that it could handle the complexity of real-world materials.

Beyond simply watching the system change, the researchers showed that this method could be used to find the lowest energy state of a material, which is often the most important property for understanding its stability and behavior. They combined their time-evolution technique with a strategy called quantum-selected configuration interaction. In this process, the quantum computer generates a series of snapshots of the system as it evolves over time. These snapshots are then used to build a compact, simplified map of the most important ways the electrons can arrange themselves. By analyzing this map on a classical computer, the team was able to calculate the ground-state energy and the density of states with high precision. To handle the errors introduced by the noisy hardware, they applied a recovery process that corrected for violations of physical laws, such as the conservation of particle number, ensuring the final results remained physically valid.

The results suggest that this approach offers a practical path forward for studying complex materials on near-term quantum computers. By avoiding the need to calculate the full, overwhelming state of the system and instead focusing on a curated set of physical properties, the method sidesteps the bottlenecks that have previously limited quantum simulations. The researchers demonstrated that even with the limitations of current hardware, it is possible to generate useful data about how correlated quantum systems behave. This work establishes a new framework where the trade-off between accuracy and computational cost can be managed systematically, allowing scientists to extract meaningful insights from noisy quantum devices. As these machines continue to improve, this technique could become a standard tool for investigating the properties of new materials, potentially leading to discoveries in superconductivity and quantum chemistry that are currently out of reach.

Technical Summary: Variational Real-Time Dynamics on Reduced Operator Manifolds

Problem Statement
Accurate real-time simulation of correlated quantum many-body systems remains a significant challenge for both classical algorithms and near-term quantum hardware. Classical methods, such as tensor networks and neural quantum states, face bottlenecks related to entanglement growth or stochastic sampling costs. On quantum hardware, standard approaches like Trotterization require circuit depths that scale unfavorably with simulation time and accuracy. While variational quantum real-time evolution (VQRTE) offers a compromise by restricting dynamics to a parameterized circuit ansatz, standard implementations based on the McLachlan variational principle require the estimation of the quantum geometric tensor (QGT). This process incurs substantial measurement overhead (scaling quadratically with parameters) and is sensitive to noise and ill-conditioning. Furthermore, existing methods often struggle to balance the trade-off between simulation accuracy and the measurement resources required.

Methodology: Operator-Projected Variational Quantum Real-Time Evolution (OVQRTE)
The authors introduce Operator-Projected Variational Quantum Real-Time Evolution (OVQRTE), a method that updates a parameterized quantum circuit by enforcing the Ehrenfest equations for a selected, polynomially sized subset of observables, rather than projecting the full state evolution onto the ansatz tangent space.

Core Principle: Instead of solving for the full state dynamics, OVQRTE tracks a set of expectation values $S = \{\hat{O}_\alpha\}$ (e.g., local Pauli strings or reduced density matrix operators). The parameter update $\dot{\theta}$ is determined by minimizing the residual between the variational rate of change of these observables and their exact Ehrenfest rates:
$\dot{\theta} = \arg \min_{v} \sum_{\alpha} \left| \dot{x}^{var}_{\alpha} - i\langle [\hat{H}, \hat{O}_\alpha] \rangle \right|^2$
This yields a linear system $A\dot{\theta} = C$ , where $A$ is the Gram matrix of the observable Jacobian and $C$ contains the commutator expectation values.
Resource Efficiency: Crucially, OVQRTE avoids the explicit construction of the QGT. The matrix $A$ is assembled entirely from expectation values and their derivatives, which can be measured simultaneously using qubit-wise commuting (QWC) groups. This reduces the measurement cost from quadratic in the number of parameters (as in QGT-based methods) to linear or even independent of the parameter count when using Simultaneous Perturbation Stochastic Approximation (SPSA).
Operator Pool Reduction: To further mitigate hardware noise and circuit depth, the authors propose a "Qubit-Excitation-Based" (QEB) variant. This approach simplifies fermionic-to-qubit mapped operators (e.g., in Jordan-Wigner transformations) by replacing non-local parity strings (Pauli-Z chains) with identity operators, retaining only the endpoint operators. This significantly reduces the locality and Pauli string weight, making measurements more robust against noise.
Ground-State Extraction via QSCI: The paper demonstrates that OVQRTE can serve as a generator for Quantum-Selected Configuration Interaction (QSCI). By sampling computational-basis states from the time-evolved trajectory, a compact, physically relevant subspace is constructed. The ground state is then approximated by diagonalizing the Hamiltonian within this subspace. To handle noise-induced symmetry violations on hardware, the authors employ a self-consistent configuration recovery (S-CORE) procedure and, for the half-filled Anderson impurity model, a particle-hole (PH) symmetry completion.

Key Results

Heisenberg Model Benchmarks: Simulations of a 1D Heisenberg model demonstrate that OVQRTE substantially reduces the measurement cost compared to standard VQRTE, requiring one to two orders of magnitude fewer circuits per time step. In state-vector simulations, OVQRTE achieves numerical precision comparable to VQRTE. Under finite-shot sampling (simulating hardware noise), OVQRTE exhibits superior robustness, maintaining lower state infidelity and more accurate magnetization dynamics than VQRTE for the same shot budget.
Anderson Impurity Model (AIM) on Hardware: The algorithm was benchmarked on the IQM Emerald superconducting processor using up to 24 qubits to simulate a single-band Anderson Impurity Model. The study utilized the Local Unitary Cluster Jastrow (LUCJ) ansatz mapped to the device's native connectivity without SWAP gates. To avoid uncontrolled feedback instabilities caused by the accumulation of hardware and sampling noise across repeated calls, the authors focused on benchmarking OVQRTE+QSCI on converged AIM instances obtained from classical gGut solutions, rather than running the full self-consistent loop on hardware.
QSCI Integration: The OVQRTE-generated states were used to construct a subspace for QSCI. The resulting ground-state energy estimates and density of states, calculated for these converged AIM instances, showed high accuracy when compared to exact classical references. The method successfully handled the extraction of ground-state properties even when the full real-time trajectory was not reproduced with high fidelity, provided the evolved states populated an informative low-energy subspace.

Significance and Claims
The paper establishes OVQRTE as a promising approach for investigating correlated condensed-matter systems on near-term quantum hardware. Its primary significance lies in:

Systematic Trade-off: It provides a controllable knob (the size and choice of the operator set $S$ ) to trade accuracy against measurement cost, allowing for resource-efficient simulations.
Noise Resilience: By avoiding the QGT and utilizing reduced operator sets (QEB), the method is less sensitive to the noise and ill-conditioning that plague standard variational real-time evolution.
Utility Beyond Dynamics: The work highlights that accurate reproduction of the entire dynamical trajectory is not strictly necessary for downstream applications. OVQRTE effectively generates informative subspaces for ground-state estimation (via QSCI), making it a viable tool for solving impurity problems within embedding theories like gGut on current hardware.

The authors conclude that OVQRTE offers a practical pathway for simulating correlated systems by leveraging reduced operator manifolds to bypass the resource bottlenecks of existing variational methods.

More like this