Quantum Hamiltonian Learning in Open Systems

Quantum Hamiltonian learning in open systems addresses the problem of inferring the governing energy operator and environmental interactions of a quantum device that exchanges information with its surroundings. Unlike closed‐system protocols that assume isolated dynamics, open‐system learning must disentangle coherent Hamiltonian evolution from dissipative processes such as decoherence and noise. Recent advances combine tools from polynomial interpolation, shadow tomography, neural networks and Bayesian inference to reconstruct both the Hamiltonian and the associated Lindbladian superoperator. These methods exploit short‐time dynamics, steady‐state measurements or continuous monitoring to achieve scalable characterisation of many‐qubit devices. Progress in this field not only enhances our understanding of fundamental quantum processes but also underpins reliable calibration, error mitigation and optimised control strategies for near‐term quantum technologies.

Research from Nature Portfolio

One study introduced an efficient protocol for many‐qubit devices that estimates time derivatives of few‐qubit observables via polynomial interpolation. This approach simultaneously reconstructs the underlying Hamiltonian and Markovian noise with exponentially relaxed time‐resolution requirements and quadratically reduced sample complexity. The implementation relies solely on product‐state preparation and single‐qubit measurements, and includes improvements to shadow tomography for quantum channels. A second work demonstrated a general method for estimating time‐dependent Hamiltonians of a single qubit by measuring its evolution under a tunable offset term. Applied to a trapped ion traversing a laser beam, the protocol recovered both the spatial intensity profile and the ion’s velocity, illustrating a broadly applicable strategy for Hamiltonian estimation in systems with non‐commuting terms.

Research from all publishers

A robust and efficient learning scheme for sparse Hamiltonians on the Pauli basis requires only local operations and short‐time dynamics, with no assumptions on eigenstates or thermal states. This method scales favourably with system size and remains resilient to state‐preparation and measurement errors, circuit noise and shot noise. Another approach targets open‐system dynamics by reconstructing the local Lindbladian from steady‐state measurements. For finite‐range interactions, the recovery protocol uses only observables within each spatial domain and shows that environmental couplings can facilitate Hamiltonian reconstruction. A complementary direction employs a recurrent neural network trained on continuous measurement records under unitary evolution, decoherence and monitoring. The network infers the Hamiltonian, measurement operators and physical parameters in real time and can perform tomography of unknown initial states, offering new routes for noise characterisation and feedback control.

Quantum Hamiltonian Learning in Open Systems publication trend

The graph below shows the total number of articles in quantum hamiltonian learning in open systems across all publications each year (not limited to Nature Index journals).

Technical terms

Hamiltonian: Operator describing the total energy and governing the time evolution of a quantum system.

Open quantum system: A system that exchanges energy or information with an external environment, leading to decoherence and dissipation.

Lindbladian: Generator of Markovian quantum dynamics under the Lindblad master equation, describing irreversible processes.

Markovian noise: Memoryless environmental disturbance characterised by exponential decay of system correlations.

Steady state: The long‐time equilibrium state of an open quantum system under continuous interaction with its environment.

References

1. Efficient and robust estimation of many-qubit Hamiltonians. Nature Communications (2024).
2. Estimation of a general time-dependent Hamiltonian for a single qubit. Nature Communications (2016).
3. Robust and Efficient Hamiltonian Learning. Quantum (2023).
4. Learning the dynamics of open quantum systems from their steady states. New Journal of Physics (2020).
5. Using a Recurrent Neural Network to Reconstruct Quantum Dynamics of a Superconducting Qubit from Physical Observations. Physical Review X (2020).