Adaptive boson sampling architectures could be used as a tool to enable the realization and quantification of optical nonlinearities beyond the capabilities of traditional linear-optical systems. These findings were recently published in Nature Photonics.
Study: Beyond quantum linear optics with adaptive boson sampling. Image Credit: Redmixx/Shutterstock.com
Adaptive Boson Sampling Foundations
Photon-based quantum computing traditionally relies on linear-optical elements, such as beam splitters and phase shifters, to manipulate photonic states. While boson sampling (BS) models implemented with linear optics present computational tasks that are thought to be challenging to simulate classically, their scope is fundamentally limited.
No time now? Download a free PDF copy to finish this news piece later!
Linear optics alone cannot generate all photonic states or perform universal quantum computation due to the absence of effective optical nonlinearities. Incorporating nonlinear interactions or measurement-based adaptivity introduces the needed nonlinear dynamics but poses significant technical challenges.
Recent research has explored intermediate regimes in which limited nonlinearities are induced via adaptive measurement-driven feedback instead of direct nonlinear optical media, offering a potential pathway to surpass linear-optical constraints.
The adaptive boson sampling (ABS) paradigm exemplifies this approach by reconfiguring interferometers conditionally on measurement outcomes, potentially generating photonic states unreachable by purely linear means.
Photonic Platform Implementation
The study focuses on experimentally accessing and quantifying nonlinear optical dynamics through ABS architectures. The core framework considers an input multi-photon Fock state injected into an m-mode interferometer, which undergoes a cascade of adaptive unitaries conditioned on intermediate photon-detection outcomes.
Mathematically, the total adaptive interferometric operation is described by a sequence of outcome-dependent unitary transformations concatenated with an initial fixed unitary matrix. Intermediate projective measurements reduce photon number and modes dynamically, generating output states that depend on the measurement trajectory.
To detect departures from linear optics, the authors formalize the concept of “passive separability”, a property of quantum states, which, if violated, witnesses genuine optical nonlinearities beyond linear optics. Starting from mode-separable Fock-state inputs, linear optics can only generate passively separable states, so demonstrating the lack of such separability acts as a nonlinearity witness.
The approach analyzes measured output density matrices in the Fock basis, leveraging Lie algebraic invariants related to unitary transformations within the linear-optical framework. The experimental protocols include both real-time adaptive feed-forward implementations and post-selection-based emulations of adaptivity.
The experiments employ QOLOSSUS-PRO, a state-of-the-art photonic platform integrating a demultiplexed quantum-dot single-photon source into simultaneous multi-photon inputs, coupled to reconfigurable integrated photonic processors with either eight or 12 spatial modes.
For real-time adaptivity, photons measured in intermediate modes trigger conditional phase shifts applied via electro-optic modulators on remaining photons, encoding adaptive unitaries on two-mode outputs.
Post-selection experiments extend these methods to configurations involving up to four input photons distributed over larger mode sets, where adaptivity is simulated by sampling over several fixed interferometric settings and recombining data accordingly.
Probing Nonlinear Photonic Dynamics
The work experimentally demonstrates that measurement-induced adaptivity in ABS setups enables the generation of photonic output states exhibiting nonlinear optical dynamics outside the reach of any purely linear-optical evolution.
In the two-photon, two-mode experiment, real-time feed-forward, adaptively controlled phase shifts conditioned on intermediate measurement outcomes realize output states that violate passive separability criteria. This observation acts as a direct signature of genuine nonlinearity emerging through the adaptive process, verified by reconstruction of the density matrix and evaluation of Lie algebraic invariants.
Scaling up to more photons and modes, ABS protocols emulated through post-selection produce complex output distributions with measured properties outside the bounds expected from passive linear optics alone. By comparing experimental data with theoretical bounds derived from group-theoretic and geometric constraints on linear-optical transformations, the authors show that ABS-generated states lie beyond the accessible linear-optical state space.
This demonstration highlights adaptive boson sampling as a versatile intermediate paradigm bridging standard linear-optics BS and fully universal photonic quantum computation that requires strong optical nonlinearities. Importantly, the adaptive approach leverages measurement outcomes to implement effective nonlinear transformations without the need for challenging deterministic optical nonlinearities in media.
The combination of integrated photonic processors, high-purity multi-photon sources, and rapid feed-forward control mechanisms forms a practical toolbox for exploring new quantum optical regimes.
Implications for Quantum Photonics
This study establishes adaptive boson sampling as a powerful approach to surpass the intrinsic bounds of linear optics by harnessing measurement-based adaptive feedback to induce nonlinear optical dynamics in photonic circuits.
By developing and applying rigorous nonlinear witnesses grounded in the properties of linear-optical transformations, the authors provide experimental evidence that ABS architectures generate photonic states inaccessible by any linear-optical unitary evolution.
The developed methods not only deepen fundamental understanding of the frontier between linear and nonlinear photonic dynamics but also provide a foundation for a feasible pathway toward scalable, programmable optical quantum devices that exploit limited nonlinearities for enhanced computational power.
This positions ABS as a promising platform for investigating new quantum optical phenomena in the intermediate regime between passive linear optics and fully universal photonic quantum computing.
Journal Reference
Rodari G., Francalanci T., et al. (2026). Beyond quantum linear optics with adaptive boson sampling. Nature Photonics. https://www.nature.com/articles/s41566-026-01959-3.