arXiv:2610.18495
Steering the Fidelity in State Preparation on Noisy Bosonic Systems with Risk-Seeking Optimization
Riccardo Piantoni, Riccardo Pellini, Maurizio Ferrari Dacrema, Paolo Cremonesi
Abstract
In order to execute quantum algorithms on bosonic quantum computers, it is necessary to encode their inputs into a desired target quantum state, with a process called state preparation. In real settings, state preparation methods operate in stochastic scenarios, where the prepared state is represented as a mixed state, that is a probabilistic ensemble of pure states. Therefore, existing state preparation methods optimize a variational ansatz by averaging over all possible sequences of photon loss events contained in the prepared ensemble, yielding an output state which approximates the desired target with a certain average fidelity. Such an approach, however, does not take into account directly the distribution of fidelities across states in the ensemble. In this work, we aim to maximize the fidelity of only a subset of pure states sampled from the prepared ensemble, at the cost of reducing the average fidelity overall. This trade-off is relevant when the goal is to obtain high-fidelity states with a certain probability, rather than maximizing the average fidelity of the prepared ensemble. We borrow and adapt objective functions from risk-sensitive optimization, a framework that optimizes specific characteristics of a distribution rather than only its expectation, and that has been successfully applied in Reinforcement Learning. In the considered experimental settings, our results show that the infidelities of some pure states in the prepared ensemble can be reduced by up to four orders of magnitude when using risk-seeking objectives, compared to those achieved using the standard optimization of the average fidelity. Furthermore, we show that it is possible to control the trade-off between the probability of obtaining high-fidelity states and the lowest infidelity threshold reached, providing the flexibility to select the operating point that best suits a certain application.
AI Summary
Standard noisy-state preparation maximizes average fidelity, but does not directly control the fidelity distribution across noise trajectories. This paper asks whether risk-seeking optimization can favor high-fidelity trajectories at the expense of ensemble-average performance, using bosonic SNAP–displacement state preparation. Two objectives apply entropic risk to trajectory fidelity $F$ or negative log-infidelity $-\log(1-F)$, with a positive risk coefficient $\beta$ controlling the emphasis on favorable outcomes. Optimization uses Monte Carlo photon-counting trajectories and stochastic-computation-graph gradients that account for parameter-dependent trajectory probabilities. Classical simulations consider one eight-level qudit, 64 additional leakage-buffer levels, four SNAP–displacement blocks, cavity lifetime $T_1=1\,\mathrm{ms}$, and a Haar-random target. Each setting uses 100 optimization runs of 30,000 steps, with ten trajectories per step and 10,000 evaluation trajectories per run. Average-fidelity optimization achieves mean fidelity $0.824$, but reaches no trajectory-infidelity threshold below $10^{-2}$. Risk-seeking objectives reach $10^{-6}$, a four-order-of-magnitude reduction in the lowest threshold reached. At this threshold, log-infidelity risk with $\beta=1$ succeeds in 12% of optimization runs; successful runs produce qualifying trajectories with mean probability $0.680$, while mean ensemble fidelity across all runs is $0.748$. Reported trajectory probabilities are conditional on successful optimization. Practical use requires suitable measurement or heralding to associate trajectories with outcomes. The logarithmic objective introduces finite-sample gradient bias. Generalization across targets, dimensions, circuit depths, and noise models remains untested; saturation and conditional-probability trends require further investigation.
the authors used GPT-5.6 Sol, GPT-5.6 Luna and Grammarly in order to paraphrase and reword, as well as grammar and spelling check.
Central claim may not hold: The abstract's probability of obtaining high-fidelity states has an operational interpretation only with access to an appropriate photon-counting measurement or heralding record, which the paper assumes conditionally but does not implement. The reported four-orders improvement compares the lowest successful thresholds on a decade-spaced grid, rather than measured minimum infidelities whose ratio is established to be 10^4.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 12 pages, 1 table, accepted at International Workshop on AI for Quantum and Quantum for AI (AIQxQIA 2026), co-located with AIxIA 2026, Perugia, Italy