An arXiv preprint reports that a quantum-switch protocol could yield a tighter calculated precision bound than fixed-order Gaussian probes in selected settings. A reported gain came when the switch was used to prepare the probe and the resulting state had odd parity: at fixed encoding settings α_x=0.5, α_p=0.4 and r=0.2, the SLD-CRB fell below 1, the Gaussian minimum used in the comparison. That is a theoretical, model-based result, not a measured error.
The SLD-CRB is the study’s main quantum precision metric. It is calculated from the quantum Fisher information matrix, which summarizes information about unknown parameters, and symmetric logarithmic derivative operators. Lower SLD-CRB values indicate a tighter bound in the paper’s comparison; they are not, by themselves, observed estimator variances.
The central question was whether indefinite causal order, or ICO, could improve the joint estimation of displacement and squeezing compared with protocols that keep the operation order fixed. In the model, a quantum switch coherently combines displacement-then-squeezing with squeezing-then-displacement. The researchers varied the switch weight and examined ICO both while preparing the probe and while encoding the unknown parameters.
The Gaussian benchmark
Before testing ICO, the authors optimized pure Gaussian probes at fixed energy. They found a unique optimum: both displacement and squeezing point along the momentum quadrature, and the same optimal Gaussian SLD-CRB is obtained across the preparation orders. This provides the fixed-resource benchmark used for the later comparisons.
Energy produced a trade-off rather than a simple precision gain. The reported optimal Gaussian SLD-CRB increased monotonically with probe energy, while the largest modeled squeezing strength also increased, from about 1.32 at probe energy 5 to 1.67 at energy 10 and 2.01 at energy 20. At each fixed energy, the most precise probe divided its resources between squeezing and displacement.
Parity changed the outcome
At equal probe energy, when the momentum-directed setting β_p exceeded the position-directed setting β_x, the odd-parity ICO preparation state had a lower SLD-CRB than the optimal Gaussian probe across approximately 0.2 to 0.8 in switch weight p. The advantage was most prominent near p=0.5.
Even parity gave a different picture. Its SLD-CRB was lower than one fixed-order Gaussian baseline for interior values between p=0 and p=1, but it did not beat the other baseline at p=1. The even-parity result was also reported as roughly an order of magnitude higher than the odd-parity result, leaving no overall advantage for even-parity probe preparation.
The switch was less powerful during encoding
When ICO was used to encode the unknown parameters rather than prepare the probe, the odd-parity advantage was confined to a smaller range of switch weights. The difference from the optimal Gaussian probe was largest near p=0.5, while momentum squeezing produced the lowest reported SLD-CRB near p=0.75.
Even-parity ICO encoding improved on the fixed encoding order represented at p=0 for every tested positive switch weight. Its broadest reported range extended to about p=0.7 when the initial displacement favored position, but no tested configuration beat the displacement-then-squeezing order at p=1. The results therefore depended on whether ICO was used in preparation or encoding, as well as on parity and parameter settings.
Non-Gaussianity was not enough to explain the edge
The researchers also tested relative-entropy non-Gaussianity, a measure of how far a state departs from Gaussian form. For ICO probe preparation, the measure was neither necessary nor sufficient for an advantage over the optimal Gaussian probe. In one selected configuration, it approached zero over roughly 0.58<p<0.79 while the ICO advantage remained.
The same broad conclusion held for ICO encoding: non-Gaussianity alone did not determine whether the protocol had an advantage. In selected more-squeezed configurations, the reported advantage did increase with non-Gaussianity, but the measure did not provide a complete explanation of the precision difference.
A calculated result, not a universal sensor
The paper offers another possible clue by quantifying sloppiness from the determinant of the quantum Fisher information matrix, with three estimable parameters. It links the lower SLD-CRBs of odd-parity ICO states with reduced sloppiness associated with noncommuting encoding operations. That link is an interpretation of the model comparisons, rather than a direct measurement consequence.
The ICO SLD-CRBs were calculated in a truncated Fock space with state indices from n=0 through 100, and no truncation-convergence analysis is reported in the supplied analysis. The evidence is limited to pure single-mode states, the specified displacement and squeezing operations, and selected numerical regimes. It does not establish universal superiority across all settings or all non-Gaussian probes, nor does a lower calculated bound demonstrate an experimentally achieved error.
The document is an arXiv version-1 preprint, arXiv:2608.25870v1, posted on 26 August 2026.
Paper data and sources
Original title: Surpassing Gaussian optimality in multiparameter estimation with indefinite causal order
Authors: Sudipta Das, Rivu Gupta, Aditi Sen De, Himadri Shekhar Dhar
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text