The Signal
Paderborn sampler compares Gaussian and non-Gaussian inputs
A photonic platform switches among Gaussian and non-Gaussian sampling configurations during one measurement run. Its researchers report stronger nonclassicality signals for heralded-photon inputs as brightness rises, while the Gaussian configuration’s signal weakens beyond a point.

One run, three sampling regimes
A photonic sampler can switch among Gaussian, heralded-photon and thermal inputs within one measurement run, allowing researchers to compare the output correlations of the different configurations. The Paderborn Quantum Sampler (PaQS), described in a paper published in PRX Quantum on November 17, supports sampling with eight input states in a 12-mode interferometer. The platform’s researchers say the configurations can be compared under otherwise identical conditions, with only the input states changing.
The study compares Gaussian boson sampling (GBS), scattershot boson sampling (SBS) and thermal boson sampling (TBS). GBS uses single-mode squeezed-vacuum states; SBS uses photon-number measurements to herald Fock-state inputs, while TBS uses thermal inputs. The integrated source and programmable interferometer let the system switch configurations during a measurement run, rather than requiring separate runs for each type.
Testing for nonclassical correlations
The researchers used a semi-device-independent test based on moments of photon-number correlations. The method does not require advance knowledge of the input states or the interferometer, and tests whether the measured data can be explained by classical resources. The paper emphasizes that a nonclassicality signal is a necessary, but not sufficient, condition for establishing quantum computational complexity.
In one comparison, GBS data at a mean photon number of 0.569 had negative minimum eigenvalues in more than 80% of recorded intervals, which the authors interpret as evidence of nonclassicality. At a mean photon number of 2.152, the data showed quantum correlations in less than 2% of the intervals. Each interval represented one second of data, with one million samples.
Brightness changes the measured signal
The paper reports a different pattern for SBS: its measured nonclassicality increased as input brightness rose. At the highest recorded mean photon number, SBS violated the classicality bound by more than 36 standard deviations. The authors say the GBS signal did not continue to strengthen with brightness: its largest reported violation exceeded 26 standard deviations at an average photon number of 0.172, while at 0.569 the violation was more than 14 standard deviations.
At a mean photon number of 2.152, the measured two-mode correlations in GBS showed no quantumness for more than 90% of input-phase values. The researchers say the cause remains unresolved. They propose that, as brightness rises, correlations may shift into higher-order measurements, but say testing this would require photon-number resolution greater than four, beyond the setup’s reliable analysis.
Scope of the result
The full measurement runs used eight modes of the installed 12-mode interferometer. The detection method distinguished up to three photons per pulse, and the authors say that cutoff was sufficient to verify nonclassicality in their SBS data using the proposed measure. The paper reports that the findings compare measured nonclassicality across configurations; it does not claim that the experiment demonstrated quantum computational advantage.
Sources
Related stories
- IonQ says DARPA will test its quantum roadmap on hardware
- Trapped-Ion Benchmark Sets Conditions for Multiqubit Gates to Win
- Nature paper reports synchronized microwave-optical superradiant combs
- PRX Quantum paper demonstrates phase-controlled photon routing
- Study models qubit sidebands as a fingerprint of antibunched light