Figure 6: Measured violations.

Observed violations as a function of the path difference |Δx|=c|Δτ| between the two photons. Blue shaded regions in the plots correspond to the cases where the hypothesis of distinguishable particles can be ruled out. Red regions correspond to the cases when both the hypotheses of distinguishable particles and mean field state can be ruled out, and true two-particle interference is present. (a) Data for the four-mode interferometer. Blue points: input (1,3). Red points: input (2,4). Blue solid line: theoretical prediction for input (1,3). Red solid line: theoretical prediction for input (2,4). Black dashed line: theoretical prediction for a Fourier matrix. (b) Data for the eight-mode interferometer. Blue points: input (1,5). Red points: input (2,6). Green points: input (3,7). Magenta points: input (4,8). Coloured solid lines: corresponding theoretical predictions for the different inputs. Black dashed line: theoretical prediction for a Fourier matrix. Tables: violations at Δx=0 and discrepancies (in sigmas) with the expected values for distinguishable particles () and MFs (), for the cyclic inputs of the two interferometers. are calculated following formula (5), while expected values for the other two cases are and . Error bars in all experimental quantities are due to the Poissonian statistics of measured events. All theoretical predictions in solid lines are calculated from the reconstructed unitaries, obtained from different sets of experimental data to ensure statistical independence. See Supplementary Note 2 for the modelling of imperfect state preparation.