Fig. 3

Demonstration of the validity of Eq. (5). All plots show the normalized change in the average photon number \(\delta \tilde n\) with N = 5 × 10⁴. a A delay in the probe controls the temporal overlap with the system inside the PCF and thus tunes the interaction strength (here with ε = 0.1). This is performed for several magnitudes of modulation, and the results trace out Gaussian shape for non-zero modulation amplitude. However, in the absence of modulation, the standard deviation due to the shot noise \({\mathrm{\Delta }}n = \sqrt N \simeq 10^2\) is too small to observe the effect of the interaction. b The standard deviation Δn is changed by controlling the magnitude of the modulation in the intensity of the probe, with ε = 0.1. c Tuning the post-selection parameter ε, while Δn = 1.4 × 10⁴. The results in both (b, and c) are fitted to Eq. 5, and an estimate of g, with an error due to the fitting quality, is shown in each panel. The error bars are shown as the uncertainty in \(\delta \tilde n\), which is written as \(\sigma {\mathrm{/}}\sqrt \nu \). Here, σ is the standard deviation of measured \(\delta \tilde n\) and ν is number of recorded probe pulses by FHO