Two identical photons enter the two input ports of a beamsplitter (BS) at nearly the same time. Classically you would expect coincident clicks (one photon in each output) 50% of the time. Quantum mechanically the two ways of getting a coincidence — both reflect, or both transmit — are indistinguishable alternatives that interfere destructively, so the photons always leave together, bunched in the same output port. This is the Hong–Ou–Mandel (HOM) effect (1987), the workhorse two-photon-interference test for photonic quantum computing and quantum networking.
Coincidence probability (BS reflectivity R, transmissivity T=1-R):
P_c(τ) = R² + T² − 2RT·g(τ)²
Photon overlap (Gaussian wavepackets, coherence time τc):
g(τ) = exp( −τ² / (2τc²) )
Perfect 50/50 BS, τ = 0 ⇒ g = 1 ⇒ P_c = 0 (the HOM dip bottoms out)
Large |τ| ⇒ g → 0 ⇒ P_c → R² + T² = 0.5 (classical limit)
- Delay τ — the relative arrival time of the two photons at the beamsplitter. τ = 0 means they arrive perfectly together.
- Coherence time τc — set by the photons' spectral bandwidth; it controls how narrow the HOM dip is. Shorter coherence time (broader bandwidth) gives a narrower dip.
- Reflectivity R — an unbalanced beamsplitter never lets the dip reach zero, because "both reflect" and "both transmit" no longer have equal amplitude to cancel.
- Photon stream — each trial samples a real outcome (coincidence vs. bunched) from P_c(τ) via Monte Carlo, so the measured coincidence rate converges to the theory curve — exactly how a real HOM experiment builds up statistics photon pair by photon pair.
Real-world relevance: HOM interference visibility is the standard benchmark for how indistinguishable single-photon sources are (quantum dots, SPDC pairs) — a deep, narrow dip is required for photonic entanglement swapping, linear-optical quantum computing (KLM scheme), and quantum repeaters.