Neutrinos · D02 · Time correlations

Reactor Antineutrino Coincidence

Acquire a twenty-second pulse record. Widen the prompt–delayed window, compare a time-shifted control, and turn off the reactor to test what remains.

Selected pairs $\text{—}$

Time-shift control $\text{—}$

Control-subtracted pairs $\text{—}$

True pairs · simulation only $\text{—}$

Accidental pairs · truth $\text{—}$

Ideal time-window acceptance $\text{—}$

Teaching simulation, not the Cowan–Reines dataset. Ideal pulse classification; 30 microsecond exponential capture lifetime; independent stationary singles; every forward pair is counted. No energy cuts, dead time, correlated neutron background or full detector response. Geometry is schematic; counts and plots use the same generated record.

Preparing a reproducible teaching record…

Read a sample of the record

Scientific reference ↗Read the worldwide history →2026 Nobel Prize in Physics →

Start with the essentials

Focus question: Does widening the window improve the evidence, or just the count?
One-sentence intuition: Keep the record fixed when comparing cuts. Change the acquisition only when testing a different source or background.

Core mathematical model

Capture-time acceptance

\epsilon(T)=\int_0^T\frac{e^{-t/\tau}}{\tau}\,dt=1-e^{-T/\tau}

Mean capture time is 30 microseconds by assumption. The window begins at zero. This analytic value is a reference, not an independently synthesized event count.

Independent accidental pairs

N_{\mathrm{acc}}\simeq R_pR_dT\,t_{\mathrm{exp}}

Both rates are total singles rates in the recorded prompt and delayed classes. All forward pairs are counted, so a pulse may participate in several pairs. The shifted control preserves class rates but destroys short physical correlations.

Common difficulties

A selection is also an experiment

Typical misconceptionA delayed pair must be a neutrino.

Better mental modelUnrelated pulses can meet the same rule. A control measurement and additional reaction-specific tests are essential.

Run the experiment

01
Keep the data fixed
Acquire once, then sweep the window from 2 to 200 microseconds.
What to observe: Accepted true pairs approach a plateau while accidental pairs keep increasing.
02
Ask the control sample
Use Crowded record. Compare time-shifted pairs with the simulated accidental truth.
What to observe: They fluctuate separately. The difference is an estimate, not an identity; a negative excess is allowed.
03
Turn the source off
Choose Reactor off, then acquire another record.
What to observe: Accidental coincidences remain. Reactor-off in this model removes signal only; real experiments must also investigate correlated backgrounds.