About Stochastic Resonance
Stochastic resonance is one of the most counterintuitive phenomena in nonlinear physics: adding noise to a noisy signal can make it clearer. In a bistable system with a double-well potential V(x) = −x²/2 + x&sup4;/4, a subthreshold periodic signal cannot on its own drive transitions between the two wells at x = ±1. However, the right level of Gaussian noise provides random kicks that help the system cross the energy barrier in synchrony with the signal, producing a peak in the output signal-to-noise ratio.
This simulation integrates the overdamped Langevin equation ˙x = x − x³ + A·f(t) + σ·ξ(t) using the Euler-Maruyama method. The top panel shows the instantaneous potential landscape and current position of the ‘particle’. The middle panel shows the time series, with the periodic signal overlaid. The bottom panel shows the SNR curve as a function of noise amplitude — the stochastic resonance peak is clearly visible. Adjust noise, signal amplitude, and frequency to explore the phenomenon.
Frequently Asked Questions
What is stochastic resonance?
Stochastic resonance is a phenomenon where an optimal level of noise enhances the response of a nonlinear system to a weak signal. Counter-intuitively, adding noise — usually considered harmful — helps the system detect or transmit a signal that would otherwise be too weak to cross a threshold. The signal-to-noise ratio (SNR) is not monotonically decreasing with noise; instead it peaks at an optimal noise level.
Why does noise help detect a weak signal?
In a bistable system with a double-well potential, a weak subthreshold signal cannot by itself push the system over the energy barrier between wells. Adding noise provides random ‘kicks’ that occasionally help the system cross the barrier. When the noise level is tuned so that transitions happen at roughly the signal frequency, the output is strongly correlated with the input — stochastic resonance. Too little noise: no transitions. Too much noise: random transitions swamp the signal. Optimal noise: transitions are signal-locked.
What is a bistable system?
A bistable system has two stable equilibrium states separated by an energy barrier. In this simulation, the double-well potential V(x) = −x²/2 + x&sup4;/4 has stable minima at x = ±1 with a barrier of height ΔV = 0.25 at x = 0. The system naturally sits in one well until noise or a signal pushes it over the barrier. Bistability appears in many contexts: neurons firing or not, ferromagnets with up/down domains, lasers, chemical reactors.
What is the Kramers escape rate?
The Kramers rate r_K = (√2/2π)·exp(−2ΔV/σ²) gives the thermally-activated escape rate from one well to the other in the double-well potential, where σ is noise amplitude and ΔV is the barrier height. At stochastic resonance, r_K matches the signal frequency, maximising synchronisation between noise-driven transitions and the signal oscillation.
Where does stochastic resonance occur in nature?
Stochastic resonance has been observed or proposed in: neurons and sensory systems (noise-enhanced nerve firing), the hearing of fish via the lateral line (hair cells with SR-like behaviour), human proprioception and balance, climate modelling (ice age cycles), electronic circuits, laser systems, SQUID magnetometers, and even social dynamics. It is particularly relevant in biological sensors operating near threshold.
How is the Langevin equation used in this simulation?
The overdamped Langevin equation dx/dt = −dV/dx + A·f(t) + σ·ξ(t) describes a particle in a potential V(x) driven by a periodic signal A·f(t) and Gaussian white noise σ·ξ(t). Here −dV/dx = x − x³ for the double-well potential. The Euler-Maruyama scheme x(t+dt) = x(t) + (x−x³+Af)·dt + σ·√dt·ξ integrates this equation numerically, where ξ is a standard normal random variable.
What is the signal-to-noise ratio (SNR) in stochastic resonance?
SNR is the ratio of signal power (power at the signal frequency in the output spectrum) to noise power density. In stochastic resonance, SNR is computed from the output x(t): the Fourier component at the signal frequency f₀ gives signal power, while the broadband floor gives noise density. SNR is plotted as a function of noise amplitude σ, showing a clear maximum — the stochastic resonance peak.
Can stochastic resonance be used in engineering?
Yes — practical applications include: dithering in audio and image quantisation (adding noise before quantising to improve digital accuracy), SR-based sensors and detectors that operate near threshold, vibrational energy harvesters, neural prosthetics that use optimal noise to improve sensory feedback, and analog-to-digital converters. Adding a small random noise before quantising can reduce systematic quantisation error.
What is the difference between noise-induced transitions and stochastic resonance?
Noise-induced transitions are simply thermal (or stochastic) crossings of a potential barrier — they occur at any noise level and are described by the Kramers rate. Stochastic resonance specifically refers to the phenomenon where a periodic signal’s SNR at the output is maximised at an intermediate noise level. SR requires both a weak periodic drive AND noise in a nonlinear (often threshold or bistable) system.
Does stochastic resonance work for any type of noise?
Stochastic resonance is most commonly studied with Gaussian white noise, which is the mathematically simplest case. However, the phenomenon is robust to the noise distribution — it also occurs with coloured noise (with correlations), Lévy noise, and even deterministic chaos acting as effective noise. The key requirement is that the noise should be broadband enough to drive transitions at the signal frequency.