Jarzynski Equality: Free Energy from a Dragged Brownian Particle

The same stochastic model as a bench: a valley slides along a rail and 14 beads jiggle inside it. Drag to orbit.

Switch to 2D
Controls

Presets

Parameters

Quality

Camera

Model checks

The science

Setup. A Brownian bead sits in a harmonic trap U = k(x - λ)²/2 in a heat bath at temperature T. The trap centre is dragged at constant speed, λ(t) = v t, from 0 to vT. The motion is overdamped (Langevin): γ dx/dt = -k(x - λ) + √(2γkBT) ξ(t). Lengths are measured in σ = √(kBT/k), times in τ = γ/k, energies in kBT.

Work. The work done on the bead by the moving trap is W = ∫ (∂U/∂λ) dλ = -v ∫ (x - λ) dt. It differs from trial to trial. Because a translation of the trap does not change the partition function, the free-energy difference is ΔF = 0.

Jarzynski equality. ⟨e-W/kBT⟩ = e-ΔF/kBT, even though the average work ⟨W⟩ ≥ ΔF (second law). The average of e-W is dominated by rare trials with small or negative work; this is why the estimator converges slowly when dissipation is large.

Exact solution used for the checks. For this linear system W is Gaussian with mean μ = v²(T - 1 + e-T) (in kBT) and variance 2μ (fluctuation-dissipation). Then ⟨e-W⟩ = e-μ + μ = 1 and the variance of e-W is e2μ - 1; that variance is used for the error bar in the first check.

Simplifications. One dimension, overdamped dynamics, constant friction, no inertia and no hydrodynamic memory. The numerical scheme is the exact Ornstein-Uhlenbeck transition per time step with a trapezoid rule for the work. The slider range keeps the dissipation low enough (μ ≤ 3.2 kBT) that the exponential average can be tested statistically within a second. Real optical-tweezer and RNA-pulling experiments (Liphardt 2002, Collin 2005) use the same estimator with non-Gaussian work distributions.

Loading model...
Loading Blender model...
-
-
Made with 🟠 Blender · Three.js