Brownian Motion & Random Walk

Explore random processes from physics to finance, Einstein's diffusion theory, and geometric Brownian motion

Time: 0.00
Steps: 0

Parameters

1.0
20
0.0
0.01

Random Walk Type

Initial Condition

Actions

Display Options

Presets

Animation Speed

30 FPS

Average Position ⟨x⟩

0.0000

Mean Square Displacement ⟨x²⟩

0.0000

Diffusion Coefficient D

0.0000
Theory: 1.0000

Variance σ²

0.0000

Position Distribution

Empirical Gaussian Theory

Mean Square Displacement vs Time

Simulated Theory: ⟨x²⟩ = 2Dt
Regression Slope: -

Mathematical Analysis

Displacement Distribution

P(x,t) = (4πDt)^(-1/2) × exp(-x²/4Dt)

Gaussian distribution with variance 2Dt

Einstein Relation

⟨x²⟩ = 2Dt

Mean square displacement grows linearly with time

Geometric Brownian Motion

dS = μS dt + σS dW_t

S_t = S_0 × exp((μ - σ²/2)t + σW_t)

Stock price model (Black-Scholes)

Simple Random Walk

S_{n+1} = S_n + ξ_n, ξ_n ∈ {-1, +1}

Var(S_n) = n

Discrete version converges to Brownian motion

Wiener Process Properties

  • W₀ = 0 (starts at origin)
  • Independent increments
  • Wₜ - Wₛ ~ N(0, t-s)
  • Continuous paths (almost surely)

Scaling Law

x ~ √t (diffusion scaling)

To double displacement, need 4× time

From Pollen to Stock Prices

1827

Robert Brown's Discovery

Scottish botanist Robert Brown observed erratic motion of pollen grains in water under a microscope. He first suspected a biological life force, but later found the same motion in inorganic particles.

1900

Bachelier's Thesis

Louis Bachelier built a theory of stock price fluctuations using random walks, five years before Einstein. His work Théorie de la Spéculation laid the foundation for mathematical finance.

1905

Einstein's Theory

Albert Einstein explained Brownian motion through kinetic theory, derived the diffusion equation, and obtained the relation ⟨x²⟩ = 2Dt. This became key evidence for atomic theory.

⟨x²⟩ = 2Dt = (k_B T / 3πηr) × t

1908

Perrin's Experiments

Jean Perrin carried out precise experiments that verified Einstein's predictions and won the 1926 Nobel Prize. His results convinced many remaining skeptics of the existence of atoms.

1923

Wiener's Mathematical Formalization

Norbert Wiener gave Brownian motion a rigorous mathematical basis by constructing Wiener measure and proving path properties. This work became central to stochastic calculus.

1973

Black-Scholes Formula

Fisher Black, Myron Scholes, and Robert Merton developed the option pricing formula using geometric Brownian motion and transformed modern financial markets.

C = S·N(d₁) - K·e^(-rT)·N(d₂)

Mathematical Foundation

1. Simple Random Walk (Discrete)

The simplest random process: at each step, move ±1 with equal probability.

S_0 = 0, S_{n+1} = S_n + ξ_n

P(ξ_n = +1) = P(ξ_n = -1) = 0.5

E[S_n] = 0, Var(S_n) = n

After n steps, the typical displacement scales like √n.

2. Continuum Limit (Scaling)

Take many small steps with step size ε and time step δ while keeping ε²/δ = 2D constant.

lim_{n→∞} S_{[nt]} / √n → W_t (Wiener process)

By the Central Limit Theorem, the limit becomes Gaussian.

3. Brownian Motion (Wiener Process)

A continuous-time stochastic process with Gaussian increments.

W_0 = 0

W_t - W_s ~ N(0, t-s) for t > s

Independent increments: W_t - W_s ⊥ W_s

Continuous paths (almost surely)

Its paths are continuous but nowhere differentiable.

4. Diffusion Equation (Fokker-Planck)

The probability density evolves according to the heat equation.

∂P/∂t = D ∂²P/∂x²

P(x,t) = (4πDt)^(-1/2) × exp(-x²/4Dt)

The solution is Gaussian with variance 2Dt.

5. Itô Calculus (Stochastic Integration)

Processes driven by Brownian noise require a new calculus.

dX_t = μ dt + σ dW_t

Itô Lemma: df(X,t) = f_x dX + (1/2)f_xx σ² dt + f_t dt

(dW_t)² = dt (quadratic variation)

The second-order term matters because (dWₜ)² = dt.

6. Geometric Brownian Motion

A stock-price model with positive values and log-normal distribution.

dS/S = μ dt + σ dW_t

S_t = S_0 × exp((μ - σ²/2)t + σW_t)

E[S_t] = S_0 e^{μt}

Var(S_t) = S_0² e^{2μt}(e^{σ²t} - 1)

log(Sₜ/S₀) follows a normal distribution.

Financial Applications

Why Geometric Brownian Motion for Stocks?

  • Prices remain positive because of the exponential form.
  • Returns, rather than prices, are modeled as additive and independent.
  • A log-normal price distribution is often a useful approximation.
  • The model stays simple enough for analytical solutions.

Drift μ vs. Volatility σ

μ is the expected return or trend, while σ measures randomness and risk.

High σ means larger price swings and a higher risk premium.

High μ means a stronger upward trend and higher expected return.

Black-Scholes Option Pricing

A European call option gives the right to buy a stock at strike K at time T.

C = S·N(d₁) - K·e^(-rT)·N(d₂)

d₁ = [ln(S/K) + (r + σ²/2)T] / (σ√T)

d₂ = d₁ - σ√T

N(·) is the standard normal CDF. The key insight is to build a risk-free hedge.

Risk-Neutral Valuation

In complete markets, price equals the discounted expected payoff under the risk-neutral measure.

Price = e^(-rT) × E_Q[Payoff]

Replace the real drift μ with the risk-free rate r.

The Greeks (Risk Measures)

  • Δ (Delta): ∂Price/∂S (hedging ratio)
  • Γ (Gamma): ∂²Price/∂S² (convexity)
  • ν (Vega): ∂Price/∂σ (volatility sensitivity)
  • Θ (Theta): ∂Price/∂T (time decay)
  • ρ (Rho): ∂Price/∂r (interest rate sensitivity)

Monte Carlo Simulation

When no closed-form solution is available, simulate many random paths.

S_{i+1} = S_i × exp((μ - σ²/2)dt + σ√dt·Z)

where Z ~ N(0,1)

This visualization uses Monte Carlo path simulation.

Virtual Laboratory

Experiment 1: Verify Einstein Relation

Test whether ⟨x²⟩ = 2Dt holds in the simulation.

  1. Set D = 1.0 and dt = 0.01.
  2. Start with 50 particles at the origin.
  3. Run until T = 10.0 (1000 steps).
  4. Check the regression slope in the MSD plot.
  5. Expected slope: 2D = 2.0.

Experiment 2: Central Limit Theorem

Watch a Gaussian distribution emerge from simple ±1 steps.

  1. Choose Simple Random Walk.
  2. Use 100 particles starting at the origin.
  3. Observe the histogram after 100, 500, and 1000 steps.
  4. Compare it with the theoretical Gaussian curve.

Experiment 3: Effect of Drift

How does a constant drift change the distribution?

  1. Set drift μ = 0.5 and D = 1.0.
  2. Run the simulation and watch ⟨x⟩.
  3. Expected result: ⟨x⟩ = μt.
  4. The variance remains 2Dt, so drift changes location but not spread.

Experiment 4: Stock Price Simulation

Compare different market scenarios.

  1. Switch to Finance Mode.
  2. Try different μ and σ combinations.
  3. Bull market example: μ = 0.15, σ = 0.2.
  4. Bear market example: μ = -0.05, σ = 0.3.
  5. Compare the probability of profit and loss.

Experiment 5: Option Pricing

Use simulation to understand Black-Scholes pricing.

  1. Set S₀ = 100, K = 100, and T = 1 year.
  2. Simulate 1000 price paths.
  3. Compute the call payoff max(Sₜ - K, 0).
  4. Average and discount it: e^(-rT) × E[payoff].
  5. Compare the result with the Black-Scholes formula.