Skip to content
Stochastic Calculus / Diffusions, SDEs, and the Feynman-Kac Formula

Derivation — Feynman-Kac via Itô's Formula

The Feynman-Kac theorem states that φ(t,x)=E[e−∫tTr(s,Xs)dsF(XT)∣Xt=x]\varphi(t,x) = \mathbb{E}[e^{-\int_t^T r(s,X_s)ds} F(X_T) \mid X_t = x] solves a parabolic PDE. This derivation note carries out the argument in full: we apply Itô's formula to e−∫tsr(u,Xu) duφ(s,Xs)e^{-\int_t^s r(u,X_u) \, du} \varphi(s, X_s), observe that the result must be a martingale (by the tower property), and conclude that the drift term must vanish — which is exactly the PDE. Evaluating at s=Ts = T recovers the terminal condition.
The same argument yields the generator form of the PDE as a byproduct. Two related approaches — the martingale method and the tower-property method — are compared.

In this lesson

  1. Setup
  2. Step 1 — The Discounted Value is a Martingale
  3. Step 2 — Apply Itô's Formula to MsM_s
  4. Step 3 — Martingale Forces the Drift to Zero
  5. Step 4 — Terminal Condition
  6. The Generator Approach
  7. The Converse — PDE to Expectation
  8. Finance: From PDE to Monte Carlo and Back

Before you start

The full lesson, with worked examples and exercise solutions, is for members.

Read the full lesson

New to SelfMadeQuant.com? See membership options or explore the curriculum.