Existence and Uniqueness Theorems: Explained with Examples

Existence and uniqueness theorems answer the most basic question about a differential equation: given a starting value y(x0) = y0, does a solution exist, and is there only one? Without these guarantees, a model could have no solution at all, or several contradictory ones, making prediction meaningless.

Every ODE model assumes its predictions are fixed by the initial conditions. These theorems say when that assumption is safe — and warn about equations like dy/dx = √|y|, where it silently fails.

Key Concepts

• Initial value problem (IVP). dy/dx = f(x, y) with y(x0) = y0.
• Picard–Lindelöf theorem. If f and ∂f/∂y are continuous near (x0, y0), a unique solution exists on some interval around x0.
• Lipschitz condition. |f(x, y1) − f(x, y2)| ≤ L|y1 − y2|; the real engine behind uniqueness.
• Local vs global. The solution is guaranteed near x0 only; it may blow up further out.
• Non-uniqueness example. dy/dx = √|y|, y(0) = 0 has both y = 0 and y = x²/4 (x ≥ 0) as solutions, since ∂f/∂y blows up at y = 0.

Formulas

• Picard iteration: y_{n+1}(x) = y0 + ∫_{x0}^{x} f(t, y_n(t)) dt converges to the unique solution.
• Blow-up illustration: dy/dx = y², y(0) = 1 gives y = 1/(1 − x), unique near 0 but gone at x = 1.
• Assumptions and limits: continuity of f alone gives existence (Peano) but not uniqueness; uniqueness needs Lipschitz, usually via continuity of ∂f/∂y.

Worked Example

Analyze the IVP dy/dx = y² with y(0) = 1, and contrast with dy/dx = √|y|, y(0) = 0.

1. Identify f(x, y) = y² and compute ∂f/∂y = 2y. Both are continuous everywhere, so Picard–Lindelöf applies at (0, 1): a unique solution exists near x = 0.
2. Solve by separation: dy/y² = dx gives −1/y = x + C. With y(0) = 1, C = −1, so y = 1/(1 − x).
3. Verify: dy/dx = 1/(1 − x)² = y², and y(0) = 1. Correct, and uniqueness says no other solution passes through (0, 1).
4. Note the solution blows up at x = 1, showing the guarantee is local: existence near 0 does not mean existence for all x.
5. Contrast: for dy/dx = √|y|, y(0) = 0, we have ∂f/∂y = 1/(2√|y|) for y ≠ 0, which is unbounded near y = 0, so the theorem does not apply.
6. Indeed y = 0 and y = x²/4 (for x ≥ 0) both satisfy the ODE and the initial condition, since d/dx(x²/4) = x/2 = √|x²/4|. Non-uniqueness confirmed.

Common Misconceptions

• Thinking existence means global existence. The theorem is local; y = 1/(1 − x) is the unique solution near 0 yet blows up at x = 1.
• Assuming continuity of f gives uniqueness. Peano's theorem gives existence from continuity alone, but uniqueness needs the Lipschitz condition on top.
• Believing non-uniqueness is a pathology to ignore. Equations like dy/dx = √|y| model real phenomena (e.g. certain draining-tank idealizations), where the extra solutions carry physical meaning.

Real-World Applications

• Model validation: before simulating an ODE model, checking Lipschitz continuity confirms the simulation targets a well-defined solution.
• Numerical analysis: error bounds for Euler's method and Runge–Kutta methods rest on the Lipschitz constant L.
• Control theory: uniqueness guarantees that a designed control input produces exactly one system trajectory.

Continuity of f and ∂f/∂y near the initial point guarantees one and only one local solution, via Picard–Lindelöf. The fine print matters: the guarantee is local, and without the Lipschitz condition uniqueness can fail, as dy/dx = √|y| demonstrates.