01 / 20

IVPs &
Existence/Uniqueness

Understand Initial Value Problems, the conditions that guarantee a solution exists, and when that solution is unique.

Differential Equations

What Is an ODE?

An ordinary differential equation (ODE) is an equation that relates an unknown function \( y(t) \) to its derivatives. The order of an ODE is the highest derivative that appears.

ODEs model countless real-world phenomena: population growth, radioactive decay, circuit currents, and motion. But to pin down a single solution we need more information.

What Is an Initial Value Problem?

An Initial Value Problem (IVP) pairs an ODE with an initial condition that specifies the value of \( y \) at a particular point \( t_0 \).

Think of the ODE as defining a slope field: every point \( (t,y) \) gets an assigned slope. The initial condition tells us where to start tracing.

Visualising an IVP on a Slope Field

t y \( y(0)=3 \) \( y(0)=1 \)

Why Existence & Uniqueness Matter

Existence
Does a solution to the IVP actually exist? Without a guarantee, we might be trying to solve an impossible problem.
Uniqueness
If a solution exists, is it the only one? In engineering and physics, a model that admits two different futures from the same state is unreliable.

The Picard-Lindelöf Theorem

Condition (i)
\( f \) continuous near \( (t_0, y_0) \) guarantees existence (Peano).
Condition (ii)
\( \partial f/\partial y \) continuous adds uniqueness (Lipschitz condition is implied).

Geometric Interpretation

R t y \( (t_0, y_0) \) \( |t - t_0| < h \) unique curve

Inside the rectangle \( R \), the theorem guarantees that through the initial point passes exactly one solution curve (shown in violet). The solution is guaranteed at least on the amber interval \( |t - t_0| < h \).

Existence vs Uniqueness: Key Distinction

Existence Only Existence + Uniqueness
Condition on \( f \) \( f \) continuous \( f \) continuous
Condition on \( \partial f / \partial y \) Not required \( \partial f / \partial y \) continuous
Guarantee At least one solution Exactly one solution
Classic theorem Peano Picard-Lindelöf

How to Apply the Theorem

1
Identify \( f(t,y) \): rewrite the ODE as \( y' = f(t,y) \) and note the initial condition \( (t_0, y_0) \).
2
Check continuity of \( f \): look for division by zero, square roots of negatives, or logarithms of non-positives near \( (t_0, y_0) \).
3
Compute \( \partial f / \partial y \): differentiate \( f \) with respect to \( y \), treating \( t \) as a constant.
4
Check continuity of \( \partial f / \partial y \): verify it is continuous on an open rectangle around \( (t_0, y_0) \).
5
State your conclusion: both continuous → unique solution exists. Only \( f \) continuous → at least one solution exists.

Example 1: Polynomial ODE

1
Here \( f(t,y) = t^2 + y^2 \). This is a polynomial in \( t \) and \( y \), so it is continuous everywhere.
2
Compute: \( \dfrac{\partial f}{\partial y} = 2y \). This is also continuous everywhere.
3
Both conditions hold on any rectangle around \( (0,1) \). By the Picard-Lindelöf theorem, a unique solution exists near \( t = 0 \). ✓

Example 2: Rational Function

1
\( f(t,y) = \dfrac{y}{t-2} \). This is discontinuous when \( t = 2 \), but at \( t_0 = 1 \) it is continuous.
2
\( \dfrac{\partial f}{\partial y} = \dfrac{1}{t-2} \). Also continuous at \( t = 1 \) (and on any rectangle that excludes \( t = 2 \)).
3
Choose \( R \) as \( 0 < t < 2 \), any \( y \). Both conditions hold, so a unique solution exists on an interval around \( t = 1 \) (at least until \( t \) approaches 2). ✓

Example 3: Uniqueness Fails

Here \( f(t,y) = 3y^{2/3} \) is continuous everywhere, so at least one solution exists. But:

!
\( \dfrac{\partial f}{\partial y} = 2\,y^{-1/3} = \dfrac{2}{y^{1/3}} \), which is undefined at \( y = 0 \). The uniqueness condition fails.

Visualising Non-Uniqueness

t y \( y=0 \) \( y=t^3 \) \( (0,0) \) 1 -1 2 -2

For \( y' = 3y^{2/3},\; y(0)=0 \), two distinct solution curves pass through the initial point. The blue line is \( y = 0 \), and the red curve is \( y = t^3 \). The uniqueness condition failed, and we see the consequence.

Common Mistakes

✕ Checking only \( f \)
Verifying continuity of \( f \) alone proves existence but not uniqueness. You must also check \( \partial f / \partial y \).
✕ Wrong partial derivative
Taking \( \partial f / \partial t \) instead of \( \partial f / \partial y \). The theorem specifically requires the partial with respect to the dependent variable.
✕ Global vs local
The theorem guarantees a solution near \( t_0 \), not for all \( t \). Solutions can blow up or leave the region \( R \) at finite time.

Summary Checklist

Step Action Result
1 Write \( y' = f(t,y) \) Identify \( f \)
2 Check: \( f \) continuous at \( (t_0, y_0) \)? Existence
3 Compute \( \partial f / \partial y \) Prepare for uniqueness check
4 Check: \( \partial f / \partial y \) continuous at \( (t_0, y_0) \)? Uniqueness
5 State conclusion Unique solution / existence only / neither

Question 1 of 5

Consider the IVP \( y' = t \cdot y, \; y(0) = 2 \). Which statement is correct?

Question 2 of 5

For the IVP \( y' = \sqrt{y},\; y(0) = 0 \), what can we conclude?

Question 3 of 5

Which condition does the Picard-Lindelöf theorem require for uniqueness (beyond what Peano's theorem requires for existence)?

Question 4 of 5

For the IVP \( y' = \dfrac{y}{t+1},\; y(0) = 5 \), on what interval can we guarantee a unique solution?

Question 5 of 5

If the Picard-Lindelöf conditions fail at \( (t_0, y_0) \), which conclusion is valid?