Differential Equations – MIT‑Level Study Notes
(Prepared for the INSPIRE Scholarship – Mathematics)
Scope – Ordinary Differential Equations (ODEs)
Audience – First‑year graduate students in mathematics or physics
Goal – Build a solid, rigorous foundation, explore advanced theory, and prepare for research‑level problems.
1. Foundations – From Calculus to ODEs
| Concept | First‑principle definition | Key properties |
|---|---|---|
| Differential equation | An equation involving an unknown function (y(x)) and its derivatives (y^{(k)}(x)). | Specifies a family of curves; solvable when enough initial data are given. |
| Order | Highest derivative appearing. | 1st order → curves; 2nd order → oscillatory/accelerated behaviour. |
| Linearity | (L[y] = \sum_{k=0}^{n} a_k(x)y^{(k)}(x) = g(x)). | Superposition holds; linear combination of solutions is a solution. |
| Homogeneous vs. Non‑homogeneous | (g(x) \equiv 0) → homogeneous. | Particular + homogeneous solution. |
1.1 Existence & Uniqueness (Picard–Lindelöf)
Theorem (Picard–Lindelöf)
Let (f(x,y)) be continuous on a rectangle (R = {(x,y): |x-x_0|\le a, |y-y_0|\le b}) and locally Lipschitz in (y). Then the IVP
[ y'(x) = f(x,y(x)),\qquad y(x_0)=y_0 ] has a unique solution on ([x_0-a,x_0+a]).
Proof Sketch – Picard iterations (y_{n+1}(x) = y_0 + \int_{x_0}^{x} f(t,y_n(t)),dt). Contraction mapping on a suitable function space.
Edge case – Peano’s theorem guarantees existence under continuity alone, but uniqueness may fail (e.g. (y' = \sqrt{|y|})).
2. First‑Order ODEs – Systematic Solution Techniques
2.1 Separable Equations
If (y' = f(x)g(y)), then
[
\int \frac{dy}{g(y)} = \int f(x),dx + C.
]
Edge case: (g(y)=0) at some point → singular solution.
2.2 Linear First‑Order Equations
[
y' + p(x)y = q(x).
]
Integrating factor (\mu(x)=e^{\int p(x)dx}).
Solution:
[
y(x) = \frac{1}{\mu(x)}\left(\int \mu(x)q(x),dx + C\right).
]
Rigorous point: (\mu(x)) never zero, so division is legitimate.
2.3 Bernoulli & Riccati Equations
Bernoulli: (y' + p(x)y = q(x)y^n). Substitution (z = y^{1-n}) reduces to linear.
Riccati: (y' = a(x) + b(x)y + c(x)y^2).
- If a particular solution (y_p) is known, substitution (y = y_p + 1/v) yields a linear first‑order ODE in (v).
- Edge case – when (c(x)=0), Riccati reduces to linear.
3. Linear Second‑Order ODEs – Theory & Techniques
General form: [ y'' + p(x)y' + q(x)y = g(x). ]
3.1 Homogeneous Equation
(y'' + p(x)y' + q(x)y = 0).
3.1.1 Reduction of Order
If (y_1(x)) is known, set (y = y_1 v). Then (v) satisfies a first‑order equation in (v').
3.1.2 Method of Variation of Parameters
Find two linearly independent solutions (y_1, y_2) of the homogeneous equation.
Particular solution:
[
y_p = -y_1\int \frac{y_2 g}{W},dx + y_2\int \frac{y_1 g}{W},dx,
]
where (W = y_1y_2' - y_1'y_2) is the Wronskian.
3.1.3 Constant Coefficients
Characteristic polynomial (\lambda^2 + p\lambda + q = 0).
- Distinct real roots → (y = C_1 e^{\lambda_1 x} + C_2 e^{\lambda_2 x}).
- Complex roots → oscillatory solutions.
- Repeated roots → (y = (C_1 + C_2 x)e^{\lambda x}).
3.2 Non‑homogeneous Equation – Green’s Function
Define (G(x,\xi)) solving [ \mathcal{L}G = \delta(x-\xi),\quad \mathcal{L} = \frac{d^2}{dx^2} + p(x)\frac{d}{dx} + q(x). ] Then [ y(x) = \int G(x,\xi)g(\xi),d\xi. ]
Rigorous note: (G) is constructed from fundamental solutions and satisfies boundary conditions.
3.3 Sturm–Liouville Theory (Eigenvalue Problems)
Consider [ -(p(x)y')' + q(x)y = \lambda w(x)y,\qquad y(a)=y(b)=0. ]
- Eigenvalues (\lambda_n)