Summary of Differential Equations
1. Foundations
Notation:
- Leibniz: dy/dx
- Lagrange: f’(x), f”(x)
- Newton: ẏ, ÿ (for time derivatives)
Differentials vs Derivatives:
- Derivative dy/dx is defined as a limit (single symbol)
- Differentials dx and dy are separate objects, defined later
- dy = f’(x)·dx (definition)
- Their ratio equals the derivative (by construction)
- This justifies “separating” dy/dx algebraically
Multivariable chain rule:
2. First-Order ODEs
Separable:
- Form: dy/dx = f(x)·g(y)
- Method: Separate → ∫(1/g)dy = ∫f dx
- Key: RHS must be product, not sum
Linear:
- Form: y’ + P(x)y = Q(x)
- Method: Integrating factor I = e^(∫P dx)
- Solution: y = (1/I)∫I·Q dx
- No constant in integrating factor; constant appears in final integration
Bernoulli:
- Form: y’ + P(x)y = Q(x)yⁿ
- Substitution: v = y^(1-n)
- Divide by yⁿ first, then substitute
- Transforms to linear equation
Homogeneous (y/x type):
- Form: y’ = F(y/x)
- Substitution: v = y/x, so y = vx, y’ = v’x + v
- Transforms to separable
Exact:
- Form: M(x,y)dx + N(x,y)dy = 0
- Test: ∂M/∂y = ∂N/∂x
- Meaning: ∃F such that dF = Mdx + Ndy
- Solution: F(x,y) = C
3. Second-Order Linear ODEs
Homogeneous with Constant Coefficients
Form: ay” + by’ + cy = 0
Method: Guess y = e^(rx), get characteristic equation ar² + br + c = 0
Three cases based on discriminant b² - 4ac:
| Discriminant | Roots | Solution |
|---|---|---|
| > 0 | Two real r₁, r₂ | y = C₁e^(r₁x) + C₂e^(r₂x) |
| = 0 | Repeated r | y = (C₁ + C₂x)e^(rx) |
| < 0 | Complex α ± βi | y = e^(αx)(C₁cos(βx) + C₂sin(βx)) |
Why dimension 2:
- Second-order → two initial conditions needed
- Solution space corresponds to R²
- Two independent solutions form a basis
Why xe^(rx) for repeated roots:
- Reduction of order method
- Assume y = v·e^(rx), find v” = 0, so v = Ax + B
Complex roots explained:
- e^(ix) = cos(x) + i·sin(x) (Euler’s formula)
- Real/imaginary parts of complex solution are each solutions (for real coefficients)
Non-Homogeneous
Form: ay” + by’ + cy = g(x)
General solution: y = y_c + y_p
Method 1: Undetermined Coefficients
- Guess y_p based on form of g(x)
- Works for: polynomials, exponentials, sin, cos, products
- If guess matches y_c, multiply by x
- Why multiplying by x works: L[xf] = x·L[f] + (extra terms) = 0 + (extra terms)
Method 2: Variation of Parameters
- Always works
- Assume y_p = u₁y₁ + u₂y₂
- Impose: u₁’y₁ + u₂’y₂ = 0 (simplifies derivatives)
- ODE gives: u₁’y₁’ + u₂’y₂’ = g(x)
- Solve system for u₁’, u₂’, then integrate
Wronskian:
- Measures linear independence
- Used in variation of parameters
4. Applications
Mixing/Brine Problems
Setup:
- y(t) = amount of substance in tank
- dy/dt = (rate in) - (rate out)
Key formulas:
- Rate in = C₁ · r₁ (concentration × flow rate)
- Rate out = (y/V) · r₂ (current concentration × flow rate)
- V(t) = V₀ + (r₁ - r₂)t
Three cases:
| Condition | Volume | Solution shape |
|---|---|---|
| r₁ = r₂ | constant | exponential decay |
| r₁ > r₂ | grows | 1/(a+t)ⁿ |
| r₁ < r₂ | shrinks | (a-t)ⁿ |
Newton’s Cooling
Solution: T(t) = T_ambient + (T₀ - T_ambient)e^(-kt)
Circuits
RL circuit (first-order):
RLC circuit (second-order):
Damping cases:
| Case | Discriminant | Behavior |
|---|---|---|
| Overdamped (sovrasmorzato) | > 0 | Smooth decay, no oscillation |
| Critically damped (smorzamento critico) | = 0 | Fastest decay without oscillation |
| Underdamped (sottosmorzato) | < 0 | Oscillates while decaying |
Springs
Hooke’s Law: F = -kx (spring force)
Newton’s Law: F = ma = mx”
Combined: mx” + kx = 0
With damping: mx” + cx’ + kx = 0
Same three cases as circuits.
5. Qualitative Analysis
Direction Fields
- At each point (t, y), draw slope = f(t, y)
- Direction vector: (1, f(t,y))
- Solution curves are tangent to the field
Autonomous Equations
Form: dy/dt = f(y) (no explicit t dependence)
Key property: Same y → same slope, regardless of t
Direction field: Horizontal bands of constant slope
Equilibrium Solutions
Definition: Values where f(y) = 0 (constant solutions)
Classification:
- Stable (attractor): Both sides push toward equilibrium
- Unstable (repeller): Both sides push away
- Semi-stable: One side toward, one side away
Method: Check sign of f(y) just above and below equilibrium
Euler’s Method
- Numerical approximation
- Follow direction field in small steps
- n = step number, not time value
6. Power Series Solutions
When to Use
- At ordinary points (punto ordinario): P(x) and Q(x) analytic
- At singular points (punto singolare): Need Frobenius method
Method
Step 1: Write y = Σcₙxⁿ and derivatives as series
Step 2: Substitute into ODE
Step 3: Get series “in phase” (same starting power of x)
- May need to pull terms outside sum
Step 4: Re-index to match powers
Step 5: Combine and set coefficient of each xᵏ to zero
Step 6: Get recurrence relation, solve for cₙ
Key Techniques
Re-indexing example:
- If sum has x^(n-1), let k = n-1, n = k+1
- Subscript cₙ becomes c_{k+1}
Series multiplication:
- For cos(x)·y, multiply term by term
- Collect like powers of x
Common Results
7. Solution Process Summary
First-order:
- Is it separable? → Separate and integrate
- Is it linear? → Integrating factor
- Is it Bernoulli? → Substitute v = y^(1-n)
- Is it homogeneous (y/x)? → Substitute v = y/x
- Is it exact? → Find F where dF = 0
Second-order linear:
- Solve homogeneous first (characteristic equation)
- If non-homogeneous, find y_p via:
- Undetermined coefficients (if g(x) is nice)
- Variation of parameters (always works)
- General solution: y = y_c + y_p
- Apply initial conditions last