Higher-Order ODEs & PDEs: Methods, Solutions & Applications

TAKEAWAYS:

• Higher-order ODEs require reduction to first-order systems (e.g., via substitution) or Laplace transforms for analytical solutions, while numerical methods like finite differences or shooting methods handle boundary value problems.
• Partial Differential Equations (PDEs) are classified by type (elliptic/parabolic/hyperbolic) and solved via separation of variables, finite difference schemes (explicit/implicit), or Fourier transforms.
• Stability in PDE solvers depends on the Courant-Friedrichs-Lewy (CFL) condition (e.g., for heat equation: $r=\frac{\alpha \Delta t}{(\Delta x{)}^2}\le 0.5$).
• Boundary conditions (Dirichlet/Neumann/Mixed) critically impact solution uniqueness and convergence; Galerkin methods extend finite differences for complex geometries.
• Applications include heat conduction (parabolic), wave propagation (hyperbolic), and electrostatics (elliptic PDEs).
• Exam focus: Derive finite difference schemes, classify PDEs, solve model problems (e.g., 1D heat equation), and discuss stability constraints.

1. Higher-Order Ordinary Differential Equations (ODEs)

1.1 Classification and Analytical Methods

Higher-order ODEs (e.g., $y^{\prime \prime \prime }+p(x)y^{\prime \prime }+q(x)y^{\prime }+r(x)y=g(x)$) are solved via:

• Reduction of order: Convert to a system of first-order ODEs (e.g., let $y_1=y$, $y_2=y^{\prime }$, $y_3=y^{\prime \prime }$).
• Laplace transforms: Useful for linear ODEs with constant coefficients and initial conditions.
• Power series solutions: For ODEs with variable coefficients (e.g., Bessel’s equation).

Example: Solve $y^{\prime \prime }+y=0$ with $y(0)=1$, $y^{\prime }(0)=0$

• Analytical solution: $y(x)=\cos x$ (via characteristic equation).
• Numerical reduction: Rewrite as: $$\{\begin{array}{l}{y}_1^{\prime }=y_2\\ y_2^{\prime }=-y_1\end{array}$$ Solve using Runge-Kutta methods (from Unit 7).

1.2 Boundary Value Problems (BVPs)

BVPs specify conditions at multiple points (e.g., $y(a)=\alpha$, $y(b)=\beta$). Methods:

• Shooting method: Convert BVP to IVP by guessing initial slopes, then adjust iteratively (like Newton-Raphson for slopes).
• Finite difference method: Approximate derivatives (e.g., $y^{\prime \prime }\approx \frac{y_{i-1}-2y_i+y_{i+1}}{h^2}$) and solve the resulting linear system.

Example: Solve $y^{\prime \prime }=f(x,y)$ with $y(0)=0$, $y(1)=1$

1. Discretize $x$-domain: $x_i=i\Delta x$, $\Delta x=0.1$.
2. Approximate $y^{\prime \prime }(x_i)\approx \frac{y_{i-1}-2y_i+y_{i+1}}{(\Delta x{)}^2}$.
3. Solve the tridiagonal system using Thomas algorithm (from Unit 2).

1.3 Comparison: IVP vs. BVP Methods

2. Partial Differential Equations (PDEs)

2.1 Classification of PDEs

PDEs are classified by the highest-order derivative and discriminant $D=b^2-4ac$:

• Elliptic ($D<0$): Steady-state problems (e.g., Laplace equation ${\nabla }^{2}u=0$).
• Parabolic ($D=0$): Diffusion problems (e.g., heat equation $u_t=\alpha u_{xx}$).
• Hyperbolic ($D>0$): Wave propagation (e.g., wave equation $u_{tt}=c^2{u}_{xx}$).

Example: Classify $3u_{xx}+2u_{xy}-u_{yy}+u_t=0$

• Rewrite as $3u_{xx}+2u_{xy}-u_{yy}=-u_t$.
• Discriminant: $D=(2{)}^2-4(3)(-1)=4+12=16>0$ → Hyperbolic.

2.2 Separation of Variables

For linear PDEs with homogeneous BCs, assume $u(x,t)=X(x)T(t)$. Steps:

1. Substitute into PDE to separate variables (e.g., for heat equation: $\frac{T^{\prime }}{T}=\alpha \frac{X^{\prime \prime }}{X}=-\lambda$).
2. Solve ODEs for $X(x)$ and $T(t)$.
3. Apply BCs to find eigenvalues ${\lambda }_n$ and eigenfunctions $X_n(x)$.
4. Superpose solutions: $u(x,t)=\sum c_n{X}_n(x)e^{-\alpha {\lambda }_{n}t}$.

Example: Solve $u_t=u_{xx}$ (0 < x < L, t > 0) with $u(0,t)=u(L,t)=0$, $u(x,0)=f(x)$

1. Assume $u(x,t)=X(x)T(t)$.
2. Separate: $\frac{T^{\prime }}{T}=\frac{X^{\prime \prime }}{X}=-\lambda$.
3. Solve $X^{\prime \prime }+\lambda X=0$ with $X(0)=X(L)=0$:
   • Eigenvalues: ${\lambda }_n={\left(\frac{n\pi }{L}\right)}^2$, $n=1,2,\dots$.
   • Eigenfunctions: $X_n(x)=\sin \left(\frac{n\pi x}{L}\right)$.
4. General solution: $$u(x,t)=\sum _{n=1}^{\infty }{c}_n\sin \left(\frac{n\pi x}{L}\right)e^{-\alpha {\left(\frac{n\pi }{L}\right)}^{2}t}$$
5. Apply initial condition $u(x,0)=f(x)$ to find $c_n$ via Fourier sine series.

2.3 Finite Difference Methods for PDEs

Approximate derivatives using Taylor expansions and solve the resulting system.

Example: 1D Heat Equation $u_t=\alpha u_{xx}$

1. Discretization:
   • Space: $x_i=i\Delta x$, $\Delta x=L/N$.
   • Time: $t_j=j\Delta t$, $u_i^j\approx u(x_i,t_j)$.
2. Finite difference approximations:
   • $u_t\approx \frac{u_i^{j+1}-u_i^j}{\Delta t}$ (forward Euler).
   • $u_{xx}\approx \frac{u_{i-1}^j-2u_i^j+u_{i+1}^j}{(\Delta x{)}^2}$.
3. Explicit scheme (forward time, centered space): $$u_i^{j+1}=u_i^j+r(u_{i-1}^j-2u_i^j+u_{i+1}^j),r=\frac{\alpha \Delta t}{(\Delta x{)}^2}$$
   • Stability condition: $r\le 0.5$ (CFL condition).
4. Implicit scheme (backward Euler): $$u_i^{j+1}-r{u}_{i-1}^{j+1}+(1+2r)u_i^{j+1}-r{u}_{i+1}^{j+1}=u_i^j$$
   • Solve tridiagonal system at each time step (unconditionally stable).

Mermaid Diagram: Finite Difference Schemes for Heat Equation

flowchart TD
    A[Heat Equation: $u_t=\alpha u_{xx}$] --> B[Discretize Space/Time]
    B --> C[Explicit Method\n$u_i^{j+1}=u_i^j+r(u_{i-1}^j-2u_i^j+u_{i+1}^j)$]
    B --> D[Implicit Method\nSolve $(I-rA)u^{j+1}=u^j$]
    C --> E[Stability: $r\le 0.5$]
    D --> F[Unconditionally Stable]
    E -->|Violation| G[Oscillations/Blow-up]

2.4 Boundary and Initial Conditions

• Dirichlet: Specify $u$ (e.g., $u(0,t)=0$).
• Neumann: Specify derivative (e.g., $u_x(0,t)=0$).
• Mixed: Combine Dirichlet/Neumann (e.g., $u(0,t)=0$, $u_x(1,t)=0$).

Example: Neumann BC for Heat Equation

• Approximate $u_x(0,t)\approx \frac{u_1^j-u_0^j}{\Delta x}=0$ → $u_0^j=u_1^j$.

2.5 Applications of PDEs

3. Numerical Techniques for Special Cases

3.1 Laplace’s Equation in 2D

For ${\nabla }^{2}u=0$ on a rectangle:

1. Use five-point stencil: $$u_{i,j}=\frac{1}{4}(u_{i-1,j}+u_{i+1,j}+u_{i,j-1}+u_{i,j+1})$$
2. Solve iteratively (e.g., Gauss-Seidel) or directly (e.g., LU decomposition).

Example: Solve ${\nabla }^{2}u=0$ on $0<x<1$, $0<y<1$ with $u(0,y)=0$, $u(1,y)=\sin (\pi y)$, $u(x,0)=u(x,1)=0$

• Discretize with $\Delta x=\Delta y=0.1$.
• Apply five-point formula and iterate until convergence.

3.2 Wave Equation

For $u_{tt}=c^2{u}_{xx}$, use leapfrog scheme: $$\frac{u_i^{j+1}-2u_i^j+u_i^{j-1}}{(\Delta t{)}^2}=c^2\frac{u_{i+1}^j-2u_i^j+u_{i-1}^j}{(\Delta x{)}^2}$$

• Stability condition: $c\frac{\Delta t}{\Delta x}\le 1$.

4. Exam Tip: How to Score Full Marks

Key Strategies

1. Classify PDEs correctly:

   • Always compute the discriminant $D=b^2-4ac$ for second-order PDEs.
   • Example: For $u_{xx}+2u_{xy}-3u_{yy}=0$, $D=4-4(1)(-3)=16>0$ → Hyperbolic.

2. Separation of variables:

   • Show every step of separating $u(x,t)=X(x)T(t)$.
   • Clearly state the ODEs for $X$ and $T$, and how BCs determine eigenvalues.

3. Finite difference schemes:

   • For heat/wave equations, derive the scheme from Taylor expansions.
   • State the stability condition (e.g., $r\le 0.5$ for explicit heat equation).
   • For implicit methods, mention tridiagonal systems and Thomas algorithm.

4. Boundary conditions:

   • Distinguish between Dirichlet, Neumann, and mixed BCs.
   • Show how to approximate Neumann BCs (e.g., $u_x\approx \frac{u_{i+1}-u_{i-1}}{2\Delta x}$).

5. Higher-order ODEs:

   • For BVPs, explain the shooting method or finite differences.
   • For IVPs, reduce to a system and apply Runge-Kutta (from Unit 7).

6. Common Pitfalls:

   • Forgetting stability conditions: Always check $r$ or CFL for PDEs.
   • Incorrect discretization: Ensure second derivatives use central differences.
   • Boundary treatment: Misapplying BCs leads to wrong eigenvalues (e.g., in separation of variables).

Model Answer Structure

1. Classification: State type (elliptic/parabolic/hyperbolic) with justification.
2. Method Selection: Choose separation of variables, finite differences, or shooting method.
3. Derivation: Show all steps (e.g., finite difference approximations).
4. Solution: Write the general solution (e.g., Fourier series) or iterative scheme.
5. Verification: Check stability, BCs, and initial conditions.

Example Exam Question: "Solve $u_t=u_{xx}$ for $0<x<\pi$, $t>0$ with $u(0,t)=u(\pi ,t)=0$, $u(x,0)=\sin x$." Model Answer:

1. Classification: Parabolic PDE (heat equation).
2. Separation: Assume $u(x,t)=X(x)T(t)$, derive $X^{\prime \prime }/X=T^{\prime }/(\alpha T)=-\lambda$.
3. Eigenvalues: ${\lambda }_n=n^2$, $X_n(x)=\sin (nx)$.
4. Solution: $u(x,t)=e^{-\alpha n^{2}t}\sin (nx)$. Apply IC to find $c_1=1$, $c_n=0$ for $n\ne 1$.
5. Final answer: $u(x,t)=e^{-\alpha t}\sin x$.

mindmap
  root((PDE Classification))
    Elliptic (D < 0)
      Example: Laplace's equation
      Applications: Steady heat, electrostatics
    Parabolic (D = 0)
      Example: Heat equation
      Applications: Diffusion, cooling
    Hyperbolic (D > 0)
      Example: Wave equation
      Applications: Vibrations, sound