Learn & Review: Partial Differential Equations

Course Overview: Partial Differential Equations and Functional Analysis

This course is divided into two main parts: the study of specific types of Partial Differential Equations (PDEs) and an introduction to Functional Analysis.

Part 1: Partial Differential Equations (PDEs)

The first part focuses on understanding various classes of PDEs.

• First-Order PDEs:
  • Linear and nonlinear first-order equations will be covered.
  • A specific example discussed is the linear transport equation:
    • Equation: $u_t + \mathbf{b} \cdot \nabla u = 0$, where $u$ is a function of time $t$ and space $\mathbf{x} \in \mathbb{R}^n$, $\mathbf{b}$ is a constant vector, $u_t$ is the time derivative, and $\nabla u$ is the spatial gradient.
    • Characteristics: The solutions are constant along lines parallel to the vector $(1, \mathbf{b})$. These lines are called characteristics.
    • Initial Value Problem: When coupled with an initial condition $u(0, \mathbf{x}) = \bar{u}(\mathbf{x})$ on the hyperplane $t=0$, the solution is given by $u(t, \mathbf{x}) = \bar{u}(\mathbf{x} - t\mathbf{b})$.
    • Non-regularizing Effect: This type of first-order linear PDE does not regularize initial conditions; the solution's smoothness matches that of the initial data.
• Second-Order PDEs:
  • Elliptic Equations: The Laplace equation ($\Delta u = 0$, where $\Delta$ is the Laplacian operator) will be studied.
  • Parabolic Equations: The heat equation will be a primary example.
  • Hyperbolic Equations: The wave equation will be studied.

Part 2: Functional Analysis

The second part delves into more abstract concepts.

• Key Concepts:
  • Hilbert Spaces: Vector spaces with an inner product, often infinite-dimensional in this context, which presents a significant challenge.
  • Banach Spaces: Complete normed vector spaces.
• Applications: Functional analysis is crucial for understanding the existence and properties of solutions to PDEs, particularly in large classes of functions (e.g., Sobolev spaces, distributions).

Suggested Books

• For PDEs (Part 1):
  • Partial Differential Equations by L.C. Evans: A standard and comprehensive reference. The course will not cover the entire book but will use it as a primary resource.
• For Functional Analysis (Part 2) and PDEs:
  • Functional Analysis and Partial Differential Equations by V.G. Maz'ya (or similar titles by Bratiz): A standard reference for the more abstract part of the course.

Key Problems in Studying PDEs

When analyzing a new PDE, several fundamental problems arise:

1. Finding Special Explicit Solutions: Looking for solutions with specific symmetries (e.g., radial, time-independent).
2. Existence of Solutions: Defining what constitutes a "solution" and proving its existence within a specific function space. Larger function spaces make existence easier but may complicate uniqueness.
3. Uniqueness of Solutions: Proving that only one solution exists within a given class. Smaller function spaces tend to ensure uniqueness.
4. Regularity of Solutions: Determining the smoothness of the solution. Often, solutions are smoother than expected based on the initial data and the PDE itself, a phenomenon known as regularization.

These points are interconnected, and a comprehensive understanding requires considering them simultaneously. Functional analysis, particularly through concepts like Sobolev spaces and distribution theory, plays a vital role in addressing existence and regularity.

Linear Transport Equation: Detailed Analysis

The course begins with a detailed examination of the linear transport equation ($u_t + \mathbf{b} \cdot \nabla u = 0$).

• Properties:
  • Linear: Derivatives of $u$ appear linearly.
  • First-Order: Involves only first-order derivatives.
  • Homogeneous: The right-hand side is zero.
  • Constant Coefficients: Coefficients are constant.
• Solution Method (Method of Characteristics):
  • The equation implies that the directional derivative of $u$ along the vector $(1, \mathbf{b})$ is zero.
  • This means $u$ is constant along lines parallel to $(1, \mathbf{b})$. These lines are the characteristics.
  • For an initial value problem with $u(0, \mathbf{x}) = \bar{u}(\mathbf{x})$, the solution is found by tracing back along the characteristics to the initial hyperplane ($t=0$).
  • The explicit solution is $u(t, \mathbf{x}) = \bar{u}(\mathbf{x} - t\mathbf{b})$.
• Remarks on the Solution:
  • Smoothness: The solution $u(t, \mathbf{x})$ has the same smoothness as the initial data $\bar{u}(\mathbf{x})$. The equation does not regularize the initial condition.
  • Transversality: The method works because the characteristic direction $(1, \mathbf{b})$ is transverse to the initial surface $\Sigma$ (e.g., the hyperplane $t=0$).
  • Transport Interpretation: The initial condition is "transported" through space and time along the characteristics.
  • Weak Solutions: If $\bar{u}$ is discontinuous, the solution $u(t, \mathbf{x})$ may also be discontinuous, requiring a notion of "weak solution."
  • Characteristics: The lines followed by the solution are solutions to a system of ordinary differential equations (ODEs).

Non-Homogeneous Linear Transport Equation

The analysis extends to the non-homogeneous case: $u_t + \mathbf{b} \cdot \nabla u = f(t, \mathbf{x})$.

• Solution Approach:
  • The method of characteristics is adapted. The derivative of $u$ along the characteristic direction is now equal to $f$.
  • Integrating along the characteristic from the initial time ($s=-t$) to the current time ($s=0$) yields the solution.
• Explicit Solution:
  • $u(t, \mathbf{x}) = \bar{u}(\mathbf{x} - t\mathbf{b}) + \int_0^t f(\tau, \mathbf{x} - (t-\tau)\mathbf{b}) d\tau$.
  • This solution incorporates the initial condition and the effect of the source term $f$ over time.
• Method of Characteristics (Generalization):
  • The process involves solving a system of ODEs (the characteristic equations) to define the characteristic curves.
  • The solution $u$ is then constructed using the initial data and the source term integrated along these characteristics. This approach is fundamental for solving more complex PDEs.