20.2.1. Finite Difference Method (FDM)

• Concept: Replace derivatives with difference quotients using Taylor series approximations.

The Finite Difference Method (FDM) is a numerical technique that approximates the values of derivatives by using difference quotients. This approach involves breaking down derivatives into simpler forms, using the Taylor series, which expresses functions as infinite sums. In FDM, instead of calculating the exact derivative of a function at a specific point, we calculate the difference between function values at adjacent discretized points, effectively creating a 'finite difference.'

Imagine you have a steep hill and you want to understand how steep it is at different points. Instead of calculating the slope mathematically, you could walk up to two points on the hill, measure the elevation at each point, and calculate the average change in height over that small distance. This gives you an approximate slope, similar to how FDM works by using nearby points to estimate derivatives.

• Grid: The domain is discretized into a grid (mesh) of points.

In FDM, the problem domain (the area where the PDE is defined) is divided into a grid or mesh of points. This grid allows us to assign numerical values to each point, which can represent values of the function we are trying to analyze (like temperature, pressure, etc.). The grid can be uniformly spaced (where all points are equally spaced) or non-uniform, depending on the specifics of the problem being solved.

Consider a map of a city. If you want to know the height of buildings across different streets, you could divide the map into sections and note the height at specific intersections. This divided map helps you visualize how the height varies across the entire area, analogous to how the grid helps us analyze the behavior of functions across a domain in FDM.

• Types: o Explicit Method – Easy to implement but conditionally stable. o Implicit Method – More stable but requires solving systems of equations. o Crank–Nicolson Method – A combination of explicit and implicit; unconditionally stable.

There are several variations of the Finite Difference Method, each with its strengths and weaknesses. The Explicit Method is straightforward to implement, as it calculates the values at the next time step directly from the known values. However, it is conditionally stable, meaning it can produce reliable results only under certain conditions, often requiring small time steps. The Implicit Method, on the other hand, is generally more stable and can accommodate larger time steps, but it involves solving a system of equations at each step, which can increase computational complexity. The Crank–Nicolson Method combines the benefits of both explicit and implicit methods, offering stability while being relatively easy to apply.

Think about making a cake. The Explicit Method is like following a simple recipe that tells you the exact next step based on your current ingredients; it’s quick but works only if certain conditions (like ingredient amounts) are met. The Implicit Method, however, is akin to knowing you have to solve a puzzle to figure out the next step, providing more flexibility in adjusting ingredients but requiring more time to think. The Crank–Nicolson Method is like a hybrid recipe that incorporates both simple and complex steps, giving you reliable results without too much hassle.

Example: Solving the 1D heat equation:

Discretize using finite differences:

To illustrate the Finite Difference Method, consider the one-dimensional heat equation, which describes how temperature changes over time along a rod. To use FDM to approximate this equation, we first discretize both time and space. The equation we derive expresses the change in temperature at a grid point over time, represented as the difference between current and future temperature values. This discrete equation allows us to compute the temperature at the next time step using known temperatures from the current time and surrounding positions.

Imagine you're tracking the temperature along a hot rod (like a metal rod warming up). Each point on the rod is like a temperature sensor that reads the current temperature. If you only know the temperatures at certain moments, you can use the readings from these sensors to estimate the future temperature at each point. This method of using current data to predict future states mimics how FDM approximates the continuous heat distribution with discrete data points.

--