Examples of Box-Counting

The box-counting technique involves covering a geometric figure with increasingly smaller boxes while tracking the number of occupied boxes at each scale. To enhance our comprehension of what a dimensional measurement signifies, we can begin by performing box-counting on a simple line.

In the first example, we cover a blue line with boxes of decreasing sizes: 1, 1/2, 1/4, and 1/8. The resulting counts, where N(s) indicates the count for a box of size s, are as follows:

• N(1) = 1
• N(1/2) = 2 = 1/(1/2)
• N(1/4) = 4 = 1/(1/4)
• N(1/8) = 8 = 1/(1/8)

From this, we observe that generally, N(s) = 1/s; each time the box size is halved, the count increases by a factor of 2.

Now, let's apply the same method to a filled blue square:

• N(1) = 1
• N(1/2) = 4 = (1/(1/2))²
• N(1/4) = 16 = (1/(1/4))²
• N(1/8) = 64 = (1/(1/8))²

Here, we find that N(s) = (1/s)², indicating that every time the box size is halved, the count quadruples.

Next, we can investigate a well-known fractal, the Sierpinski gasket:

The Sierpinski gasket, named after the Polish mathematician Wacław Sierpiński, can be constructed in several ways and is recognized for its self-similar structure.

We can cover the gasket with the fewest number of non-overlapping boxes at various scales:

• N(1) = 1
• N(1/2) = 3
• N(1/4) = N((1/2)²) = 9 = 3²
• N(1/8) = N((1/2)³) = 27 = 3³

The overarching rule here becomes N((1/2)ⁿ) = 3ⁿ, meaning that each time we halve the box size, the count increases by a factor of 3. This behavior deviates from the expectations for one- or two-dimensional objects, prompting further exploration.

Power-laws in Box-Counting

The relationships established in these examples are defined by a power-law, where one quantity varies in proportion to a power of another. By taking the logarithm of both sides, we can derive a useful equation for determining the dimension of an object based on its scaling properties.

Equation 1 provides a formula for calculating the dimension when the scaling relationships are known. This is known as the similarity dimension, which can be computed for self-similar shapes made of N non-overlapping copies, each scaled by a factor of s.

For the Sierpinski gasket, we find its dimension to be approximately d = Log(3)/Log(2) ≈ 1.58496, indicating it occupies more space than a line (1-dimensional) but less than a plane (2-dimensional).

When dealing with self-similar objects with unknown scaling characteristics, we can utilize log-log plots of box sizes against their respective counts to check for a power-law. If a power-law is present, the data points will align along a regression line, where the slope of this line can be interpreted as the dimension d of the object being measured.

Here are the log-log plots for the previously discussed examples:

Coastlines and Fractal Dimensions

One of the most famous applications of box-counting comes from Benoit Mandelbrot’s 1967 research titled "How Long is the Coast of Britain?" Mandelbrot discovered that as he decreased the size of his measuring tool, the length of coastlines increased due to the increased ability to measure finer details. This phenomenon highlights the unique nature of fractal measurements, which differ from conventional Euclidean measurements that remain constant regardless of the scale.

For example, Mandelbrot estimated the dimension of Britain’s coastline to be approximately d ≈ 1.25, as illustrated in the box-counting process shown below:

I conducted my own box-counting analysis, resulting in a dimension of approximately d ≈ 1.2458, closely aligning with Mandelbrot's findings.

Another interesting comparison can be made with the coastlines of different countries; for instance, South Africa has a relatively smooth coastline with a dimension of d ≈ 1.05, while Norway, with its intricate fjords, has a dimension of d ≈ 1.52.

Applications of Box-Counting

The box-counting method has numerous variations and can be extended to higher dimensions. Its applications span various fields, including the analysis of river networks, biological structures, geological formations, and even atmospheric phenomena.

In industrial contexts, box-counting assists in examining the microstructures of materials, which is essential for understanding their mechanical properties and quality control processes.

The Koch Curve: A Fractal Example

To conclude, let's briefly explore the Koch curve, developed by Swedish mathematician Niels Fabian Helge von Koch. This fractal involves a continuous, non-differentiable curve constructed by iteratively removing the middle third of a line segment and replacing it with a triangular shape.

The Koch curve illustrates how box-counting can quantify the dimensional complexity of shapes:

Similar to the Sierpinski gasket, the Koch curve also exhibits a dimension greater than one, underscoring its ability to fill space more than a simple line would.

Chapter 2: ARIMA Models and Box-Jenkins Method in Eviews

This video provides a complete guide to understanding ARIMA models and the Box-Jenkins method, demonstrating step-by-step how to utilize these techniques in Eviews for effective time series analysis.

Chapter 3: RNA-Seq Data Processing Workshop

Explore the first part of our R Workshop Series, which delves into the RNA-Seq process from raw data to processed output, providing practical insights into data handling in R.