Tag Archives: Fractal Dimension

Fractals & Chaos Lesson Recap for 10/24

Today, used the Box Count method to find again the dimension of Great Britain (report your findings here) then completed one last project to find calculate the dimension of one of the spiral fractal seen on the last dimension calculation sheet (this took most of the remainder of the period).

For Monday, read pages 83-92 in Fractals: The Patterns of Chaos (about fractal math limitations)

If you’d like to rewatch Adam Neely’s Coltrane Fractal video we saw in class (or check out some of his related videos), click the link.

Fractals & Chaos Recap for 10/23

After discussing the reading from the text and the answer to yesterday’s question of the border between Spain and Portugal, we moved on to the last method of finding dimension, the Box Count method.

This method of finding dimension produces the same table of values and log-log plot that we made with the Richardson plot, but the values of S and C are found using a different method. Imagine overlaying a grid on top of a fractal image. We then count (C) the number of boxes of that grid that contain some portion of the fractal. We then repeat this process using a grid with smaller boxes, the sizes of which relative to the original give us S.

The YouTube channel 3Blue1Brown has a great video summarizing all of this.

After enough counts are collected at different scales of boxes, we can create a log(c) vs log(s) plot and find the dimension using the slope as we did before. Your homework is to make the necessary counts with the coastline of Great Britain.

Also homework: For Monday, read pages 83-92 in Fractals: The Patterns of Chaos (about fractal math limitations)

Fractals & Chaos Recap for 10/22

We continued yesterday’s applications of the Richardson Plot to the Koch Curve and finally to the coastline of Great Britain, largely confirming Richardson’s findings as included in Mandelbrot’s article. The results of these can be found here.

Also, at the end of class today, we discussed the border between Spain and Portugal and looked at three maps.  Take the data below and answer the following questions:

  1. What is the dimension of the border between the two countries?
  2. One country has historically given the length of the border as 987 km, while the other has given a length of 1214 km.  Which country is which, and why might this difference have a logical basis (in other words, why might the countries have truly measured the borders in this way? The answer isn’t political!)
Step Size S C Distance measured
100 km 1 7.3 730 km
50 km 2 16.2 810 km
25 km 4 35.4 885 km
10 km 10 93.2 932 km
5 km 20 200.6 1003 km

Remember: for Friday please read pages 61-73 of your new book Fractals: The Patterns of Chaos

Fractals & Chaos Recap for 10/21

In class today, we derived a new method for finding a dimension of a fractal, the method proposed by Mandelbrot in his article and originally conceived of by mathematician and meteorologist Lewis Fry Richardson.  We observed that if we make a “log-log” plot (a so-called “Richardson plot”) of the step sizes and counts of steps that “fit” in a curve, the distribution of points comes out to a roughly linear association, the slope of which is the dimension of the fractal.

We concluded class today by testing this theory to find the dimension of a circle, the results of which can be found here.

We also got a new book, Fractals: The Patterns of Chaos, and our first reading assignment: pages 61-73 (on fractal dimension)

Fractals & Chaos Recap for 10/16

One more chance to sign up for snacks for tomorrow’s Fractal Art Show!

In class today, we finished up the dimension classwork sheet and practiced how the segments of each fractal design that you measure to calculate its dimension can be used to recreate the fractal in FractaSketch (aka “stealing the template”). Neat!

We also have an important reading assignment: By Friday, October 18, please read Mandelbrot’s revolutionary paper that sparked the recognition of fractals and fractal geometry How Long is the Coast of Britain? You might also want to read this version, where Mandelbrot himself explains how he originally wrote this paper as a “Trojan Horse” to introduce his vision of fractal dimension into the scientific community conversation.

Fractals & Chaos Recap for 10/15

Please sign up for snacks for this week’s Fractal Art Show on Thursday, October 17.

In class today, we worked on our fourth and final dimension classwork sheet (while a handful of stragglers turned in their Fractal Art Show designs) and looked at how the analysis we do to find the dimension of the shapes can be used to recreate them in FractaSketch. Neat!

We also have an important reading assignment: By Friday, October 18, please read Mandelbrot’s revolutionary paper that sparked the recognition of fractals and fractal geometry How Long is the Coast of Britain? You might also want to read this version, where Mandelbrot himself explains how he originally wrote this paper as a “Trojan Horse” to introduce his vision of fractal dimension into the scientific community conversation.

Fractals & Chaos Recap for 10/9

We spent some more time in class working on practicing our Generalized Hausdorf Dimension formula on the second batch of fractals you got on Monday, then started in on a third batch. We’ll spend tomorrow working on our FractaSketch designs (remember that they are due on Friday!)

I would also like you to read Pollack’s Fractals, an article from Discover Magazine about the math underlying Jackson Pollack’s famous paintings.  For some interesting follow-up reading, check out this article from the New York Times about the use of fractal analysis to examine the authenticity of supposed Pollack paintings and this article from the Science Daily blog suggesting that such an analysis is not scientifically valid.

Fractals & Chaos Lesson Recap for 10/7

We started class today with a quick visit to H Courtyard to observe and analyze the dimension of the fractals found there, then came back to class and practiced our Generalized Hausdorf Dimension formula on a new batch of fractals. We’ll spend tomorrow working on our FractaSketch designs (remember that they are due on Friday!)

Tonight, take a look at this article from Nature: Fractal secrets of Rorschach’s famed ink blots revealed

Fractals & Chaos Recap for 10/1

We had a brief discussion on Stephen Jay Gould’s essay “Size and Shape,” seeing some common threads with prior reading assignments. For one last follow-up, I recommend this NPR piece from 2007 – Size Matters: The Hidden Mathematics of Life (and/or this followup to the  Kurzgesagt video from last week).

From there, we moved on to revisit the problems with the current Hausdorff Dimension formula that we identified yesterday, specifically about what to do with fractals that are self-symmetric, but at inconsistent scales. We developed a new, Generalized Hausdorff Dimension formula:

Annotation 2019-10-01 084135
Generalized Hausdorff Dimension formula

Here, S_1, S_2, …, S_N are all the differing ratios each iterating piece form with the original whole (note that here, each S is a fraction, so a piece that is 1/2 the length of the whole has S = 1/2 instead of S = 2 as it was in the previous definition)

We then practiced this definition with a few more examples, found on this sheet.

Fractals & Chaos Recap for 9/30

I decided to postpone our discussion of Gould’s essay, assigned on Friday, so that we could finish our discussion of the first dimension calculation practice sheet, which we did so with an observation that the S^d = N definition for dimension has some weaknesses. In particular:

  • What do you do with fractals with non-iterating stems (e.g., 2 and 13)?
  • What if the fractal is not exactly self-similar (like 5)?
  • What if it is self-similar, but at differing scales (like 15)?

We will continue to work on our definition of dimension to accommodate these issues. You have also been issued a challenge: use what you’ve learned from the S^d=N definition to design a fractal with a dimension of exactly 1.5.

We spent the rest of the period working on our FractaSketch designs.

Fractals & Chaos Recap for 9/27

We started class with a writing/discussion prompt based on Haldane’s On Being the Right Size, a response to the meme “Would you rather fight one horse-sized duck or 100 duck-sized horses?” The horse-sized duck won the argument, as an application of the central takeaway from Haldane’s article is that artificially scaling a duck up to the size of a horse would increase its weight exponentially faster than its strength could accommodate, resulting in a duck incapable of walking around, let alone flight. We followed this up with a relevant Kurzgesagt (In a Nutshell) video: What Happens if we Throw and Elephant From a Skyscraper (as promised, the follow up — How to Make an Elephant Explode — can be found here).

In the second half of the period, we continued our discussion of the dimension of the various fractals found on our first dimension practice sheet. We found that one of the templates was a cleverly disguised version of the Sierpinski Carpet, and that another–Sierpinski’s Pyramid–has the interesting property of being simultaneously defined as having 1, 2, or 3 dimensions, depending on what type of dimension you use.

Monday, we go back to the computer lab to work on our fractal designs. By Tuesday, read Size and Shape, paleontologist and evolutionary biologist Stephen Jay Gould‘s follow up to Haldane’s essay. If Gould’s name sounds familiar, it’s because he co-developed the idea of (and coined the term for) punctuated equilibrium, a theory of evolution.

Fractals & Chaos Lesson Recap for 9/23

We wrapped up Thursday’s lesson with an explanation of the idea of the Hausdorff Dimension of a fractal. In brief, the Hausdorff Dimension is the solution to the equation S^d = N, where S is the scale by which a fractal is being broken up into pieces, and N is the number of such pieces. So for the Sierpinski Triangle, we cut the original triangle into pieces that are all half the length of the original (S = 2), but keep 3 of those pieces. Solving the equation 2^d = 3 gives d ≈ 1.585, which is the Hausdorff dimension of the Sierpinski Triangle. For the Koch Curve, we get the equation 3^d = 4, producing d ≈ 1.26, and for the Cantor Set, we get 3^d = 2, so d ≈ 0.631.

For homework, you’ve been asked to think about the dimension of the Dragon Curve, and have been reminded that the base template for the curve involves a right isosceles triangle, as well as to design a fractal called the Sierpinski Carpet (essentially the same design as the Sierpinski Triangle, but with a square as the starting shape) and to find the dimension of that.

Fractals & Chaos Recap for 9/19

Today was an important day.

We continued a thought started yesterday, acknowledging the difference between intrinsic dimension—the dimension of an object itself—and extrinsic dimension—the dimension of the space the object occupies. For example, the edge of a circle is one dimensional, since there is only one axis of movement around the circle, but the circle exists in two dimensional space. We further observed that the extrinsic/containing dimension of an object must necessarily be greater than or equal to the intrinsic dimension of the space.

We also discussed how the dimension of an object affects the dimension with which we can measure it. We cannot find the volume of a square, and we can’t find the area of a cube. Put another way, if we tried to use a cubic centimeter measuring device to measure the area of a circle, we’d end with a measure of zero. And if we tried to use a square centimeter measuring device to measure the volume of a cube, we’d get a measure of infinity. In general, too small a dimension of measure results in a measure of infinity, and too large a dimension results in zero.

Finally, we considered the implications of this to the Sierpinski Triangle. When we attempted to measure the area of the Sierpinski triangle, we resulted in a measure of zero (an infinity of triangles each with area zero results in a total area of zero). When we attempted to measure its length, we resulted in a measure of infinity (an infinity of segments each with length zero results in a total length of infinity). Thus, 2 is too high a dimension to describe the Sierpinski Triangle and 1 is too low. It must therefore mean that the dimension of the Sierpinski Triangle is strictly between 1 and 2, and is therefore fractional.


Of course, this raises an immediate question: If the dimension of the Sierpinski Triangle is somewhere between 1 and 2, what is it?

Fractals & Chaos Lesson Recap for 9/17

Today something weird happened.

We’ve already seen the Sierpinski Triangle appear in two ways, by iterating the process of breaking a triangle into four equal pieces and removing the central one, and by coloring the even elements of Pascal’s triangle. Today we found a third way, by Iterating a Curve.

By starting with a solid, filled in triangle, and step-by-step removing stuff from that area, we reach the same exact figure as we get by starting with a simple line segment and lengthening/bending it. The area clearly starts as 2-dimensional, the line segment clearly starts as 1-dimensional. Yet both processes have the same end result. What does this mean for the dimension of Sierpinski’s Triangle? Is it 2D or 1D? It certainly can’t be both, so the only option is… neither?

When I first took this course (back in the days of Mr. Drix), this was the first moment where I realized all of this talk about fractional dimension may be more than nonsense. Maybe there’s something to it after all…

Your homework: think critically about what dimension really means. What does it mean to say something is 1-dimensional or 2-dimensional. We’ll dig into this in the next few days…