Lissajous Curve

From Math Images

Revision as of 22:26, 17 June 2012 by Pzhao1 (Talk | contribs)
Jump to: navigation, search
Lissajous Box
Field: Geometry
Image Created By: Michael Trott
This image is protected by copyright. Click here to
ask for permission before using it elsewhere!

Lissajous Box

This is a beautiful Lissajous Box. The Lissajous Curves on its sides have an angular frequency ratio of 10:7.


Basic Description

Lissajous Curves, or Lissajous Figures, are patterns formed when two harmonic vibrations along perpendicular lines are superimposed. The parametric equations of Lissajous Curves have the following form:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(a*t + \phi) \\ y & \mbox{=} & \sin(b * t) \end{array}\right.

in which A and B are magnitudes of two harmonic vibrations, a and b are their angular frequencies, and \phi is their phase difference.

For example, let's see the case A = B = 10, a = 2, b = 3, and \phi = 0:

In the animation above, points X and Y are simple harmonic oscillators in perpendicular directions. They follow the equations of motion x = \sin(3t), and  y = \sin(2t), respectively. When t goes from 0 to 2\pi, the x - vibration completes 3 cycles, and the y - vibration completes 2.

However, what we really care about is the superposition of these two vibrations. To get this superposition, we can draw from X a line perpendicular to x-axis, and from Y a line perpendicular to y-axis, and get their intersection P. By simple geometry, P will have the same x-coordinate as X, and y-coordinate as Y, so it combines the motion of X and Y. As we can see in Figure 1, the trace of P turns out to be a complicated and beautiful curve, which we refer to as a "Lissajous Curve". More specifically, it's the Lissajous Curve with frequency ratio a / b = 3 / 2 and phase difference \phi = 0, since we can easily generate more Lissajous Curves with other angular frequencies and phases using the same mechanism.

The following image shows some of these Lissajous Curves:

Figure 2-aLissajous Curve: a/b = 1/2
Figure 2-a
Lissajous Curve: a/b = 1/2

Figure 2-bLissajous Curve: a/b = 3/4
Figure 2-b
Lissajous Curve: a/b = 3/4

Figure 2-cLissajous Curve: a/b = 5/4
Figure 2-c
Lissajous Curve: a/b = 5/4

To see exactly what determines the appearance of Lissajous Curves, please go to the More Mathematical Explanation section.

A Dip Into the History

Figure 3-aPhotograph of Joules Lissajous. Year and Photographer Unknown
Figure 3-a
Photograph of Joules Lissajous. Year and Photographer Unknown

Lissajous Curves were named after French mathematician Jules Antoine Lissajous (1822–1880)[1], who devised a simple optical method to study compound vibrations. Lissajous entered the Ecole Normale Superieure in 1841, and later became a professor of physics at the Lycee Saint-Louis in Paris, where he studied vibrations and sound.

During that age, people were enthusiastic about standardization in science, and the science of acoustics was no exception, since musicians and instrument makers were crying out for a standard in pitch. In response to their demand, Lissajous invented the Lissajous Tuning Forks, which turned out to be a great success since they not only allowed people to visualize and analyse sound vibrations, but also showed the beauty of math through interesting patterns.

The structure and usage of Lissajous Tuning Forks are shown in Figure 3-b. Each tuning fork is manufactured with a small piece of mirror attached to one prong, and a small metal ball attached to the other as counterweight. Two tuning forks like this are placed besides each other, oriented in perpendicular directions. A beam of light is bounced off the two mirrors in turn and directed to a screen. If we put a magnifying glass between the second tuning fork and the screen (to make the small deflections of light beam visible to human eyes), we can actually see Lissajous Curves forming on the screen.

Figure 3-bDomonstration of Lissajous Tuning Forks
Figure 3-b
Domonstration of Lissajous Tuning Forks

The idea of visualizing sound vibrations may not be surprising nowadays, but it was ground-breakingly new in Lissajous' age. Moreover, as we are going to see in the More Mathematical Explanation section, the appearance of Lissajous Curves are extremely sensitive to the frequency ratio of tuning forks. The most stable and perfect patterns only appear when the two forks vibrate at frequencies of simple ratios, such as 2:1 or 3:2. These frequency ratios correspond to the musical intervals of the octave and perfect fifth, respectively. So, by observing the Lissajous Curve formed by an unadjusted fork and a standard fork of known frequency, people were able to make tuning adjustments far more accurately than tuning by ear.

Because of his contributions to acoustic science, Lissajous was honored as member of a musical science commission set up by the French Government in 1858, which also featured great composers such as Hector Berlioz (1803-1869) and Gioachino Rossini (1792-1868)[2].

A More Mathematical Explanation

In previous sections, we have encountered this problem for many times:

  • What determines the appear [...]

In previous sections, we have encountered this problem for many times:

  • What determines the appearance of Lissajous Curves?

In this section, we are going to answer this question in two ways. The first method is simple and direct, but is limited to several special cases. The second one applies to almost all Lissajous Curves, but as a result it's more subtle and complicated.

First Method: Direct Elimination of t

Since Lissajous Curves are defined by the following parametric equations:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(a*t + \phi) \\ y & \mbox{=} & \sin(b*t) \end{array}\right.

In principle, one can use trigonometric formula to eliminate t from these equations, and get a relationship between x and y. See the following examples:

(Note: in all examples below, we are going to assume that A = B = 1, since changing these magnitudes will only make the curves dilate or contract in horizontal or vertical direction. They don't affect the structure of Lissajous curves.)

Example 1: line segment

Figure 3-aLissajous Curve 1: Line Segment
Figure 3-a
Lissajous Curve 1: Line Segment

If in addition to A = B = 1, we have a = b = 1, and \phi = 0, then the parametric equations will become:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(t) \\ y & \mbox{=} & \sin(t) \end{array}\right.

from which we can easily get:

x = y

Moreover, since the range of \sin(x) is from -1 to 1, we have:

-1 \leq x \leq 1

Together, they give us the line segment shown in Figure 3-a.

Example 2: circle

(Starting in this example we will use some trigonometric formula to help us reduce the equations. These formula, together with some explanations, can be found here[3].)

Figure 3-bLissajous Curve 2: Circle
Figure 3-b
Lissajous Curve 2: Circle

In this case, we still have a = b = 1. But instead of \phi = 0, we change it to \pi \over 2:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(t + {\pi \over 2}) \\ y & \mbox{=} & \sin(t) \end{array}\right.

Using the trigonometric identity sin(t + {\pi \over 2}) = \cos(t), we will get:

\left\{ \begin{array}{rcl} x & \mbox{=} & \cos(t) \\ y & \mbox{=} & \sin(t) \end{array}\right.

Using the trigonometric identity \sin^2(\theta) + \cos^2(\theta) = 1, we will get:

x^2 + y^2 = 1

Which gives us the circle shown in Figure 3-b.

Example 3: parabola

Figure 3-cLissajous Curve 3: Parabola
Figure 3-c
Lissajous Curve 3: Parabola

This time, we change the parameters into a = 1, b = 1, and \phi = {\pi \over 4}, then the parametric equations will become:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(t + {\pi \over 4}) \\ y & \mbox{=} & \sin(2t) \end{array}\right.

from Eq.1 we can get:

2x^2 - 1 = 2\sin^2(t + {\pi \over 4}) - 1

Using the trigonometric identity cos(2\theta) = 1 - 2 \sin^2(\theta), we can get:

2x^2 - 1 = - \cos (2t + {\pi \over 2})

Apply the formula \cos(\theta + {\pi \over 2}) = - \sin(\theta), we can get:

2x^2 - 1 = \sin(2t)

Combine it with Eq.2: y = \sin(2t), we can get:

 y = 2x^2 - 1

with x confined between -1 and 1. This gives us the parabola in Figure 3-c.

Conclusion: pros and cons

In the examples above, we can clearly see some advantages of the direct elimination method: it's clear, accurate, and easy to understand. However, these advantages are quickly shadowed by the complexity of calculation when we get to larger frequency ratios. For example, see the following parametric equations of a Lissajous Curve:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(9t) \\ y & \mbox{=} & \sin(8t) \end{array}\right.

In principle, this could be solved by expanding the x- and y- function into powers of \sin(t) and \cos(t):

x = \sin 9t = \sin^9 t - {9\cdot8 \over 2!}\sin^7 t\cos^2 t + {9\cdot8\cdot7\cdot6 \over 4!}\sin^5 t \cos^4 t -{9\cdot8\cdot7\cdot6\cdot5\cdot4 \over 6!}\sin^3 t\cos^6 t + {9! \over 8!}\sin t\cos^8 t

y = \sin 8t = \sin^8 t - {8\cdot7 \over 2!}\sin^6 t\cos^2 t + {8\cdot7\cdot6\cdot5 \over 4!}\sin^4 t \cos^4 t -{8\cdot7\cdot6\cdot5\cdot4\cdot3 \over 6!}\sin^2 t\cos^6 t + {8! \over 8!}\cos^8 t

These equations come from a general formula given in 16th century by French mathematician Vieta. For more information, please go to this page.

Notice that in these equations, if we consider \sin t and \cos t as unknowns, then we will have a set of two polynomial equations with two unknowns, and in principle we can solve \sin t and \cos t in terms of x and y. Then, the identity {\sin^2 t} + {\cos^2 t} = 1 will give us a direct relationship between x and y. However, in practice, few people are willing to carry on with the algebra, because the calculations involved are just so cumbersome and annoying. To make things worse, as group theory tells us, not all polynomial equations of powers higher than 5 can be solved with exact expression of roots [4]. So there is no guarantee that our effort will lead us to the answer. Even if they can, the relationship between x and y is going to be too complicated to tell us anything useful about the shape of the curve. So the method of elimination fails here, and we would like a new way to study these curves.

Second Method: Experiment and Observation

As shown in the previous discussion, the attempt to directly solve Lissajous Curves failed when we try to deal with large angular frequencies, so we have to find another way to study them. One such way is to study them through experiment and observation. That is, we can use computer software to draw some Lissajous Curves with different parameters, and see how they affect the appearance of Lissajous Curves.

When we study something with multiple variable parameters, it's much easier to study these parameters separately, rather than together. There are three variable parameters, a, b, and\phi, in Lissajous Curves. So in the rest of this section we will fix the phase difference \phi to study the angular frequencies a and b, and then fix the angular frequencies to study the phase difference.

Study a and b with \phi fixed

The following table shows the Lissajous Curves:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(a*t + \phi) \\ y & \mbox{=} & \sin(b*t) \end{array}\right.

with angular frequencies a and b varying from 1 to 5, and phase difference \phi fixed at 0:

Figure 4-aA table of Lissajous Curves with different angular frequency ratios
Figure 4-a
A table of Lissajous Curves with different angular frequency ratios

There are many interesting properties associated with this table:

1. All Lissajous Curves in the table are confined in a 2 * 2 square box. The curves can touch, but cannot go beyond, the lines x = 1, x = –1, y = 1, and y = –1, because the amplitudes of both horizontal and vertical vibrations are set to 1.

2. The Lissajous Curve with a = b = 1 is identical to the curves with a = b = 2, a = b = 3 ... Similarly, the Lissajous Curve with a = 1, b = 2 is identical to the curve with a = 2, b = 4, as shown in Figure 4-b. In other words, the only thing that matters is the ratio between a and b. It can be shown that Lissajous Curves with the same angular frequency ratio must have the same appearance. For example, if we do the substitution t = 2u in the Lissajous Curve with a = 1 and b = 2:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(t) \\ y & \mbox{=} & \sin(2t) \end{array}\right.

we will get:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(2u) \\ y & \mbox{=} & \sin(4u) \end{array}\right.

which is nothing different from the Lissajous Curve with a = 2 and b = 4, because whether we use the symbol t or u doesn't matter here. This analysis can be generalized to all Lissajous Curves with rational frequency ratios.

Figure 4-bProperty #2
Figure 4-b
Property #2

3.The Lissajous Curve with a = 1 and b = 2 is the reflection of the Lissajous Curve with a = 2 and b = 1 about line y = x, as shown in Figure 4-c. In fact, if we exchange the values of a and b in a Lissajous Curve, the result will be the original curve "flipped" about line y = x. To prove this, let's see the Lissajous Curve:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(at) \\ y & \mbox{=} & \sin(bt) \end{array}\right.

if we replace a with b, and b with a, we will get the following Lissajous Curve:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(bt) \\ y & \mbox{=} & \sin(at) \end{array}\right.

However, the same resulting curve could also be achieved by replacing x with y, and y with x in the original curve. In other words, the exchange of a and b is equivalent to the exchange of x and y. Moreover, in Cartesian coordinates, exchanging x and y in the equation of the curve is equivalent to flipping the curve about line y = x. So exchanging a and b is also equivalent to flipping about line y = x.

Figure 4-cProperty #3
Figure 4-c
Property #3

From these properties, we can see that many Lissajous Curves with different angular frequencies are actually the same thing, and we do not need to study all of them. In fact, we can use the following family to represent all Lissajous Curves with rational frequency ratios:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(rt) \\ y & \mbox{=} & \sin(t) \end{array}\right.

in which r is a rational number standing for angular frequency ratio. The argument for this goes as following: for any Lissajous Curve

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(at) \\ y & \mbox{=} & \sin(bt) \end{array}\right.

in which a and b are integers, we can assume that  a \leq b, since if a > b we can exchange their values, and according to property #3 the curve will only be flipped about line y = x. This doesn't affect the curve's structure, which is what we really care about.

The next step is to divide both angular frequencies by b. According to property #2, the curve will not change, and we will get:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin({a \over b}t) \\ y & \mbox{=} & \sin(t) \end{array}\right.

The last set of parametric equations belongs to the family we mentioned above. So we only need to study this family of Lissajous Curves, since others can all be reduced to this case.

The following animation shows some of the Lissajous Curves in this family, with the frequency ratio a / b varying continuously from 0 to 1:

Click to stop animation.

Figure 4-d
Lissajous Curves with varying frequency ratio
Click to stop or replay animation

Surprisingly, as we can see in the animation, most of these Lissajous Curves are rather convoluted. But there are some simple and beautiful ones scattered in them. A more careful examination shows that, when these simple patterns occur, the frequency ratio must be equal to a simple fraction. This phenomenon is not hard to understand if we look at the generating process of Lissajous Curves once again. Suppose the two component vibrations start at t = t0. As long as the frequency ratio is rational, the moving point will eventually return to its starting place, and make a closed Lissajous Curve. Suppose this happens at t = t1. So the time period between t0 and t1 is a complete cycle of this Lissajous Curve. Moreover, since the starting point and ending point overlap, we must have:

\left\{ \begin{array}{rcl} x (t_0) & \mbox{=} & x (t_1) \\y(t_0) & \mbox{=} & y(t_1) \end{array}\right.

Substitute into the parametric equations of Lissajous Curve with rational frequency ratio, we can get:

\left\{ \begin{array}{rcl} \sin({a \over b}t_0) & \mbox{=} & \sin({a \over b}t_1) \\ \sin(t_0) & \mbox{=} & \sin(t_1) \end{array}\right.

which leads to:

\left\{ \begin{array}{rcl} {a \over b}(t_1 - t_0) & \mbox{=} & 2k_1 \pi ---------------- Eq.1 \\ (t_1 - t_0) & \mbox{=} & 2k_2 \pi ---------------- Eq.2 \end{array}\right.

in which k1 and k2 are integers. Substitute Eq.2 into Eq.1, we get:

{a \over b} = {k_1 \over k_2}

since a / b is assumed to be an irreducible fraction (if not, we can divide them by their common factor without changing the Lissajous Curve), the smallest k1 and k2 that satisfy this equation are k1 = a and k2 = b. Substitute back into Eq.2, we can get:
(t_1 - t_0) = 2b \pi

So the larger b is, the longer it's going to take before the Lissajous Curve closes and repeats itself, and the more convoluted it's going to be. For a simple angular frequency ratio like 1/2, the vibrations soon start to repeat, and the Lissajous Curve is simple, as shown in the previous table. However, a ratio like 37/335 will make the curve much more complicated. In an extreme case, if the ratio is irrational, then both a and b will be infinitely large, and the Curve is no longer closed. This special case is treated later in this section. Click here to see.

In conclusion, the angular frequency ratio a / b, reduced to simplest fraction, determines the complexity of Lissajous Curves. Large a and b lead to complicated Lissajous Curves; small a and b give us simple ones. This is why Lissajous Tuning Forks are so suitable for tuning notes. In music theory, the interval of a perfect octave is 1:2, perfect fifth is 2:3, perfect fourth is 3:4, and so on[5]. These intervals all correspond to simple Lissajous Curves with distinctive features.

Study \phi with a and b fixed

In the previous subsection we have figured out how angular frequencies of a Lissajous Curve affect its appearance. Now we are going to fix the angular frequencies to study the third, and last, variable parameter: the phase difference \phi.

The following animation shows the Lissajous Curve

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(t + \phi) \\y & \mbox{=} & \sin(3t) \end{array}\right.

with \phi varying continuously from 0 to 2\pi:

Click to stop animation.

Figure 5-a
Lissajous Curve with varying phase difference \phi
Click to stop or replay animation

An interesting fact to notice is that, the animation above looks more like a rotating 3-D curve, rather than a changing 2-D one. The reason for this illusion is related to another way to define Lissajous Curves. In the beginning of this page, we introduced the following definition:

Lissajous Curve is the superposition of two harmonic vibrations.

However, this is not the only definition for Lissajous Curves. These curves can also be viewed as the projection of a 3-D harmonic height function over a circular base. The following set of images explain this definition in more detail:

Figure 5-bCircular base of harmonic height function
Figure 5-b
Circular base of harmonic height function

Figure 5-cRaising process
Figure 5-c
Raising process

Figure 5-dProjection onto y-z plane
Figure 5-d
Projection onto y-z plane

The first step to generate this harmonic height function is to draw a circular base in x-y plane, as shown in Figure 5-b. The parametric equation of this circular base is:

\left\{ \begin{array}{rcl} x & \mbox{=} & \cos(t + \phi) \\ y & \mbox{=} & \sin(t + \phi) \end{array}\right.

The variable parameter \phi here doesn't change the shape of the circle, as we still have the relationship x^2 + y^2 = 1. But if we change the value of \phi, then the circle will rotate about the origin O. Of course we can't see the motion here, because O is also the center of the circle. However, this rotation is going to make a difference later.

In the next step, we raise (or lower) each point in the circular base to a certain height. This height is determined by the function:

z = \sin(3t)

The raising process is shown in Figure 5-c. Note that if we change \phi now, the rotation is visible, since the curve's rotational symmetry is broken in the raising process.

Finally, if we make the projection of that rotating height curve onto the y-z plane, as shown in Figure 5-d, we can see that it's exactly same as the animation in Figure 5-a. In other words, this Lissajous Curve can be viewed as the projection of this 3-D height function. Changing the value of \phi makes the 3-D curve to rotate, and in turn changes the 2-D curve. In fact, this is why we had the 3-D illusion in Figure 5-a.

Algebraic analysis agrees with this result. As we have seen, the parametric equations of this harmonic height function are:

\left\{ \begin{array}{rcl} x & \mbox{=} & \cos(t + \phi) \\ y & \mbox{=} & \sin(t + \phi) \\ z & \mbox{=} & \sin(3t)\end{array}\right.

To project it onto the y-z plane, we can fix its x component to be 0:

\left\{ \begin{array}{rcl} x & \mbox{=} & 0 \\ y & \mbox{=} & \sin(t + \phi) \\ z & \mbox{=} & \sin(3t)\end{array}\right.

Compare this projection to the Lissajous Curve we had before:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(t + \phi) \\y & \mbox{=} & \sin(3t) \end{array}\right.

we can see that they are indeed the same thing.

Although we used a special case a = 1, b = 3 in our discussion, the result applies to all Lissajous Curves with rational frequency ratios. The following images show the 3-D height function of some other Lissajous Curves:

Figure 5-ea = 2, b = 3,  = 0
Figure 5-e
a = 2, b = 3, \phi = 0

Figure 5-fa = 3, b = 5,  = /10
Figure 5-f
a = 3, b = 5, \phi = \pi/10

To put it all together: a java applet

So far we have talked much about the appearance of Lissajous Curves. We know that some simple cases can be solve by direct elimination of t in the parametric equations, that the frequency ratio of a Lissajous Curve determines its complexity, and that the phase difference \phi affects a Lissajous Curve by rotating its corresponding 3-D height function. Here is an interactive java applet that puts these all together. It allows the user the change both angular frequencies from 1 to 9, and animate the curve by changing \phi[6]:

What happens when things get irrational?

We have limited previous discussions to Lissajous Curves with rational frequency ratios. So, one may naturally wonder, what happened to all those with irrational frequency ratios? Well, they have all died painfully because of their irrationality ...

Just kidding. They are still there, waiting for us to study. For example, see the following Lissajous Curve:

\left\{ \begin{array}{rcl} x & \mbox{=} & \sin(2t) \\y & \mbox{=} & \sin(\pi t) \end{array}\right.

It's a known fact that \pi is irrational. So the frequency ratio 2 \over \pi here is also irrational, and this curve is going to be radically different from any one we have encountered so far. See the animation below to get a sense of what it looks like:

Click to stop animation.

Figure 6-a
Lissajous Curve with irrational angular frequency ratio
Click to stop or replay animation

Figure 6-a shows the trace this Lissajous Curve in accelerating motion. In the beginning, it looks just like an ordinary Lissajous Curve. However, soon we can see that the

Why It's Interesting


How the Main Image Relates


Teaching Materials

There are currently no teaching materials for this page. Add teaching materials.

Related Links

Additional Resources



  1. Jules Antoine Lissajous, from Wikipedia. This is a biography of Jules Lissajous, discoverer of Lissajous Curves.
  2. Lissajous tuning forks: the standardization of musical sound from Wipple Collections. This is a brief introduction to Lissajous' Tuning Forks and his contribution in acoustic science.
  3. List of Trigonometric identities, from wikipedia. This page lists some of the trigonometric formula we used the derive the shape of Lissajous curves.
  4. Polynomial, from wikipedia. This briefly explains why we can't find a general solution for equations of powers higher than 5.
  5. Interval (music), from Wikipedia. This article explains more about musical notes and their frequency intervals.
  6. Animated Lissajous figures.This is the source of the embedded java applet.

Future Directions for this Page


If you are able, please consider adding to or editing this page!

Have questions about the image or the explanations on this page?
Leave a message on the discussion page by clicking the 'discussion' tab at the top of this image page.

Personal tools