Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Parametrization of functions

  1. Nov 18, 2008 #1
    could someone explain to me what exactly parametrization of a function in more than parameter means?
    so i know that for f(x)=x^2
    there are two parameters,

    but how does that lead to a circle being

    what does this actually mean??
    i.e. in the first example, i get it, in the second, i really don't.

    is there a general way to take a parametrization and turn it into a "regular" function?
  2. jcsd
  3. Nov 19, 2008 #2


    User Avatar
    Science Advisor
    Homework Helper

    Hi soandos! :smile:

    Parametrization means that instead of y = f(x), we use another variable t (the parameter), and write x = g(t), y = h(t).

    For example, a circle can be y = ±√(1 - x2),

    or it can be x = cos(t), y = sin(t).

    One parameter means one dimension (a curve), two parameters means two dimensions (a surface), and so on. :smile:
    No, that's not a parametrisation.
    sorry, but a circle isn't that … that's just wrong …

    see the equations above.
  4. Nov 19, 2008 #3
    sorry for all the errors, but how is it ploted in terms of two functions?
  5. Nov 19, 2008 #4


    User Avatar
    Science Advisor
    Homework Helper

    Two parameters would give you a surface …

    for example, a sphere z = ±√(1 - x2 - y2) could be described by using the usual latitude and longitude as parameters:

    x = cosθcosφ, y = cosθsinφ, z = sinθ. :smile:
  6. Nov 19, 2008 #5
    sorry, i meant that with one parameter what is actually happening,
    ex: Sin(t),Cos(t) plots as a circle. where are the points coming from?

    are values of t plugged in and then both Sin(x) and Cos(x) plotted?
    because that does not give a circle.
    Last edited: Nov 19, 2008
  7. Nov 19, 2008 #6


    User Avatar
    Science Advisor
    Homework Helper

    Yes, you plug in values of t, and plot sin(t) and cos(t) …

    and that must be a circle, because sin2t +cos2t = 1.
  8. Nov 19, 2008 #7
    I must be missing something because if i plot y=sin(x) and y=Cos(x) on the same set of axes i do not get a circle
  9. Nov 19, 2008 #8


    User Avatar
    Science Advisor
    Homework Helper

    What are you doing? :confused:

    y = sin(t), x = cos(t).
  10. Nov 19, 2008 #9
    Oh. got it.
  11. Nov 19, 2008 #10
    Any function f can be easily "parametrized" this way by setting:

    y(t) = f(t)
    x(t) = t

    But you can't always "undo" this parametrization. For example, a graph of a circle is not a function (because f(0) would have to take one two values, 1 and -1).

    Parametrization is very useful because it corresponds to what you imagine when you think about a car or a guy walking around on a map. "t" represents time and x(t), y(t) represent his x,y coordinates at time t.
  12. Nov 19, 2008 #11
    Unrelated, I read somewhere that all objects of rotation are easily parametrized. how so?
  13. Nov 19, 2008 #12


    User Avatar
    Science Advisor
    Homework Helper

    Use θ as one of the parameters …

    so lines of constant θ are all the same shape, adn going parallel to the axis of rotation. :smile:
  14. Nov 19, 2008 #13
    Another way of thinking about it is giving a coordinate system to the shape. As a curve is 1-dimensional, you only need one parameter to locate a point on the curve. Ie., that parameter may be distance from some point (an origin) on the curve.
    For example, the curve generated by the graph of y = f(x) = x^3 can be parametrized by letting the point (0,0) be the origin of our 1-dimensional coordinate system and then noting that x(t) = t and y(t) = t^3 when describing the curve using the Cartesian xy-coordinate system.
    Another valid parametrization is to choose the origin arbitrarily, ie., let the point (1,1), the coordinates of the point on the curve with respect to the Cartesian plane, be the point 0 with respect to our 1-dimensional curve coordinate system. Then x(t) = t + 1 and y(t) = (t + 1)^3.
    You are now free to embed curves in any dimension, ie., curves in R^3, which would be rather clumsy to define in terms of functions. Any 1-parameter family of functions x(t), y(t), z(t) should describe a curve, as long as they're well-behaved. Try to describe a right-circular helix about the z-axis. (Hint: x(t) and y(t) should describe a circle, while z(t) just stretches it out. Drawing a picture always helps).
    Surfaces are by definition 2-dimensional objects, so we will need 2 coordinates to locate a point on a surface. We usually go for the variables (u,v). That is to say, if we are talking about the parametrization of a surface as a subset of R^3, we would have 3 functions x(u,v), y(u,v), and z(u,v). Just imagine laying a grid down on the surface you want to parametrize; one of the directions will be the u-direction, the other will be v.
  15. Nov 19, 2008 #14
    OK so the circle part is easy,
    for z(t) doesn't any linear function work (assuming that the space between "coils" remains a constant)?
  16. Nov 19, 2008 #15
    Yep. By varying z(t), you can make the coils do any wacky thing along the z-axis that you want. With 3-d software that renders curves from parametrizations and a little bit of physics knowledge, you should be able to animate a bouncing spring.
    You should even be able to parametrize a closed ring-like helix, where the "axis" is a circle.
    As for surface parametrizations, a simple but non-trivial one would be parametrizing a torus in R^3. One common parametrization would be to let one coordinate (u) represent angle from the center of the torus and let the second coordinate (v) represent angle from the center of the tube of the torus. If we place our torus around the origin in the xy-plane with inner radius r and outer radius R, we then have x(u, v) = ((r + R)/2 + (R - r)sin(v))cos(u) by applying some trigonometry. You may get a slightly different form depending on exactly how you defined your u and v. You can then derive the expressions for y(u, v) and z(u, v).
    For a sanity check, you will want to use some graphing software that accepts parametrizations. http://blog.viamentis.com/articles/2008/01/18/plotting-mathematical-functions-with-gnuplot/ [Broken] raytracer for aesthetic manipulation/animation. If you are already experienced in POV-Ray scripts, you can input parametric equations right into it, and even visualize curves in 3d using a #while loop inside a sphere_sweep object.
    Last edited by a moderator: May 3, 2017
  17. Nov 19, 2008 #16
    Just curious, if a helix is a circle that is "extended" what is a cylinder?
  18. Nov 19, 2008 #17
    A helix is a curve, while a cylinder is a surface. It can only be described by 2 coordinates. One easy parametrization is having height be the second parameter: x(u,v) = sin(u), y(u, v) = cos(u), z(u,v) = v.
  19. Nov 20, 2008 #18
    You'd also need to put restrictions on z, so it isn't infinitely long.
  20. Nov 20, 2008 #19


    User Avatar
    Science Advisor

    Why shouldn't it be infinitely long?
  21. Nov 20, 2008 #20
    Because an infinitely long can of soda would be expensive and would go flat before you ever finished =-)
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook