MathGroup Archive 1996

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: [Q] Programming & "Not Floating Point Number" Error

Noel T. Doromal wrote:
> Hello out there!  This is my first time on this group, so I apologize for
> breaking any netiquette rules.
> I have a Mma question that I hope someone out there can help me with.  I
> recently started using Mma for a class and have had a hard time
> "programming" with it. Here is the script/program that i wrote:
> --
> epsilon = .01
> gamma = 7
> dt = .125
> stim = 0
> y0 = -.5
> w0 = 0.0
> For[i=0, i<50, i++,
> f = y0 * (1 - y0 *y0);
> v = y0 + (f - w0) * dt;
> w = w0 +(epsilon*(gamma*y0 - w0)) * dt;
> vee[i] = v;
> doubleu[i] = w;
> y0 = v;
> w0 = w;
> ]
> ListPlot[{vee,doubleu}]
> ---
> First question:  Are there examples of Mma programs out there that are
> longer than a line or two?  (Or was mathematica not really made for
> programming?)  I bought the book and can't find any in there.  (It
> describes the control structures: do, if, for... but then gives one line
> examples only)
> Second: In the script above I'm essentially calculating a bunch of points
> that I want to plot (V vs. W - I had to use "vee" and "doubleu" to get it
> to work).  The only way I could figure to do it was to store the numbers
> in a list and then use ListPlot[] but I can't get it right.  I keep
> getting:
> --
> Graphics::gptn:
>    Coordinate vee in {1, vee} is not a floating-point number.
> Graphics::gptn:
>    Coordinate doubleu in {2, doubleu} is not a floating-point number.
> --
> What am I doing wrong?
The basic thing you are doing wrong is that you are not making a
list--you are making vee and doubleu as subscripted variables, which can
alternatively be viewed as functions with values at 0, 1, 2, etc.  

The for loop you set up is fine--all you need to do to see your result
is to make a List. The easiest way is with Table.  If you replace your
ListPlot command with


you'll get the plot you desire.

However, there are much easier ways to achieve what you desire with
Mathematica.  For example, assuming that your constants epsilon, gamma,
and dt are defined the same,

y0 = -.5;
w0 = 0.0;

		ytmp = y;
		f = y0 * (1 - y0 *y0);
		v = y0 + (f - w0) * dt;
		w = w0 +(epsilon*(gamma*y0 - w0)) * dt;
		{y0,w0} = {v,w},

gives exactly the same thing.

If your goal is simply to get a numerical approximation to the solution
of the differential equation, the easiest way (and far more accurate and
fast that Euler's method) is

	v'[t] == v[t]*(1-v[t]^2) - w[t], v[0] == -.5,
	w'[t] == epsilon*(gamma*v[t] -w[t]), w[0] == 0.},

the solution can tehn be plotted by:

ParametricPlot[Evaluate[{v[t],w[t]} /. %],{t,0,50*.125}];

My guess is this might not help you much because your assignement may
have been to write a program to compute the solution using Euler's
method?  Well, here is where the power of Mathematica comes in--you can
easily write a program which will compute the solution of any equation
using Euler's method:

Eulers[f_,init_,vars_,{t_,tmin_,tmax_,dt_}] := 
Module[{vals = init, time},
	Table[vals = N[vals + dt*
	 (f /. 	{t->time,vars[[1]]->vals[[1]],vars[[2]]->vals[[2]]})],

f is the lhs of the ODE, init are the initial values, vars are the
dependent variables, t is the time variables which is to go between tmin
and tmax in steps of dt.  With this definition, 


Gives the same plot as you could get via your For loop.  You can try any
system, however.  (This is a crude definition because the is no error

To make it even more general, you don't need to restrict the number of
dependent variables to two:  the following definition will take a system
of any numebr of equations:

Eulers[f_,init_,vars_,{t_,tmin_,tmax_,dt_}] := 
Module[{vals = init, time},
	Table[vals = N[vals + 
		dt*(f /. Join[{t->time},Thread[vars->vals]])],

then, for example,


Produces a list of values for the scalar equation x' = x.

{{1.1}, {1.21}, {1.331}, {1.4641}, {1.61051}, 
  {1.77156}, {1.94872}, {2.14359}, {2.35795}, 
  {2.59374}, {2.85312}}

(Note that this list can be made a simple list of numbers using Flatten)
Rob Knapp
Wolfram Research, Inc.


  • Prev by Date: Simplifying Expressions containing SQRT
  • Next by Date: Re: [Q] Programming & "Not Floating Point Number" Error
  • Previous by thread: [Q] Programming & "Not Floating Point Number" Error
  • Next by thread: Re: [Q] Programming & "Not Floating Point Number" Error