error: Math script code giving only zeros..

[Kris Matrix]

I have this code . I know it works in MATLAB. I tried to make it work in MathScript but in the last step, i only get M to be a zero matrix. however it should have values then. What is wrong??

t = (0:0.0000001:.01);

f1 = .0011312*cos(2*pi*49400*t);

f2 = .0033936*cos(2*pi*100*t);

H = (f1 + f2);

x = ((2.93464*10^-10)*H);

ms=1000;

M = ms * (coth(x) - (1./x));

[Jarrod_S.]

I think someone suggested entering the code into the Mathscript Window line-by-line, so that you can see the progression of your variables. For a small script like this, it's a great way to  debug it.

[Kris Matrix]

I have input the code one at a time. It shows values of all them except M. I get H, x, ms, everything except M.

[smercurio_fc]

I tried your script in Matlab 6.5.1 and I essentially got zeros for M. It sort of makes sense. The array x is an array of exceedingly small numbers (on the order of 10E-12), which is effectively zero. The coth of 0 is undefined.

[hrh1818]

On Dec 11, 11:40 am, Kris Matrix <x...@no.email> wrote:
> I have this code . I know it works in MATLAB. I tried to make it work in MathScript but in the last step, i only get M to be a zero matrix. however it should have values then. What is wrong??
> t = (0:0.0000001:.01);
> f1 = .0011312*cos(2*pi*49400*t);
> f2 = .0033936*cos(2*pi*100*t);
> H = (f1 + f2);
> x = ((2.93464*10^-10)*H);
> ms=1000;
> M = ms * (coth(x) - (1./x));

I wouldn't trust the answer in either Matlab or Labview. Wirh X being
approximately 1 x 10^-12 the coth of x and 1/x are both approximately
1 x 10^12 you are trying to calculate the difference between two very
large numbers. This is often an invitation to a disaster. I suggest
you rework your equations to eliminate the problem of determining the
difference between two large numbers.

Howard

[hrh1818]

On Dec 11, 11:00 pm, hrh1818 <hr...@att.net> wrote:
> On Dec 11, 11:40 am, Kris Matrix <x...@no.email> wrote:
>
> > I have this code . I know it works in MATLAB. I tried to make it work in MathScript but in the last step, i only get M to be a zero matrix. however it should have values then. What is wrong??
> > t = (0:0.0000001:.01);
> > f1 = .0011312*cos(2*pi*49400*t);
> > f2 = .0033936*cos(2*pi*100*t);
> > H = (f1 + f2);
> > x = ((2.93464*10^-10)*H);
> > ms=1000;
> > M = ms * (coth(x) - (1./x));
>
> I wouldn't trust the answer in either Matlab or Labview. Wirh X being
> approximately 1 x 10^-12 the coth of x and 1/x are both approximately
> 1 x 10^12 you are trying to calculate the difference between two very
> large numbers. This is often an invitation to a disaster. I suggest
> you rework your equations to eliminate the problem of determining the
> difference between two large numbers.
>
> Howard


I took another look at your question this morning and noticed most of
the values Matlab gave for M were equal to zero. Then I visually
compared some values of x that produced an answer of zero for M and
those that gave a non-zero value. Here is the comparison for Matlab.

show = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10];
>> x(show)

ans =

1.0e-011 *

Columns 1 through 7

0.1328 0.1328 0.1327 0.1326 0.1325 0.1324 0.1322

Columns 8 through 10

0.1320 0.1318 0.1315

>> M(show)

ans =

Columns 1 through 7

0 0 0 0 0 0.1221 0

Columns 8 through 10

0 0 0.1221

Notice how close the x values that gave a zero result are to the x
values that gave a non-zero result. That doesn't make sense. Hence I
conclude your problem as presently defined isn't suitable for Matlab
or Labview.

Howard

[hrh1818]

On Dec 11, 11:40 am, Kris Matrix <x...@no.email> wrote:
> I have this code . I know it works in MATLAB. I tried to make it work in MathScript but in the last step, i only get M to be a zero matrix. however it should have values then. What is wrong??
> t = (0:0.0000001:.01);
> f1 = .0011312*cos(2*pi*49400*t);
> f2 = .0033936*cos(2*pi*100*t);
> H = (f1 + f2);
> x = ((2.93464*10^-10)*H);
> ms=1000;
> M = ms * (coth(x) - (1./x));

Another update. I ran your code in Matlab using variable precision
arithmetic as follows.

>> digits(25)
>> y = vpa(x);
>> z = y([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]);
>> 1000 * (coth(z) - (1./z))

ans =

[ .5000e-9, .4000e-9, .4000e-9, .4000e-9, .4000e-9, .4000e-9, .
4000e-9, .4000e-9, .5000e-9, .5000e-9]

The answer given by variable precision arithmetic for the first ten x
values is completely different than the values given by double
precision arithmetic.

Howard