how to save variables not in scientific notation

I have certain variables saving in scientific notation. I am trying to use them in creating a matrix, for example:

nc = 19840; %saves the variable as 1.9840e+04

A = ones(nc,nc)

When I try to do this, I get the following error:

Error using ones

Size inputs must be integers.

I guess Matlab isn't reading it as an integer? Is there a way to make it save the variable not in scientific notation? (I don't have the problem with anything not saved in sci notation) I've tried things like sprintf and num2str without luck.

[Walter Roberson]

What shows up for

nc - 19840

[Nick Thomas]

In the workspace the value shows up as 1.9840e+04

[Stephen23]

@Nick Thomas: please show the output of these operations:

>> nc - 19840
>> whos nc

[Nick Thomas]

>> nc-19840

ans =

     3.6380e-12

>> whos nc

    Name      Size            Bytes  Class     Attributes

    nc        1x1                 8  double              

This is interesting, now I see that it's somehow storing a small decimal amount with the value. It doesn't make sense to me why though, here is my code for calculating nc along with the variables used.

nb = 128;
h = 0.2;
t_s = 0.004;
t_i = 0.05;
s = 500;

nc = nb+(nb*((h+2*t_s+2*t_i)*s));

So there is no reason I should be having a weird decimal at the end.

Knowing this I fixed the problem using the round(x,0) command. However, I'm still curious as to why this happened?

Answer 1

[the cyclist]

The reason is that the intermediate values in your calculation cannot be stored exactly. Here is one starting point for understanding why that is.

Answer 2

[Walter Roberson]

http://matlab.wikia.com/wiki/FAQ#Why_is_0.3_-_0.2_-_0.1_.28or_similar.29_not_equal_to_zero.3F

[Walter Roberson]

syms deltah
nb = sym(128, 'r');
h = sym(0.2, 'r') + deltah;
t_s = sym(0.004, 'r');
t_i = sym(0.05, 'r');
s = sym(500, 'r');

nc = nb+(nb*((h+2*t_s+2*t_i)*s));

nc
subs(nc, deltah, eps(0.2))
vpa(ans)

nc =
64000*deltah + 19840
ans =
1396115884484853885/70368744177664
ans =
19840.00000000000177635683940025

You can see from this that a difference of 1 bit in the representation of 0.2 makes a difference of

>> vpa(subs(nc,deltah,eps(0.2)) - 19840)
ans =
0.0000000000017763568394002504646778106689453

0.2 is not exactly representable in binary floating point: it is an infinite repeating decimal, just like 1/7 is not exactly representable in finite decimal.