Round double value to 2 decimal

Hi all,

I'm trying to round an array of double values to 2 decimals.

This is my code:

for k=1:n,
    y(k)=k*st;
    y(k)= sprintf('%0.2f', y(k));
    x(k) = sqrt(d^2+(2*y(k))^2)-d;
    x(k)= sprintf('%0.2f', x(k));
end

I got the following error message:

In an assignment A(I) = B, the number of elements in B and I must be the same.

BTW! the value n is 10.

I want the arrays y and x with 2 decimal values. Ex: 1.23

Thanks a lot.

Raúl. In an assignment A(I) = B, the number of elements in B and I must be the same.

[the cyclist]

What is the size of your variable d? If it is a matrix, then it looks like the line

x(k) = sqrt(d^2+(2*y(k))^2)-d;

is trying to assign a matrix into a single element, and that will give the error you see.

[Raúl]

The size of d is 1, it is a double value.

[Walter Roberson]

Are you trying to change how the number is displayed, or to assign a new value that is the old one rounded to 2 decimal digits?

[Raúl]

Hi Walter, the second approach.

Thanks.

Answer 1

[Matt Tearle]

The reason you're getting the error is that sprintf creates a char array -- in this case, a 1-by-4 char array made up of the 4 characters '1', '.', '2', and '3'. But your assignment is to a single element y(k).

Possible solutions are to index into the kth row of a char array, which you should keep separate from your double array y (otherwise you'll get another error), or to use cell arrays as Wouter suggests, or to just round everything as doubles and avoid strings altogether (as Wouter also suggests).

I'd recommend the last approach. You can always display things with sprintf or fprintf at the end, and this can be done without a for-loop: fprintf(1,'%0.2f\n',x)

[Raúl]

Great, the last approach works. However, it rounds the number but it keeps the rest of the number as zeros. I would like to delete all this extra zero.

For example:

From 1.3400 leave it as a 1.34.

Thanks again.

[Matt Tearle]

I don't understand what the problem is. MATLAB, like any numerical language stores all floating-point numbers to a certain precision (in MATLAB, the default is double, which is about 16 decimal places), so rounding 1.234567... to 2 dp means -- by definition -- changing the stored value to 1.230000...

How the number is stored is different to how it's displayed. MATLAB's default display is 4 dp. If you want the number displayed in a certain format, use something like fprintf. In the special case of 2 dp, you could also use the MATLAB command format bank, which will set the Command Window display style to 2 dp instead of 4. This is a universal change until you use format to change back to a different style.

Answer 2

[Wouter]

How about this:

a = rand(1,10);
y ={}; 
n=length(a);

for k = 1:n
 y{k} = sprintf('%0.2f',a(k)); 
end

Because sprintf returns a string, you need to put it in a cell: {}.

You could also round a like this (if you do not want strings):

a = rand(1,10);
y = round(a * 100)/100; % two decimals remain; more decimals are set to 0.

[Raúl]

Hi Wouter,

I've modified in this way, but still getting error message.

w ={};
z ={};
for k=1:n,
    y(k)=k*st;
    w{k}= sprintf('%0.2f', y(k));
    x(k) = sqrt(d^2+(2*w{k})^2)-d;
    z{k}= sprintf('%0.2f', x(k));
end

The error is the same

Then, i tried with:

for k=1:n,
    y{k}=k*st;
    w{k}= sprintf('%0.2f', y{k});
   x{k} = sqrt(d^2+(2*w{k}).^2)-d;   
    z{k}= sprintf('%0.2f', x{k});
end

It works, but the cell z is composed by 10 cell of 1x21char... instead of doubles in the way X.XX.

Thanks.