Find the number of digits after the decimal point in a number, please help!

Say you have a number, x, and you want to find the number of digits after the decimal point. I know that in matlab, one way to do it would be to convert the number to a string and find the length from after the decimal point, but I think num2str has issues. How would you find the number of digits after the decimal point in a number then (the question is complicated by the fact that the numbers are floating point).

    %
function y = decimalpoint(x)
x = abs(x); %in case of negative numbers
y = 0;
while (floor(x)~=x)
y = y+1;
x = x*10;
end

This code doesn't work for some cases, such as decimalpoint(11.111). How would I make the code work for all cases? I think recursion could be useful too, but probably not necessary.

[Daniel Shub]

This is essentially a duplicate of http://www.mathworks.com/matlabcentral/answers/52068-machine-precision-problem-in-code

Answer 1

[Azzi Abdelmalek]

x=11.111
x = abs(x); %in case of negative numbers
n=0
while (floor(x*10^n)~=x*10^n)
n=n+1
end

[SB]

Wow, thanks so much! Out of curiosity as well, I saw an article about using eps(X), or ceil(-log10(eps(x))), do you know how that would work to solve the same question?

[Daniel Shub]

What about x=11.11100000000000065????

[Walter Roberson]

Afterwards what is the difference between n and y ? what is the purpose of calculating u ?

[Azzi Abdelmalek]

• u is not used.
• y is the same as n

[Azzi Abdelmalek]

Daniel, I think the max you can get is 15. But I can't explain why your code does'nt work

[Walter Roberson]

Why does the code return 5 for x = 11.111 * (1-eps)

[Azzi Abdelmalek]

 x=11.111*(1-eps)
 x*10^4=1.1111e+05
 floor(x*10^4)=111109

and

 x*10^5= 1111100

I know it has something to do with the way Matlab codes floting point numbers. Errors caused by quantification

[Daniel Shub]

The point is that MATLAB cannot represent any of these numbers exactly. Therefore the number of decimal places is poorly defined.

Answer 2

[Chris Taylor]

a = 11.111;
b = numdecpoints(a);

function f = numdecpoints(x)
i = -1;
X = 0;
while X ~= x
    i = i + 1;
    X = round(x, i);
end
f = i;
end

This seems to work, making use of MATLAB's round function. Not sure if it is the most efficient but seems to work for me. Simply compare the output of your original number and the number rounded to i decimal places. When the values match, i is the number of demical places in the orginal number.