How can I test if an input value is an integer?

I want to know how to test whether an input value is an integer or not. I have tried using the function isinteger, but I obtain, for example, isinteger(3) = 0. Apparently, any constant is double-precision by Matlab default, and is therefore not recognized as an integer. Can anyone tell me what function I'm supposed to use?

[Torsten]

Please explain in further detail the reason why you want to perform the test.

Best wishes

Torsten.

[Hung Chu]

For the function I'm trying to write, if any input of the function is not an integer, the function returns an error.

Answer 1

[James Tursa]

E.g.,

>> isaninteger = @(x)isfinite(x) & x==floor(x)
isaninteger = 
    @(x)isfinite(x) & x==floor(x)
>> isaninteger(3)
ans =
     1
>> isaninteger(3.1)
ans =
     0
>> isaninteger(inf)
ans =
     0
>> isaninteger(nan)
ans =
     0
>> isaninteger(uint8(4))
ans =
     1
>> isaninteger([-inf -1.2 0 1.3 5 inf nan])
ans =
     0     0     1     0     1     0     0

[Guillaume]

Another option,

isaninteger = @(x) mod(x, 1) == 0

which works for NaN and +- Inf

[tqml]

I think using equal with floating point values might be error prone.

> mod((1:5)+eps,1)==0

ans =

  1×5 logical array

   0   1   1   1   1
   

For double value eps should be aroung 1e-15, so small rounding errors could mess up the check.

Therefore it might be helpful to introduce an additional parameter to control the tolerance and only check if the result is CLOSE to zero (and not equal)

function [b] = is_integer_value(x, tolerance)
%IS_INTEGER_VALUE Returns true if the value is an integer
% Checks if x modulo 1 is close to 0.

if nargin < 2
    tolerance = eps;
end

b = ( mod(x,1) <  tolerance ) ;

end

[Steven Lord]

When you call the eps function with no inputs, it returns the distance from 1 to the next largest number.

The distance from (for example) 2 to the next largest number is greater than eps. When you call eps with an input, it returns the distances from the elements of the input to the next largest numbers.

>> x = 1:5;
>> y = x + eps;
>> y > x
ans =
  1×5 logical array
   1   0   0   0   0

>> z = x + eps(x);
>> z > x
ans =
  1×5 logical array
   1   1   1   1   1

You can see the value of eps for each element of x:

>> eps(x)
ans =
  Columns 1 through 3
      2.22044604925031e-16      4.44089209850063e-16      4.44089209850063e-16
  Columns 4 through 5
      8.88178419700125e-16      8.88178419700125e-16

Answer 2

[Jan]

According to your comment and Guillaume's suggestion:

assert(mod(x, 1) == 0, '%s: X must be an integer', mfilename);

Or safe this as M-file:

function T = isIntegerValue(X)
T = (mod(X, 1) == 0);
end

And then:

if ~isIntegerValue(X)
  error('%s: X must be an integer', mfilename);
end

Answer 3

[Hung Chu]

Thank you! It works! But I don't know why LOL :P Could you explain to me how that function handles works?

Answer 4

[Adam]

I use e.g.

validateattributes( myInteger, { 'double' }, { 'scalar', 'integer' } )

for validating my input arguments. Not so useful if you want a boolean/logical returned though to do something with after, although you can catch the exception that gets thrown but its untidy and inefficient to do that.

Answer 5

[randell mullally]

I needed the same thing. these answers are Blaah for me so here was my soloution.

clear;
    clc;
A=[3.6,4,4.25,5,6.175,nan,inf];
SizeA=size(A);
B=zeros(SizeA);
for i=1:SizeA(1,2)
    if A(i)/round(A(i))==1
    B(i)=1;
    end
end
AB=[A;B]

AB =

3.6000 4.0000 4.2500 5.0000 6.1750 NaN Inf

0 1.0000 0 1.0000 0 0 0

I suppose if you want to test just one at a time... and let y be the test subject.

    y=5;
    if y/round(y)==1
        fprintf('%3g is an intiger!! hooray\n',y)
    else
        fprintf('%3g is not an intiger!! sorry\n',y)
    end

5 is an intiger!! hooray

testing this with A yeilds good results.

    clear;
    clc;
A=[3.6,4,4.25,5,6.175,nan,inf];
SizeA=size(A);
B=zeros(SizeA);
for i=1:SizeA(1,2)
    if A(i)/round(A(i))==1
    B(i)=1;
    end
    y=A(i);
    if y/round(y)==1
        fprintf('%3g is an intiger!! hooray\n',y)
    else
        fprintf('%3g is not an intiger!! sorry\n',y)
    end
end
AB=[A;B];

3.6 is not an intiger!! sorry

4 is an intiger!! hooray

4.25 is not an intiger!! sorry

5 is an intiger!! hooray

6.175 is not an intiger!! sorry

NaN is not an intiger!! sorry

Inf is not an intiger!! sorry

If you really really really like this, ALOT!

find me on linkedin and endorse my matlab skill.

www.linkedin.com/in/randell-mullally

Best of luck!

[Cris Luengo]

This does not work if the input is 0.

What is wrong with the other answers? They actually work!

Answer 6

[Weimin Zhang]

For an array x, perhaps use this:

assert(~any(mod(x,1)),'x must have integer(s) only');

or a function allIntegers = @(x)~any(mod(x,1))

For example,

>> allIntegers([1,2,3])

ans = logical 1

>> allIntegers([1,2.5,3])

ans = logical 0