type conversion, precision loss

I observed a strange behaviour when working with data read from a binary file. Here a small example:

>> a=int32(1);

>> b=12.5;

>> c=a+b

c =

    14

The value of c is now 14 "instead" of 13.5. Why is (int32+double) silently typecasted to int32? Tried it with different matlab versions (2010, 2011, 2012, 2013) with the same result. So maybe I just don't understand the logic behind it? I guess I'd prefere a result without precision loss?

Answer 1

[Azzi Abdelmalek]

It's what Matlab do. I don't think, there is a logic behind it. If the result was double, you can ask the same question.

you can wrtie

a=int32(1);
b=12.5;
c=double(a)+b

Answer 2

[Walter Roberson]

http://www.mathworks.co.uk/matlabcentral/answers/110812-j-k-gives-wrong-answer-in-a-loop