sym(2/pi*0.01, 'd') * cos(t)
The reason you are getting what you are getting, is that the default symbolic conversion of floating point number to symbolic number, is for the floating point number to be converted to rational. Well, not really rational as such: it is extended to include sqrt(), and pi.
You can specifically tell MATLAB to convert floating point into software floating point by using the 'd' flag.
Or you can convert after the fact by using vpa() on the result you got.