does matlab have a problem with modular integer arithmetic?

>> x = -78907292 * 1941317253; >> y = 2^32 +1

y =

    4.29496729700000e+009

>> x

x =

   -153.184087347109e+015

>> mod(x,y)

ans =

    1.51932828600000e+009

correct answer:

= 1519328274 (python) and others...

[KSSV]

With format long g as option mod(x,y) is 1519328286

Answer 1

[Walter Roberson]

Try

 X = int64(-78907292) * int64(1941317253)

Remember that the default data type is double not one of the integer data classes.

Answer 2

[Roger Stafford]

As has so often been pointed out in this forum, matlab’s “double” in everyone’s computers possesses a significand (mantissa) consisting of 53 binary digits. Consequently it is incapable of representing the above product -78907292*1941317253 exactly. For that reason the errors it must necessarily make will certainly be manifest using the mod function as given here. Have a heart! Or better still use the symbolic forms of numbers for such calculations.

Answer 3

[alexander sharp]

an integer is an integer is an integer - by any other name. it is wrong

[Walter Roberson]

MATLAB frequently allows people to use abbreviated forms. In MATLAB your line

x = -78907292 * 1941317253;

is considered to be an abbreviated form of

x = times(-78907292.0, 1941317253.0);

An integer might, as you say, be an integer, but you did not enter any integers.