I am trying to implement a basic pelin noise function, but I have some problem with the integer to noise function used to generate deterministic noise from integer inputs.

the function I am trying to implement is defined there:
http://libnoise.sourceforge.net/noisegen/index.html#continuousnoise

and my code so far look like this:

noise2d :: (Int32, Int32) -> Double
noise2d (x, y) =
    let m = x + y * 57
        n = (shiftR m 13) ^ m
        j = (n * (n * n * 15731 + 789221) + 1376312589) .&. 0x7fffffff
    in  1.0 - (fromIntegral j / 1073741824.0)

the code compile but I get the same result for any input, due to the fact that n is evaluated to 0.

Is there a better way to do that?