Question: Maple's calculations are non-deterministic ?

Hello,

I've performed the exact same calculation (using the exact same worksheet, with absolutely no changes made) last night and today and got these results:

y=[0, .4646313991, .9499094532, .5799050874, .1092604065, .4031507404, .8483735670, .6138799333, .1913933456, .3724618402, .7707837830];

y=[0, .4646313991, .9499094532, .5799050874, .1092604061, .4031507404, .8483735643, .6138799354, .1913933461, .3724618343, .7707837938];

Which are exactly the same modulo the last 3 digits.

Digits=10 in both cases.

It seems as if maple is truncating digits based on something dynamic,

(such as, maybe, the CPU usage of my laptop. If I have a lot of things running, it will truncate more digits.. if the CPU is completely free, it won't trucate as many)

I learned in my symbolic computation course that maple uses Las Vegas algorithms for things like "prime number verification" and "factoring polynomials exactly" ..... but those algorithms have 0-sided error and always give the same result.

My code is only doing sums and products of the functions cos(2.2) and,

sqrt(-1)* sin(2.2)  ... then taking the Real part of the final answer. So the code is very simple, although it does millions of iterations of the above operations.

Does anyone know what kind of things may be causing the discrepancy in the last 3 digits ? Might it be truncating  more or less digits based on the machine's CPU and memory requirements ?

Any comments or discussion would be greatly appreciated,

thanks