Darren New <dne### [at] sanrrcom> wrote:
> Warp wrote:
> >   I don't think so. The very definition of "deterministic" is predictability.

> That's where we disagree. It's completely possible to be both deterministic 
> and unpredictable. Indeed, that's exactly what the halting problem is all about.

  But a deterministic program will always behave the same way. It doesn't
change its behavior from one execution to another. Thus it behaves predictably.
"It will do the same thing it did the last time."

> > The very word itself is saying so. It's the opposite of "non-deterministic",
> > which is unpredictability.

> Also not quite true. Non-deterministic turing machines are very predictable 
> in their behavior.

  If it's non-deterministic, you cannot say how it will behave the next time
it will be executed.

> >   A chain of events is deterministic if it happens in a certain way because
> > there's no other way it could have happened. If the exact same initial setup
> > can be replicated, then the chain of events will happen in the exact same
> > way again, completely predictably. That's the very definition of
> > deterministic.

> Yet, oddly, NDTMs behave that way. ;-)

  If it always behaves the same way, isn't it by definition deterministic?

-- 
                                                          - Warp

Post a reply to this message