[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: rewriting CSP processes

"Campbell, John" wrote:
> Hi All
> I've a CSP question. The words around the mathematics
> talk about "events" being instantaneous transactions that
> are handshaken (both the Roscoe and Schneider books
> use those terms).  In real life, that's not possible.  Handshaking
> protocols don't execute in zero time, and signals don't
> propagate instantly.  Is there anything in the CSP literature
> that deals with this problem?
> It seems that you could re-write any set of CSP processes
> in terms of lower level processes.  For example, a shared
> event at the macro level would be re-written as a set of
> events at the micro level, where the hand shaking would be
> implemented as micro-processes (Ignore that pun).  At some
> point you get to logic gates that intrinsically respond to
> "events" at their input terminals.

I ought to add to my previous note that one need not stop at the logic
gate level, but continue beyond to the analogue circuit implementing
that gate. With the potential to prove the whole system correct using a
common CSP framework. That was the main motivation which drove my ideas
for a "continuous" CSP. In THCSP (continuous CSP), I have both
instantaneous events mainly to accomodate the higher level abstractions,
but also events of arbitary duration. There is no lower bound on the
time between events as in TCSP. That does mean that you can write
recursions that make no progress (the main reason why TCSP requires that
lower bound), but we often allow CSP to describe "unreasonable" things -
like divergence. There is just an extra proof obligation to show that
the real process does describe something with a sensible time evolution.
That is not a problem in real applications.

Just as important, we can now think about an occam compiler generating
analogue circuits: with defined behaviour. 

Since others have picked up on my mention of quantum computing in the
fringe presentation, I should add that most models of quantum computers
are essentially discrete, working with quantum bits (qubits) manipulated
by quantum gates. But THCSP does allow us to capture Schrodinger's
Heisenberg picture evolution, or a ket evolution descriptions if need
be. I don't know about relativistic quantum mechanics, but most work on
quantum computers seems to be non relativistic. I hope that we will
learn much more about the ultimate parallelism (something about which
this community is supposed to know) of quantum computation at CPA-2001
at Bristol.


[Trying to catch up with this flood of CSP postings. It always happens
after the conference. Then things tend to dry up. We should manage
things better. But it is evidence also of how stimulating are our
meetings :-) ]

Dr A E Lawrence