[CST-2] OptComp 93/9/6
M.Y.W.Y. Becker
mywyb2@cam.ac.uk
Mon, 21 May 2001 17:29:51 +0100
Has anyone done this (rather hard) question?
Is this correct?
(a)
VB(n) = empty if succ(n)=empty
VB(n) = INTERSECTION over succ(n):
VB(s) UNION gen(s) \ kill(s), o/w
where
gen(s) = set of expressions freshly computed at s
kill(s) = set of expressions containing vars modified at s
(b)
* backwards flow equation
* neither VB(n)<=AVAIL(n) nor the other way round
* what else???
(c) as usual, find max. fixed point of continuous transformation on finite
lattice by repeated application etc.
Second part: ???
If f is strict in y it doesn't necessarily mean that y is very busy
somewhere.
If y is very busy somewhere it doesn't necessarily mean that f is strict in
y.
So what's the relation between strictness and very-busy-ness?
Mo