07.01.2013 Views

Lecture Notes in Computer Science 3472

Lecture Notes in Computer Science 3472

Lecture Notes in Computer Science 3472

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

182 Machiel van der Bijl and Fabien Peureux<br />

Example. Another similar difference is illustrated <strong>in</strong> Figure 7.4. We have the IOA<br />

p1 and p2, both can perform an arbitrary number of a <strong>in</strong>put actions followed<br />

by one y output action, followed aga<strong>in</strong> by an arbitrary amount of a actions. p1<br />

and p2 are equivalent accord<strong>in</strong>g to the quiescent preorder (both p1 ⊑Q p2 and<br />

p2 ⊑Q p1), because they have the same external traces and their quiescent traces<br />

conta<strong>in</strong> at least a y action. However, p1 and p2 are not equivalent accord<strong>in</strong>g to<br />

the fair preorder, when consider<strong>in</strong>g the same partition<strong>in</strong>g as before: {y,τ}. This<br />

is because a ω isafairtraceofp1 but not of p2. This might not be easy to see<br />

at first glance, but remember that the partition of locally controlled actions is<br />

{y,τ}. Inp1 we can do the fair execution (q0·a·q1·τq0) ω (and thus the fair trace<br />

a ω ).<br />

q1<br />

q0<br />

!x !x<br />

τ<br />

q2<br />

q3<br />

!y<br />

Fig. 7.5. Quiescent versus fair preorder example 3<br />

Divergence (i.e., the possibility for a system of to do an <strong>in</strong>f<strong>in</strong>ite number of<br />

<strong>in</strong>ternal transitions) shows another difference between the quiescent and fair<br />

preorder. Because of divergence, a fair trace is not necessarily a quiescent trace,<br />

as is illustrated <strong>in</strong> the next example.<br />

Example. In Figure 7.5, we have two IOA p1 and p2 with output actions x and<br />

y. Accord<strong>in</strong>g to the quiescent preorder both automata are equivalent. But they<br />

are not equivalent accord<strong>in</strong>g to the fair preorder, s<strong>in</strong>ce x isafairtraceofp1, but<br />

not of p2.<br />

As we can see from these examples, the fair preorder is a stronger relation<br />

than the quiescent preorder. Because of the def<strong>in</strong>ition of the quiescent preorder,<br />

this is not very surpris<strong>in</strong>g. As a side step, Segala shows a way to make the<br />

quiescent preorder and the fair preorder equivalent by mak<strong>in</strong>g some restrictions<br />

on the IOA’s that we allow [Seg97]. Basically, the property we are look<strong>in</strong>g for is<br />

that we can approximate an <strong>in</strong>f<strong>in</strong>ite fair trace by a f<strong>in</strong>ite trace and can extend a<br />

f<strong>in</strong>ite fair trace to an <strong>in</strong>f<strong>in</strong>ite fair trace. This is expressed by the properties fair<br />

cont<strong>in</strong>uity and fair approximability. AnIOAp is fair cont<strong>in</strong>uous if the limit of<br />

any cha<strong>in</strong> of fair traces of p is also a fair trace. Fair cont<strong>in</strong>uity is nicely illustrated<br />

<strong>in</strong> Figure 7.3. The sequence a n is a fair trace of p2, but if we take n to <strong>in</strong>f<strong>in</strong>ity<br />

this is not the case. In other words the trace a n is not fair cont<strong>in</strong>uous. An IOA<br />

p is fair approximable if each <strong>in</strong>f<strong>in</strong>ite trace of p is the limit of a cha<strong>in</strong> of fair<br />

traces of p. This is illustrated <strong>in</strong> Figure 7.4. The trace a ω is a fair trace of p1<br />

but it is not fair approximable because the f<strong>in</strong>ite trace a n is not fair.<br />

t0<br />

t1<br />

t2<br />

!x<br />

!y

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!