Here you can ask questions and find or give answers to organizational, academic and other questions about studying computer science.

1.1k questions

1.3k answers

1.7k comments

557 users

0 votes

https://es.cs.uni-kl.de/teaching/vrs/exams/2019.08.27.vrs/2019.08.27.vrs.solutions.pdf

In this exam paper Question 2 a) I am not able to understand the use of hint. How to utilize this hint, otherwise solving without hint is very lengthy.

Thank you in advance.

in * TF "Emb. Sys. and Rob." by (870 points)

1 Answer

+1 vote

It is about the fact that the bisimulation relations on the same structure are equivalence relations and therefore reflexive and symmetric. Reflexive means that for every state s, there is a pair (s,s) included from the beginning and will remain there, so that it is nasty to always write it down. You may call the set of such pairs ID (for identity), and always write {...} ∪ ID without listing ID explicity.

Also, the pairs are symmetric  which means that if there is (s,s'), then there is also (s',s). Again, you may skip half of the pairs if you note down that one should add the symmetric closure. 

For example, the first step was  

    (S0,S0);(S0,S5);
    (S1,S1);(S1,S2);(S1,S3);(S1,S4);
    (S2,S1);(S2,S2);(S2,S3);(S2,S4);
    (S3,S1);(S3,S2);(S3,S3);(S3,S4);
    (S4,S1);(S4,S2);(S4,S3);(S4,S4);
    (S5,S0);(S5,S5)

and becomes then

    {(S0,S0);(S1,S1);(S2,S2);(S3,S3);(S4,S4);(S5,S5)}
    ∪
    {(S0,S5);
     (S1,S2);(S1,S3);(S1,S4);
     (S2,S3);(S2,S4);
     (S3,S4)}
    ∪
    {(S2,S1);
     (S3,S1);(S3,S2);
     (S4,S1);(S4,S2);(S4,S3);
     (S5,S0)}

which would be written just

    ID
    ∪
    {(S0,S5);
     (S1,S2);(S1,S3);(S1,S4);
     (S2,S3);(S2,S4);
     (S3,S4)}^

where R^ should denote the symmetric closure.

by (170k points)
Ok. I understand now. So it is just to save time while writing.

1) If pair (s0,s1) satisfies then can we eliminate checking (s1,s0) with the reasoning that if (s0,s1) satisfies then (s1,s0) will also satisfy?

2) For quotient structure if in greatest Bisimulation result if we get pairs like (so,s1), (s1,s2), (s2,s3), (s3,s4) (and some other pairs) so in merging similar states step will we merge all and make single state (s0,s1,s2,s3)? Because s0 is similar to s1, s1 is similar to s2, and so on.

3) During finding the greatest bisimulation relation, if we get pair (s0,s1) and successor of s0 is s0 and successor of s1 is s1. So should we keep this pair? Because it is checking on itself.

Thank you in advance :)
Also, it saves redundant checks.

1) For bisimulation with itself, every relation is reflexive. So if you keep a state pair, you also need to keep its symmetric twin. If you kick it out, you also need to kick out its twin. Thus, you check whether (s1, s0) stays in the relation or gets kicked out. The result also applies for (s0, s1)

2. We merge so that the merged state represents the biggest set of states all similar to each other. Technically, your relation should be transitive. So if you have (so,s1), (s1,s2), you should also have (s0, s2).

3. The computation of the (bi)simulation relation is an iterative fixpoint procedure. You *always* use the *previous* relation as your reference. If a pair (s0, s1) is in R5, then this is what you consider when computing R6. So if s0 is the only successor of s0, and s1 the only successor of s0, then having (s0, s1) in R5 means that we keep it in R6. Note that this rule (to always consider the previous relation), has to be applied with all consequences. For example, you have (s4,s5) in R3, you kick out (s4,s5) while computing R4, you then have (while still computing R4) a relation (s3,s7) that relies on (s4,s5). You would then keep  (s3,s7) in R4 because you only consider R3 for now. When computing R5, you kick out  (s3,s7) because (s4,s5) wouldn't be in R4.
ok. I understood now.
One more doubt if for my pair (s0,s1)
1) Greatest Simulation relation - if s0 has no successor but s1 has (let's say s2) then we will consider this pair

2) Greatest BiSimulation relation - if s0 has no successor but s1 has (let's say s2) then we will kick out this pair.
Right, since for the simulation relation, we just have to make sure that the second structure is at least able to do what the first can do. If it can do more, it's not forbidden. For bisimulation, however, it is forbidden, there both structures must be bisimilar, i.e., equally powerful.
1) Simulation relation would work, depending on the direction. See also: https://q2a.cs.uni-kl.de/1527/simulation-and-bisimulation
2) Bisimulation relation would not work. As s0 has no successors, there's nothing s2 could be in relation with.
Thank you :)

Related questions

Imprint | Privacy Policy
...