Transitive closure of a relation

  • Thread starter Thread starter icantadd
  • Start date Start date
  • Tags Tags
    closure Relation
Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
5 replies · 4K views
icantadd
Messages
109
Reaction score
0

Homework Statement


Let R be a relation and define the following sequence
[tex]R^0 = R[/tex]
[tex]R^{i+1} = R^i \cup \{(s,u) \vert \exists t, (s,t) \in R^i, (t,u) \in R^i \}[/tex]
And
[tex]R^{+} = \bigcup_{i} R_i[/tex]
Prove that [tex]R^{+}[/tex] is the transitive closure.


Homework Equations





The Attempt at a Solution


1)
[tex]R = R_0 \subset R^{+}[/tex]

2) show transitive
Assume [tex]xR^{+}y,yR^{+}z[/tex] Then there is a
k, such [tex]\exists u, xR^{k}u, uR^{k}y[/tex], and
l, such [tex]\exists w, yR^{l}w, wR^{l}z[/tex]
Therefore [tex]\exists m \geq k+l, xR^{k+l}z[/tex], because ??
I don't know why I think this is the way to go, and I don't know how to finish this part of the proof. I will continue on. Please suggestions here??

3) show smallest transitive relation (by induction)
Let T be any other transitive relation containing R.
(I) [tex]R^0 = R \subset T[/tex]
(II) Assume [tex]R^n \subset T[/tex] and let [tex]xR{n+1}y[/tex]. Then there is a z, such that
[tex](x,y) \in R^n \cup \{(x,y) \vert (x,z) \in R^n, (z,y) \in R^m \}[/tex]
Now if [tex](x,y) \in R^n[/tex] then we are done. Assume [tex](x,y) \in \{(x,y) \vert (x,z) \in R^n, (z,y) \in R^m \}[/tex]. Then because [tex]R^n \in T[/tex] , then [tex](x,z) \in T, (z,y) \in T[/tex]. Then because T is transitive (x,y) is in T.
?Is my last statement here correct?

Therefore [tex]R^n \subset T[/tex]

Any help on the above two questions would be greatly appreciated.
 
Physics news on Phys.org
Hi icantadd! :smile:
icantadd said:
2) show transitive
Assume [tex]xR^{+}y,yR^{+}z[/tex] Then there is a
k, such [tex]\exists u, xR^{k}u, uR^{k}y[/tex], and
l, such [tex]\exists w, yR^{l}w, wR^{l}z[/tex]

Try it with k = l. :wink:
 
Do you mean max{k,l}?
I got credit for changing my answer to that. Meaning I had the answer right. I seem to have trouble in this area, a lot of times, I know that something is working, but I don't know how to justify it to myself. For example, with this problem I can look at it, and say "yes, that is obvious" and write a proof that is technically right, but then I look back at it and I don't know why. Why does that answer work? Because you are guaranteeing recursively that the relation is transitive one step at a time. So then I realize my question is okay, how do you know you are making the relation transitive one step at a time? And to that I simply answer, "because it is" and that just isn't good enough for me.

How do I correctly state what I am thinking on this problem, what is it that makes this work?
 
icantadd said:
… How do I correctly state what I am thinking on this problem, what is it that makes this work?

mmm … all I can suggest is that, in a problem like this, you start by translating the definition into words before you do anything else

in this case, stare at the definition until you satisfy yourself that Rn is the "pairs of people with 2n degrees of connection" … that is, Bud know Fred knows … … … Ginger knows Lou (with 2n names)

then you can safely predict what will happen, and translate it back into symbols :smile:
 
Then (somewhat removed from the original question) it is appropriate to say something like,
at R^n for any (a,b) there is a set[tex]\{(a,a_1) (a_1,a_2) , ... , (a_{n-1},b)\}[/tex]whose size is at most n.

Back to the problem:
At any rate, if i take m = max {k,l}, then clearly we have
[tex]xR^{m}y[/tex] and also [tex]yR{m}z[/tex], and therefore in [tex]R^{m+1}[/tex] we will have [tex]xR^{m+1}z[/tex]. I think I am getting it. Thank you!
 
icantadd said:
At any rate, if i take m = max {k,l}, then clearly we have
[tex]xR^{m}y[/tex] and also [tex]yR{m}z[/tex], and therefore in [tex]R^{m+1}[/tex] we will have [tex]xR^{m+1}z[/tex]. I think I am getting it. Thank you!

Yes, that's exactly right! :biggrin: