[-] q5, is there any constrain on the e' edge?
by eial - Thursday, 14 May 2009 17:30:45
e.g. can it be any edge from the tested cut?
[-] Re: q5, is there any constrain on the e' edge?
by algo092 - Thursday, 14 May 2009 18:03:12
As mentioned in the question:
1. e' is an edge in F (F=The edges of T).
2. e'=(u,v) such that u is in S and v is in V\S.
[-] Re: q5, is there any constrain on the e' edge?
by eial - Thursday, 14 May 2009 18:17:01
so I understand that the answer is no, e.g. e' can be any edge in the cut.
thanks.
Re: q5, is there any constrain on the e' edge?
by algo092 - Thursday, 14 May 2009 19:48:41
As long as it is in F.
[-] Re: q5, is there any constrain on the e' edge?
by eial - Friday, 15 May 2009 19:04:36
there is something I don't quite understand, does e' is in F in the start or e?
if e' is in F, we selected it randomly? are we switching the wrong edge (e') with the right edge (e) or e is selected randomly too?
[-] Re: q5, is there any constrain on the e' edge?
by algo092 - Sunday, 17 May 2009 13:03:54
e' is an edge in F. We select it following these conditions:
1. e' is an edge in F (F=The edges of T).
2. e'=(u,v) such that u is in S and v is in V\S.
e is not selected randomly, but according to the algorithm rule of choosing the next edge.

The rest is for you to figure out.
[-] Re: q5, is there any constrain on the e' edge?
by eial - Monday, 18 May 2009 12:39:30
ok, I think I understand it abit more, now F is the edges set of the final tree or the tree in each iteration? and e' is any edge from F and e is any edge from the cut that is defined by e'.
also T is mst but T' isn't an mst, right?
just to understand the question better in order to tackle it with more ease.

edit: after looking at the question again, I've
deduce that the change in after T was completed (thus removal of e' creates 2 sub trees) but, if e'!=e, e'=(u,v) so u in S and v in V\S and e=(x,y) so x in S and y in V\S, then e' must be e or else there is a vertex l in V which isn't in S so T isn't a mst but T is, so in light of this, I must deduce that the change is created in a single iteration, but removal of e' doesn't necessarily creates two sub trees.

can you clarify when does the change occurs? and what exactly are the conditions?

thanks.

Re: q5, is there any constrain on the e' edge?
by algo092 - Tuesday, 19 May 2009 15:16:12

The question is not new. It's based on the proof made in class.
The question suggests a different "switching" argument.
If you feel uncomfortable with the background (The current status in the step of the induction) please review the lecture first.
Then you'll see that the argument is just a different suggestion.

You are not supposed to understand it without knowing the details.