Jump to content

Talk:Dinic's algorithm

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Source for alt name Chaim?

[edit]

This article uses the name "Yefim (Chaim) A. Dinitz", but I can't find any reference for the alternate name "Chaim", and I don't see it in the Russian version of the article. (There's no linked Hebrew version of the article at the moment. The source cited is just a translation of the original paper on the algorithm, which names the author as "E. A. Dinic".) Since sourcing is especially important for facts about living people, I'll change the name to "Yefim Dinitz" tomorrow if we don't have a source by then. If you find a source later, please change it back! —Vectornaut (talk) 21:31, 22 December 2023 (UTC)[reply]

Dinic or Dinitz?

[edit]

Why is the Algorithm called Dinic's algorithm, when the guy who invented it is called Yefim Dinitz? That doesn't make sense to me - I think it should be both written the same. Wadi oo (talk) 09:17, 2 November 2009 (UTC)[reply]

Look up his personal homepage on BGU - http://www.cs.bgu.ac.il/~dinitz/ He refers to his own work as Dinic's Algorithm. Strange. 79.176.200.40 (talk) 14:01, 6 May 2011 (UTC)[reply]

Well he explains it right there in his article. Shimon Even first popularized the algorithm in the west under the name "Dinic'c algorithm", which was "rendered incorrectly as [dinik] instead of [dinits]". That explains why alternative spellings of Dinitz's name are so rarely seen. -- X7q (talk) 16:55, 6 May 2011 (UTC)[reply]

Complexity of blocking flow calculation

[edit]

Isn't it possible to find a blocking flow in a level graph in time using the Malhotra-Kumar-Maheshwari blocking flow (MPM) algorithm? This would yield an overall running time of . 82.130.21.149 (talk) 20:20, 7 December 2009 (UTC)[reply]

Former Russian?

[edit]

Do we have any evidence to believe that he had renounced his Russian citizenship? If not, should we state that he is "former Russian"? Unless we are talking about ethnicity, in which case, he should have always been Jewish, and not either Israeli or Russian... just a thought ... — Preceding unsigned comment added by Gabiteodoru (talkcontribs) 17:43, 13 December 2010 (UTC)[reply]

I suggest that the segment in the first paragraph that refers to Dinitz as formerly Soviet be removed from the article. Also, consider changing Israeli to Jewish or removing the reference to his ethnic background altogether. Finally a link to his personal homepage may be in order (see here: http://www.cs.bgu.ac.il/~dinitz/). 79.176.200.40 (talk) 14:03, 6 May 2011 (UTC)[reply]

I agree, it's not nice to discuss his ethnicity in this article. But the fact that the algorithm was invented in the USSR and not in the West looks to me historically significant and it should be somehow mentioned in the article, perhaps at least by mentioning the journal in which the algorithm was originally published - "Soviet Math. Doklady". As for a link to personal homepage - there's already a direct link to the paper about the algorithm on his website. That's sufficient. Another link to his homepage would be against guidelines (items 13 and 19). -- X7q (talk) 16:39, 6 May 2011 (UTC)[reply]

Running Time

[edit]

I might be missing something, but it seems as if the running time should be O(VE^2) rather than O(V^2E)? That's what the information in the Analysis seems to suggest. 18.111.103.83 (talk) 04:08, 6 December 2011 (UTC)[reply]

I don't see how it could possibly suggest that. It says that there are at most n-1 = O(V) blocking flows to be found and finding each one takes O(VE) time. Multiplying these quantities you get O(V^2 E). -- X7q (talk) 19:11, 6 December 2011 (UTC)[reply]

inputs

[edit]

where did you get the inputs used in every path...? i'am applying this algorithm with my case study.and my problem is that i don't know where will i derived the inputs? please help me... — Preceding unsigned comment added by 122.52.159.76 (talk) 10:37, 6 January 2012 (UTC)[reply]

Definition of blocking flow?

[edit]

The current text reads like this (emphasis mine):

   A blocking flow is an s-t flow f such that the graph G' [...] contains no s-t path.

It would be useful to show G' in the example too, otherwise the no remains a bit cryptic in the definition. Unless, of course, it should be read as an or one.

--213.203.132.119 (talk) 08:06, 12 January 2018 (UTC)[reply]

Example unclear

[edit]

I don't understand how the Example ("simulation") is constructed starting the very first step, the initial "guess"(?) G_L in row "1." of the table. Why is there no flow from node (1) to node (2), but maximum flow of 4 from (1) to (3)? I understand that from (s) I can send 10 units to (1). I could also send 10 units to (2), but since this isn't done, I assume that we go on (in "depth first" mode) with outgoing arrows from node (1) and don't consider the edge (s)-->(2) for the moment. But if the incoming 10 units are distributed over the outgoing arrows...

... in order of the label of destination nodes, then one should first have a flow of 2 units to node (2), then 4 units to node (3), and the remaining 4 units to node (4).
... in order of capacity of the edges, then one should first have a flow of 8 units to node (4), next a flow of the remaining 2 units to node (3), and nothing to node (2).

So, how is the initial guess constructed? Thank you in advance! — MFH:Talk 21:52, 22 April 2022 (UTC)[reply]