I just finished a major expansion of the Wikipedia article on path decomposition and pathwidth of graphs. As always, I'd be grateful for constructive criticism.

Something that I wasn't able to fit in, because it doesn't seem to be in the published literature: although they aren't the same, there seems to be a close connection between the pathwidth of a tree and its Strahler number, an invariant of trees invented in the 1940s and 1950s by hydrologists to categorize rivers. They're not the same (even ignoring the fact that Strahler number is defined for rooted trees and pathwidth is for unrooted trees) but I think they're always within a constant factor of each other.

While working on this, I came across a bibliographic curiosity that I don't really understand: how and why did Lopez and Law get the same paper published simultaneously in two different IEEE journals? Both versions exist on the IEEE web site (I downloaded pdfs of both, though I was only able to find one doi) and the biggest difference between them seems to be in the lines at the tops of the pages saying which journal they're in. Explained thanks to help from [info]justinwsmith — it was a joint issue of both journals – see comments.

On an unrelated subject, my posting yesterday about per-paper registration fees kicked up quite a few comments. The consensus seems to be that, if it is a scam at all, it's one that a lot of respected conferences are pulling. In my own personal case, the situation is resolved, my co-author has registered to go, and my paper will be there. But my take-away message is that such things need to be clearly stated in the call for papers (or at least somewhere on the conference web site at the time the call for papers becomes available). It's reasonable to assume as a default without saying so explicitly that every paper needs to have a registered co-author. But if you want to have a one-talk-per-speaker policy, or you are using per-paper fees in a small-attendance conference to reduce the variability of the budget, or you are departing from the default assumptions for whatever other good reason, say so. That way potential authors can plan ahead and don't feel like they've been the subject of a bait-and-switch.





Comments:

justinwsmith:
2010-05-06T04:17:43Z

Nice work! (I'm a little tired now, so I'll save a more careful reading for tomorrow.)

That Lopez-Law publication appears suspicious. (Submitting the same paper to two places is, at best, frowned upon.) I think I found the other DOI: 10.1109/T-ED.1980.20086

11011110:
2010-05-06T04:37:07Z

If the authors did it without telling the journal editors, it's definitely frowned upon. But there might be some other reason why it could be a legitimate thing to do, I suppose. For instance, sometimes old journal papers are reprinted (clearly marked as a reprint) in a special issue of another journal that is collecting them for some historical purpose.

justinwsmith:
2010-05-06T04:41:22Z

Ok, there's probably no "funny business" here. The tables of contents for the two issues appear, at first glance, to be the same:
http://ieeexplore.ieee.org/xpl/tocresult.jsp?isnumber=22572
http://ieeexplore.ieee.org/xpl/tocresult.jsp?isnumber=31805

justinwsmith:
2010-05-06T04:46:25Z

Yes. The forwards (e.g., http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1051410) in both say that they are a "joint issue" with the other.

11011110:
2010-05-06T04:48:57Z

Right, I just saw the same thing. So now the only minor mystery is why there are two different DOIs, since a single paper is supposed to have a single DOI. But that can be adequately explained as IEEE not paying attention to that level of detail when assigning DOIs to their publications.

None: Gene
2010-05-10T07:34:32Z

Is there any chance you'll actually update this product for OSX? The website gives a 2003 update that says you are doing it. I used it primarily for it's wonderful chart generating abilities, but no more. I cannot use it any longer since classic no longer is available...