pseudoPython

[Gareth McCaughan]

Duncan Smith wrote:

>                          I seem to have forgotten to post my last function.
> Just wondering whether a non-Pythonista would be able to follow this as
> 'tis, or whether I should be popping and pushing the cliques (pointers to
> cliques) to a stack, or something.
..
> graphThin(DJT):
>
> cliques = a list of cliques in elimination order
>
> while cliques:
>     nextClique = cliques[-1]
>     cliques = clique[:-1]
>     cliques = cliques + cliqueThin(nextClique)
>
>
> or maybe,
>
>
> graphThin(DJT):
>
> cliques = a list of cliques in elimination order
>
> while cliques:
>     nextClique = cliques.pop()
>     for newClique in cliqueThin(nextClique):
>         cliques.append(newClique)

graphThin(DJT):

    cliques = a list of cliques in elimination order

    while cliques is not empty:
        nextClique = last element of cliques
        delete last element of cliques
        for newClique in cliqueThin(nextClique):
            append newClique to cliques

is surely clearer. Instead of "while cliques is not
empty", you could use a phrasing you used before:
"while there are cliques". I don't like it much,
though :-).

I think it's worth indenting the body of each function
as well as the structures within the functions, as I've
done here.

Returning briefly to the original pseudocode, I think
it's a mistake to use "\" to denote both set-theoretic
difference and line continuation. I suggest "..." for
the "continued on next line" meaning. And you probably
want to replace "cliques.append(Cv)" with "append Cv to
cliques", not that it matters much. I'm also puzzled
by the inner loop in cliqueThin, which goes like this:

    for each edge in thinnableEdges:
        if edge is incident to v:
            remove edge \
            from thinnableEdges
        elif edge is in Cv:
            remove edge \
            from thinnableEdges

Why not

    for each edge in thinnableEdges:
        if edge is incident to v, or in Cv:
            remove edge from thinnableEdges

instead? Er, and is mutating thinnableEdges meant
to have a lasting effect? I mean, does "remove edge
from thinnableEdges" actually alter the structure
of the graph? If not, then this whole loop could
be deleted :-). If it does, then it might be worth
flagging the fact more explicitly when you say
"thinnableEdges = thinEdges(clique)". Anyway, I'm
now quibbling about style, which is pretty stupid
since I haven't read the accompanying text that may
make sense of everything that's puzzling me :-),
so I'll stop.

-- 
Gareth McCaughan  Gareth.McCaughan at pobox.com
.sig under construc