[geeks] Pentominos continued

[microcode at zoho.com]

On Mon, Oct 28, 2013 at 01:15:15PM -0400, Mouse wrote:
> > Deleted the posts since the math went over my head so sorry for
> > screwing up the threading after reconsidering.
> 
> Well, as far as I personally am concerned, no big deal; my list
> threading happens in wetware and doesn't depend on References:. :-)

You use all the cool expressions. I need to buy a copy of your dictionary!

> > Have you tried / is it possible to apply some parallelization stuff
> > like OpenMP to make this faster?  Do you have a multicore box to try
> > it on?
> 
> I have two dual-core machines readily available; I have another
> two-core and a four-core that would require digging them out.  But a
> speedup of small factors, like available core count, isn't going to
> make much difference.  Ten _orders of magnitude_ improvement (not to be
> confused with factor of ten) still wouldn't put the full problem within
> reach of brute force.

Ok. Then maybe the suggestion of Amazon was really on target. 10^34 is not a
big deal compared to some of the stuff I've read was done for $1.50
American.

> > If you would share your algorithm it might be interesting to see what
> > other people could come up with using other languages, tools etc.
> 
> Well, it's all available in git repos. 

I'm not likely to get past the C but if somebody posts a solution in
formulas I could probably whip up something in Fortran eventually and it
would be interesting to try OpenMP since I never have as it's now well
supported, but I don't have any number crunching hardware.

> rectangle is (as it appears in 5x12.out)
> 
> (WLLLLYNFFPPP) (WWZZLYNNFFPP) (TWWZYYVNFXUU) (TTTZZYVNXXXU) (TIIIIIVVVXUU)
> 
> Feed this line to sgraphic and the result is
> 
>  _______________________
> | |_____  | | |_  |_    |
> |_  |_  |_| |_  |  _|___|
> | |___| |_  | | |_| |_  |
> |  ___|___|_| |_|_   _| |
> |_|_________|_____|_|___|
> 
> It can do line-drawing characters as well, for VT-100s or TVI-955s or a
> few others - see the charsets[] array in the source - by running, eg,
> "sgraphic -chars vt100".  There are other options - there isn't any
> real documentation at the moment, but there will be soon; I'm going to
> write up a comment header for sgraphic.c and will commit/push it as
> soon as it's done.

That is really, really cool. Some other interesting stuff could come out of
this. I'm thinking traffic simulations for city streets etc.

> > If it was the math you're after then get back to us after somebody
> > gives you the equations ;-)
> 
> :-)
> 
> I was mostly wondering if there's some work already done that had
> determined what shape had the most solutions, somehow.  Secondarily,
> wondering if anyone had any suggestions for how to approach it.

There are places where people with heads like that hang out. I guess the
links on the wikifeebya page didn't lead to any formulaic solutions?