I don't have any performance information. First, my OS is constantly running background programs that I can't control -- especially security software. This impacts my wall-clock execution timing. Secondly, the prioritization algorithm in my solver already flags freecell-to-empty_column as a "low priority" move. As such, any attempt for a comparison would be skewed because the control run would already be biased against such moves.
What I do know is that I perform considerably more operations in my freecell-to-empty_column logic using your constraint. This has to be a drawback on execution time because this routine is executed every time I process a layout.
---In fc-solve-discuss_at_yahoogroups.com, <shlomif_at_...> wrote:
Hi Danny,
thanks for the investigation [and the text portion of this E-mail came out
fine - maybe Yahoogroups fixed something on their end.]
On 16 Oct 2013 18:14:23 -0700
<dannyjones183_at_... mailto:dannyjones183_at_...> wrote:
> It's been awhile since I've altered my FreeCell software, so it's possible
> that I made a mistake in my current effort. Take it with a grain of salt.
>
> I altered my move_freecell_to_column routine to come close to what Shlomi
> proposed. A card was blocked from being moved from a freecell to an empty
> column if a subsequent single-card on top of it wasn't possible. I did not
> check for multi-card moves on top of it. Here's my results for the first 1M
> deals/games. There were only two deals/games added to my list of unsolvables.
>
> _at_00011982 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 42577
> _at_00146692 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 7044
> _at_00186216 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 14250
> _at_00255317 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 15230 ***
> _at_00455889 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 9443
> _at_00495505 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 25857
> _at_00512118 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 17542
> _at_00517776 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 42093
> _at_00739671 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 3713 ***
> _at_00781948 Attempt: 1 NumFcs=4 (WKR Super) 0 moves 2933
>
All I see here is the list of unsolvable deals. Can you share some other
statistics such as how much that prune increased or decreased performance?
(Unless I missed something with the E-mails.).
In any case, in this branch:
https://bitbucket.org/shlomif/fc-solve/branch/prune-of-dont-move-cards-in-freecells-to-empty-columns https://bitbucket.org/shlomif/fc-solve/branch/prune-of-dont-move-cards-in-freecells-to-empty-columns
I started to implement the safe version of the prune into the atomic move
functions, and it made absolutely no difference in the iterations count (but
possibly made things slower wallclock-time-wise). So I didn't continue working
on that branch.
Regards,
Shlomi Fish
--
-----------------------------------------------------------------
Shlomi Fish
http://www.shlomifish.org/ http://www.shlomifish.org/
Best Introductory Programming Language -
http://shlom.in/intro-lang http://shlom.in/intro-lang
The X in XSLT stands for eXtermination.
—
http://www.shlomifish.org/humour/bits/facts/XSLT/ http://www.shlomifish.org/humour/bits/facts/XSLT/
Please reply to list if it's a mailing list post -
http://shlom.in/reply http://shlom.in/reply .
Received on Thu Oct 17 2013 - 14:37:26 IDT