computer-go: new address of GoTools

[Thomas Wolf <twolf@xxxxxxxxxxxxxxxxxxx>]

The two replies below are late, as they bounced back
(my email address had changed). So here again.

Thomas
=====================================================================
>>    GoTools Online has moved to:
>>    http://lie.math.brocku.ca/gotools/applet.html
>>    Please update your bookmarks!
>
>First question - why is it so much slower now?  The problem shown at
>http://www.britgo.org/links/brits.html#wolfth used to take it one
>second.  Now it thinks for five minutes and then gives up.

As described in all documentation to GoTools, this program is only
efficient for closed problems. It may solve open problems but no
guarantee. If you add a black stone at B9 the problem is still open
but sufficiently closed to be solved in under 1 second.

>Second question - where is Thomas's web site now?  The link given from
>the above page no longer works.

The GoTools home page address has always been

http://www.qmw.ac.uk/~ugah006/gotools/t.wolf.html

whereas the link from

http://www.britgo.org/links/brits.html#wolfth

points at

http://www.qmw.ac.uk/~ugah006/gotools/t.wolf.html/

which is clearly wrong. The address to be used in
future will be

http://lie.math.brocku.ca/GoTools

which is also working now.

>Best wishes,
>          Nick
>--
>Nick Wedd    nick@xxxxxxxxxxxxxxxxx

Thomas.
=====================================================================
>> The odd thing is, a while ago I heard that GoTools had become much
>> better at handling unenclosed positions.
>> I tested it on that one at
>> http://www.britgo.org/links/brits.html#wolfth exactly as it is, and
>> found that it solved it in a second.
>
>I suppose Thomas has improved the support for unclosed positions, which
>means it works better but slower. :-(

I have not worked on the open position issue in the last months (just
fixing bugs). If one really want to solve open positions with the
current version  of GoTools then one has to increase the search depth
successively, as described in the help and FAQ subpages of the
applet. The only reason that it solved the problem once and not now
that I can see is that after fixing some bugs related to the linking
issue, or other related bugs, it is now more carefully and
investigates moves which it did not before. But as said in the
previous email, one does not have to close the mentioned problem
rigidly. Just a single black stone on B9 closes it enough to be solved
quickly. 
  Another comment is that this problem is not computationally expensive
and there are probably a number of programs that can solve it under 1 sec.
Carpender Square (but fully enclosed) in 5 sec is not so easy to beat.

Thomas Wolf