From: Colin Paul Adams <colin <at> COLINA <dot> DEMON <dot> CO <dot> UK>
Date: 3 feb 2000
Subject: CHU: revised solution to A4, after 325 years

It appears that the published solution to A4 is not correct.

I worked through the published solution, and determined that the Lion
mate mentioned occurred on move 13 (using the Japanese count
throughout). I accordingly set the search depth to the computer to 14,
but it said there was no mate within 14 moves.

So I increased the depth to 16. This time, in less than a minute, it
came up with the following beautiful mate in 15:

1. +DH x 8b,        BT - 10b
2. +DH x 10b,       C x 10b
3. +FL x 9d,        K - 12c
4. +Ky x 10f x 11e, P x 11e
5. +VM - 8g,        VM - 11d
6. FK - 7c,         C - 11c (or C - 9c/10c, 7. FK x 9c/10c Mate)
7. +VM x 11d,       K x 11d
8. FK - 10c Mate.

Now ain't that something?

Now for the sidelines:

If 1. ..., BT 8 b
2. +VM x 11c,       K x 11c
(if K - 12c, then - get this -
    FK - 10a, K x 11c, FK - 11a, K - 12c, FK - 10a Mate (by the
    repetition rule!))
3. FK - 11a,        K - 12c (K - 10c?, +Ky x 10f - 9f Mate)
4. +Ky x 10f x 11e, P x 11e
5. +FL - 10e,       C - 11d
6. FK x 11d Mate

If 2. ..., K - 12c,
3. FK - 10a then Mate next move

If 2. ..., K x 10b,
3. +VM x 11c, K x 11c,
   if K - 9b,
   4. +VM - 9a, K - 10a
   if 4. ..., K - 10c,
      5. +Ky x 10f - 9e Mate)
   5. +VM - 11c is Mate (K - 9b would be repetition).
4. FK - 11a etc.


If 3. ..., C - 10c
4. +VM x 11c, K x 11c
   if K - 12c,
   5. FK - 10a, then +Ky x 10f - 9e Mate
5. +Ky x 10f - 9e, K - 11b/12c
6. +Ky x 10c Mate

If 3. ..., VM - 10c
4. +FL x 10c, C x 10c
   if K x 10c,
   5. +Ky x 10f - 9e, and lion Mate next move
   if K - 12c,
   5. +Ky x 10f x 11e, P x 11e
   6. FK - 7h Mate
5. +VM - 11c, K x 11c
   if K - 12c
   6. FK - 10a, K x 11c 7. +Ky x 10f - 9e Mate
6. +Ky x 10f - 9e, K - 11b
   if K - 12c, 7. +Ky x 10c Mate
   if K - 10b, 7. +Ky - 9c Mate
7. +Ky x 8d - 9c, K - 12c
8. +Ky x 10c Mate

Well, I thought that was really something!

But what is wrong with the published solution?
Well, 1. ..., K - 12c isn't a refutation, as the computer promptly
finds a mate in a further 15 moves (but that needs to be published to
make the solution complete! - there are several minor sidelines as
well that result in quick mates). But this is of course a longer mate
than the computer finds, so is not optimal.

However, in response to 3. +DH x 11e, K - 11b and there is no Mate!

I think, in the light of all this, that this is a really first class
problem - it appears to have withstood all efforts for 325 years!
--
Colin Paul Adams
Preston Lancashire