| View previous topic :: View next topic |
| Author |
Message |
daj95376
Joined: 23 Aug 2008
Posts: 3854
|
 Posted: Sun May 30, 2010 4:27 pm Post subject: Puzzle 10/05/30: (C) Advanced |
 |
|
Probably a BBDB as well.
| Code: | +-----------------------+
| 6 . . | . 3 . | . . . |
| . 5 . | 8 . . | 6 . . |
| . . 3 | . . 6 | 7 . . |
|-------+-------+-------|
| . 4 . | . 5 . | 8 . . |
| 7 . . | 4 2 . | . . 5 |
| . . 5 | . . 7 | . . . |
|-------+-------+-------|
| . 9 4 | 3 . . | . 7 . |
| . . . | . . . | 2 . 4 |
| . . . | . 7 . | . 6 1 |
+-----------------------+
|
Play this puzzle online at the Daily Sudoku site |
|
| Back to top |
|
 |
peterj
Joined: 26 Mar 2010
Posts: 974
Location: London, UK
|
| Posted: Sun May 30, 2010 7:12 pm Post subject: |
|
|
First time through as played - four steps with a nice UR
| Quote: | x-wing(2); r1c6<>2, r369c1<>2
type 6 UR(49); r2c1<>4, r3c5<>4
kite(9); r4c6<>9
xy-chain (9=3)r4c1-(3=1)r4c6-(1=2)r7c6-(2=9)r2c6; r2c1<>9 |
Went back to see if my last step could be used earlier - two steps..
| Quote: | x-wing(2); r1c6<>2, r369c1<>2
AIC (9=3)r4c1-r6c1=r6c7-(3=1)r5c7-(1=8)r5c3-(8=2)r9c3-(2=9)r1c3; r23c1<>9 |
|
|
| Back to top |
|
|
Mogulmeister
Joined: 03 May 2007
Posts: 1151
|
| Posted: Mon May 31, 2010 7:18 am Post subject: |
|
|
| Quote: | A different 2 step:
1. Saw the swordfish which eliminates six 2's
2. (8)r8c6=(8-3)r5c6=r5c7-r6c7=(3-8)r6c1=r6c5-r8c5=(8)r8c6; r8c6 is 8 |
|
|
| Back to top |
|
|
daj95376
Joined: 23 Aug 2008
Posts: 3854
|
| Posted: Mon May 31, 2010 8:30 am Post subject: |
|
|
Sometimes, I can't get to sleep and come up with crazy thoughts that probably should have been forgotten. An example ...
Consider the following four rows and a way to use ANP() to get a strong link on two different values in two different cells.
| Code: | +-----------------------------------------------+
| . . . | . . . | . . . |
| . 23 . | . 23 . | . 23 . | 3 cells and 2 values
| . . . | . . . | . . . |
|---------------+---------------+---------------|
| . 123 . | . 23 . | . 23 . | 3 cells and 3 values
| . . . | . . . | . . . |
| . 23 . | . 23 . | . 234 . | 3 cells and 3 values
|---------------+---------------+---------------|
| . 123 . | . 23 . | . 234 . | 3 cells and 4 values where
| . 123 . | . 23 . | . 24 . | two values appear only once
| . 123 . | . 23 . | . 34 . | and in different cells
+-----------------------------------------------+
row 2: invalid
row 4: ( 23)r4c58 - (23=1)r4c2 => Naked Pair forces a Naked Single
row 6: ( 23)r6c25 - (23=4)r6c8 => Naked Pair forces a Naked Single
row 7: (1=23)r7c25 - (23=4)r7c8 => SL equivalent: (1)r7c2 = (4)r7c8
row 8: (1=23)r8c25 - (2 =4)r8c8 => SL equivalent: (1)r8c2 = (4)r8c8
row 9: (1=23)r9c25 - ( 3=4)r9c8 => SL equivalent: (1)r9c2 = (4)r9c8
|
Now, why would anyone want the SL equivalent when the original chain segment will always work? Someone who has a chain() routine in their solver that only allows for two single-digit values to be stored for a strong link.  _
BTW: I used this logic to crack Puzzle 10/05/30: (C) Advanced in one step.  _ |
|
| Back to top |
|
|
peterj
Joined: 26 Mar 2010
Posts: 974
Location: London, UK
|
| Posted: Mon May 31, 2010 9:40 am Post subject: |
|
|
Danny, I tried hard to find a one-step here so I had to go back to look after reading your post...
Is this it?! Very nice!
| Quote: | (8=19)r45c3-(9=3)r4c1-(3=19)r4c6,r6c4-(19=8)r6c5; r5c6<>8, r6c12<>8
or in Danny chain() solver terms...
(8)r5c3=(3)r4c1-(3)r4c6=(8)r6c5 |
[Edit(daj): added missing cells to first chain.] |
|
| Back to top |
|
|
daj95376
Joined: 23 Aug 2008
Posts: 3854
|
| Posted: Mon May 31, 2010 3:42 pm Post subject: |
|
|
Peter: Congratulations! You almost matched my results exactly. I think yours is more interesting because you used the approach twice. (note: I added 2x missing cells to your first chain.) Nice going on the second version of the chain.
| Code: | (8=19)r45c3 - (9=3)r4c1 - r4c6 = (3)r5c6 => r5c6<>8
(8 )r 5c3 = ( 3)r4c1 - r4c6 = (3)r5c6 => r5c6<>8 SL equivalent version
+--------------------------------------------------------------+
| 6 128 1289 | 7 3 1259 | 4 58 29 |
| 249 5 7 | 8 49 29 | 6 1 3 |
| 2489 128 3 | 1259 149 6 | 7 58 29 |
|--------------------+--------------------+--------------------|
| 39 4 19 | 6 5 139 | 8 2 7 |
| 7 6 189 | 4 2 1389 | 139 39 5 |
| 2389 1238 5 | 19 189 7 | 139 4 6 |
|--------------------+--------------------+--------------------|
| 12 9 4 | 3 6 12 | 5 7 8 |
| 135 7 6 | 159 189 1589 | 2 39 4 |
| 2358 238 28 | 259 7 4 | 39 6 1 |
+--------------------------------------------------------------+
# 70 eliminations remain
|
|
|
| Back to top |
|
|
Mogulmeister
Joined: 03 May 2007
Posts: 1151
|
| Posted: Mon May 31, 2010 5:57 pm Post subject: ALMOST ALMOST ALS |
|
|
I had that <189> ANP but didn't follow that particular path round! Nice one chaps. That ANP proves that r5c3 is 8, however, I was looking at it from a different structure - the almost ALS XZ. (see diagram with apologies for the double "almost" but that's not new!  )
Everything hinges around that same 8 in r5c3
X ={1,2,3,8,9} shown in green and in block 4.
Z ={1,3,8,9} shown in orange and in block 5.
a) IF that 8 at r5c3 is false then the ALS XZ is TRUE and restricted common is 8 and the common candidate is the 3 in r4c1 which is eliminated (in blue) and puzzle solved
b) IF the 8 at r5c3 is true puzzle is solved
PS - Yes I know that the contradiction in scenario (a) is that a locked pair <19> is formed if r5c3 <> 8 which in turn should mean that r4c1 <> 9. Either way, there is an inherent contradiction between the ALS hypothesis and the locked pair which indicates that the hypothesis r5c3 <> 8 is also false!! 
Last edited by Mogulmeister on Tue Jun 01, 2010 5:39 am; edited 2 times in total |
|
| Back to top |
|
|
Mogulmeister
Joined: 03 May 2007
Posts: 1151
|
| Posted: Tue Jun 01, 2010 4:51 am Post subject: |
|
|
| This is another example of the AALS that Ronk has referred to previously? |
|
| Back to top |
|
|
daj95376
Joined: 23 Aug 2008
Posts: 3854
|
| Posted: Tue Jun 01, 2010 6:15 am Post subject: |
|
|
Forcing Network: r5c6=8 r4c6=3 ... leaving contradiction in cells r4c1, r4c3, and r5c3
| Code: | +--------------------------------------------------------------+
| 6 128 1289 | 7 3 1259 | 4 58 29 |
| 249 5 7 | 8 49 29 | 6 1 3 |
| 2489 128 3 | 1259 149 6 | 7 58 29 |
|--------------------+--------------------+--------------------|
| 9 4 19 | 6 5 3 | 8 2 7 |
| 7 6 19 | 4 2 8 | 139 39 5 |
| 2389 1238 5 | 19 19 7 | 139 4 6 |
|--------------------+--------------------+--------------------|
| 12 9 4 | 3 6 12 | 5 7 8 |
| 135 7 6 | 159 189 159 | 2 39 4 |
| 2358 238 28 | 259 7 4 | 39 6 1 |
+--------------------------------------------------------------+
|
Too bad it qualifies as T&E. However, it does show the critical ALS cells that I used.
Last edited by daj95376 on Tue Jun 01, 2010 6:29 am; edited 1 time in total |
|
| Back to top |
|
|
Mogulmeister
Joined: 03 May 2007
Posts: 1151
|
| Posted: Tue Jun 01, 2010 6:26 am Post subject: |
|
|
Danny, even when the contradiction is expressed as:
(8-3)r5c6=r4c6-(3=19)r4c13-(19=8)r5c3-(8)r5c6; contradiction so r5c6 <>8 ?
Excuse notation but It looks like another discontinuous loop....
[edit to include missing cell ref] |
|
| Back to top |
|
|
daj95376
Joined: 23 Aug 2008
Posts: 3854
|
| Posted: Tue Jun 01, 2010 6:50 am Post subject: |
|
|
| Mogulmeister wrote: | Danny, even when the contradiction is expressed as:
(8-3)r5c6=r4c6-(3=19)r4c13-(19=8 )r5c3-(8)r5c6; contradiction so r5c6 <>8 ?
It looks like another discontinuous loop.... |
Yes, and that's the hook that draws everyone into using ALS and ANP().
However, if you examine the (3=19)r4c13-(19=8 )r5c3 logic closely, you'll see that the subset operation (=19)r4c13-(19)r5c3 is really a forcing network in disguise. All subset operations are really forcing network logic. So, what you have is an embedded forcing network in the middle of what looks like a discontinuous chain.
I've learned to accept the use of subsets in chains because they are a natural extension to chain logic. I also accept ALS and UR segments, even though I consider them less natural.
Last edited by daj95376 on Tue Jun 01, 2010 7:17 am; edited 1 time in total |
|
| Back to top |
|
|
Mogulmeister
Joined: 03 May 2007
Posts: 1151
|
| Posted: Tue Jun 01, 2010 7:13 am Post subject: |
|
|
Understood - I watched many debates unfold on T&E (due to my background I wondered what time and expenses had to do with anything) and stayed out of them as some of the raciocination in the debates descended to the level of "angels on a pinhead".
Like you, I am happy to use the interesting ANPs, ALSs but I won't be worrying if I am not restoring the table back to the natural wood. |
|
| Back to top |
|
|
daj95376
Joined: 23 Aug 2008
Posts: 3854
|
| Posted: Tue Jun 01, 2010 7:34 am Post subject: |
|
|
Everyone: Please forgive my soap-box rantings on chains vs. forcing networks.
When I wrote my first Sudoku solver, I missed an important constraint in the definition of a chain. What resulted, after much effort, was a useless chain() routine that was no better than forcing network logic. Since then, I've been very sensitive to what's being called a chain.
Regards, Danny |
|
| Back to top |
|
|
peterj
Joined: 26 Mar 2010
Posts: 974
Location: London, UK
|
| Posted: Tue Jun 01, 2010 7:55 am Post subject: |
|
|
For me it all boils down to the mental attitude I had when I went looking for a 'chain'. If I spot a pattern that looks like a ANP/ANT/ALS etc or a useful crossover between strong links on single digits and then see that I can use it in a chain to disprove something useful... I'm happy.
If however I have descended into mentally plugging numbers into cells in promising parts of the grid and following the consequences ... it's really T&E.
I'm sure I am not alone in having sufficient self-awareness to know when I am really just guessing! |
|
| Back to top |
|
|
Mogulmeister
Joined: 03 May 2007
Posts: 1151
|
| Posted: Tue Jun 01, 2010 8:40 am Post subject: |
|
|
| Speaking purely personally, it's always good to hear your POV Danny and those of the others here. From what I've seen thus far I'd say this is a pretty broad church which is not always true at some other places I have visited. |
|
| Back to top |
|
|
ronk
Joined: 07 May 2006
Posts: 398
|
| Posted: Tue Jun 01, 2010 11:28 am Post subject: |
|
|
| Mogulmeister wrote: | | This is another example of the AALS that Ronk has referred to previously? |
Not in this case. Even when candidate <8> exists at r5c3, there are only five candidate values in your four cells.
I see your deduction as an ALS-xy-wing with endpoint overlap, where the overlap is illustrated below with bicolor cells. Using some of this forum's AN[PTQ] lingo ...
ANP(19=3)r4c13 - ANT(3=189)[r4c6,r6c45] - ANQ(8=1239)[r4c13,r6c12] --> r5c3<>19
It avoids such awkwardness as ...
a) IF that 8 at r5c3 is false then the .... puzzle solved
b) IF the 8 at r5c3 is true puzzle is solved |
|
| Back to top |
|
|
tlanglet
Joined: 17 Oct 2007
Posts: 2468
Location: Northern California Foothills
|
| Posted: Tue Jun 01, 2010 1:23 pm Post subject: |
|
|
A two step solution without chains or forcing networks........
w-wing 39 r4c1 & r5c8 SL3 r45c6; r5c3<>9
Flightless xy-wing -128 with vertex 28 in r9c3 plus transport: (1)r5c3 - r4c3 = (1)r4c6; r7c6<>1
Ted |
|
| Back to top |
|
|
peterj
Joined: 26 Mar 2010
Posts: 974
Location: London, UK
|
| Posted: Tue Jun 01, 2010 1:52 pm Post subject: |
|
|
| Ted, lovely xy-wing - a satisfying find! |
|
| Back to top |
|
|
Mogulmeister
Joined: 03 May 2007
Posts: 1151
|
| Posted: Tue Jun 01, 2010 2:46 pm Post subject: |
|
|
| Mogulmeister wrote: | | This is another example of the AALS that Ronk has referred to previously? |
| Ronk wrote: | | Not in this case. Even when candidate <8> exists at r5c3, there are only five candidate values in your four cells. |
Thanks - it remains an ALS and the extra "almost" not required. It is the presence or otherwise of the 8 that determines if the ALS XZ works.
| Ronk wrote: | | I see your deduction as an ALS-xy-wing with endpoint overlap, where the overlap is illustrated below with bicolor cells. |
V Interesting - see what you mean.
ANP(19=3)r4c13 - ANT(3=189)[r4c6,r6c45] - ANQ(8=1239)[r4c13,r6c12] --> r5c3<>19
I was also interested in the inherent logical contradiction that says:
The ALS XZ when true (ie without an 8 in r5c3) implies that r4c1 <>3 AND that r4c1 <> 9 due to the locked pair. This contradiction implies that the putative ALS XZ is False and so 8 must be in r5c3 solving the puzzle. |
|
| Back to top |
|
|
Marty R.
Joined: 12 Feb 2006
Posts: 5770
Location: Rochester, NY, USA
|
| Posted: Tue Jun 01, 2010 10:54 pm Post subject: |
|
|
I finished up with Ted's flightless XY-Wing, but before that I had difficulties.
X-Wing (2), ERs (1,1), Multi-coloring twice (2,2) and Type 6 UR (49) before the ending XY-Wing. |
|
| Back to top |
|
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
