Author: John Merlino
Date: 09:08:09 11/09/01
Go up one level in this thread
On November 09, 2001 at 10:44:53, William Penn wrote: >On November 09, 2001 at 10:13:16, Jonas Cohonas wrote: > >>On November 09, 2001 at 10:11:14, Uri Blass wrote: >> >>>On November 09, 2001 at 08:20:36, Jonas Cohonas wrote: >>> >>>>On November 09, 2001 at 07:26:21, William Penn wrote: >>>> >>>>>On November 09, 2001 at 04:28:37, Chessfun wrote: >>>>> >>>>>[snip] >>>>> >>>>>>I agree, I think Fritz 7 is a great engine. >>>>> >>>>>[snip] >>>>> >>>>>Are you overlooking the analysis bugs? I continue to find examples of stupid >>>>>blunders involving Queens when using the Fritz 7 engine (the one with MMX) in >>>>>Infinite Analysis mode. It attempts to give the Queen away for free, which >>>>>opponent ignores! >>>>>WP >>>> >>>>Could you post the positions in question? >>>> >>>>Regards >>>>Jonas >>> >>>He already did it >>> >>>see http://www.icdchess.com/forums/1/message.shtml?195455 >>> >>>Uri >> >>Thank's Uri >> >>Regards >>Jonas > >That's one of three such obvious blunders with the Queens in Fritz 7 (with MMX) >Infinite Analysis that I've noticed. So far I've only noticed it in positions >involving active Queen play. I analyze lots of positions and haven't tried to >keep track of the particular positions. I imagine if you pick out almost any >position involving Queen(s) and Pawns and pieces with active play, and analyze >it in Infinite Analysis mode with that engine, you will get some fairly obvious >goofs too. Of course I don't know, but at least it is goofy on my >system/installation. [W98se, Deep Fritz GUI, Celeron 500, 256MB RAM] I have not >seen such serious analysis errors with any other engine, and I've tried quite a >few of them in different versions (Fritz, Hiarcs, Shredder, Junior). Someone >suggested that the analysis may be faulty beyond the first half-move, but that >the engine is actually working OK to find the best move. Maybe, maybe not, I >have no idea. But my main interest is accurate analysis, so it is a serious >fault to me. >WP When looking at any analysis output, I generally am very guarded about trusting any moves beyond the reported search depth. So, for Chessmaster (for example), if the search depth says 7/11, but the PV is 18 moves long (as it is in the example below, which is based on the position quoted above), I would only consider the first 7 moves to be "best play". However, I would EXPECT that there would not be any OBVIOUS BLUNDERS in any moves past those first seven. Chessmaster 8000 (default personality) on a PIII-733: Time Depth Score Positions Moves 0:00 2/5 1.86 10839 1. Qb5 Qb6 2. Qe8 Nf6 3. Qxf7 0:00 2/6 1.89 35986 1. Qb5 Kd6 2. Bxd5 exd5 3. Qxb4+ Kc6 4. Qc3+ Kb6 5. Qxg7 Kxa6 6. Qxh6+ Kb5 0:01 3/7 1.89 81636 1. Qb5 Kd6 2. Bxd5 exd5 3. Qxb4+ Kc6 4. Qc3+ Kb6 5. Qxg7 Kxa6 6. Qxh6+ Kb5 0:02 4/8 2.25 253205 1. Qb5 Kd6 2. Bxd5 exd5 3. Qxb4+ Kc6 4. Qc3+ Kb6 5. Qxg7 Kxa6 6. Qxh6+ Kb5 7. h4 0:07 4/8 3.64 800266 1. Bxd5 Kxd5 2. Qb5+ Kd6 3. Qxb4+ Kc6 4. Qa5 g5 5. b4 g4 0:12 5/9 3.83 1285851 1. Bxd5 Kxd5 2. Qb5+ Kd6 3. Qxb4+ Kc6 4. Qa5 Kd6 5. b4 f6 6. Qd8+ Kc6 0:24 6/10 3.74 2911414 1. Bxd5 Kxd5 2. Qb5+ Kd6 3. Qxb4+ Kc7 4. Qa5+ Kd6 5. Qb5 e5 6. Qd3+ Ke6 7. g4 g6 1:14 7/11 3.79 9232747 1. Bxd5 Kxd5 2. Qb5+ Kd6 3. Qxb4+ Kc7 4. Qa5+ Qb6 5. Qe5+ Kd7 6. Qa1 Qa7 7. Qa4+ Kd6 8. Qb5 f5 4:08 8/12 3.69 30876535 1. Bxd5 Kxd5 2. Qb5+ Kd6 3. Qxb4+ Kc7 4. Qa5+ Qb6 5. Qe5+ Kc8 6. Qxg7 Qxa6 7. Qf8+ Kc7 8. Qxf7+ Kd6 9. Qf4+ Kc5 10. Qxh6 Qa1+ 11. Kg2 Note that White has won three pawns (four for one) in the final PV. This may not be "best play" for Black, but I'm pretty confident that White should win. jm
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.