Home * Search * Alpha-Beta * Fail-Soft

Fail-Soft is a term related to an Alpha-Beta like search. Returned scores might be outside the bounds:

• an upper bound less than alpha at All-Nodes
• a lower bound greater than beta at Cut-Nodes

History

In his 1983 paper Another optimization of alpha-beta search ^[1], John Philip Fishburn introduced Fail-Soft Alpha-Beta as an improvement of Fail-Hard without any extra work ^[2]. Fail-Soft has the reputation for searching less nodes than Fail-Hard, but might also require some care regarding to search instability issues in conjunction with transposition tables and various pruning-, reduction- and extension techniques.

Mate Scores

In Chrilly Donninger's initial null move pruning implementation there was a deep search extension ^[3], if the null move was refuted by a mate attack, thus relying on Fail-Soft of a null window search, where many "random" moves may refute the null-move with or without score in the mate range.

See also

• Fail-Hard
• Fail-High
• Fail-Low
• Principal Variation Search
• NegaScout
• NegaC*
• MTD(f)

Publications

• John Philip Fishburn (1983). Another optimization of alpha-beta search. SIGART Bulletin, Issue 84, pdf

Forum Posts

1995 ...

• New improvement to alpha/beta + TT? by Heiner Marxen, rgcc, January 13, 1997
• Fail-soft with PVS? by Will Singleton, CCC, March 09, 1999 » Principal Variation Search
• What is the advantage of fail-soft? by Bas Hamstra, CCC, August 25, 1999

2000 ...

• Fail-soft or Fail-hard ? by Teerapong Tovirat, CCC, November 04, 2000 » Fail-Hard
• Alpha beta fail soft, pruning & hash bounds? by Steve Maughan, CCC, December 14, 2000 » Pruning, Bound
• Fail Soft Alpha Beta & Transpositions by Orhan Öztürk, rgcc, September 3, 2003
• (Fail soft) alpha beta by Delphi, rgcc, September 4, 2003
• Fail soft alpha-beta by Russell Reagan, CCC, September 08, 2003
• Is this a correct fail-soft? by Mikael Bäckman, CCC, September 21, 2003
• My fail soft reduces quality of collected PV. Help needed by Volker Böhm, CCC, April 20, 2004
• Fail-hard, fail-soft question by Dan Honeycutt, CCC, May 06, 2004

2005 ...

• Fail Soft question by David B. Weller, CCC, February 17, 2006
• Search questions by Sven Schüle, Winboard Forum, July 17, 2007 » Mate Distance Pruning, Principal Variation Search
• mate detetion issue by Mike Adams, CCC, October 24, 2008 » Checkmate, Connect Four
• fail soft vs fail hard by cyberfish, CCC, November 19, 2008
• Return eval or upper bound? by John Merlino, CCC, October 26, 2009

2010 ...

• First post (and FailHigh question!) by Natale Galioto, CCC, June 14, 2013 » Fail-High, Fail-Hard
• Fail soft vs fail hard by Sergei S. Markoff, CCC, February 15, 2014 » Fail-Hard, Fail-Low, Internal Iterative Deepening
• Implications of Lazy eval on Don Beal effect in Fail Soft by Henk van den Belt, CCC, November 19, 2014 » Lazy Evaluation

2015 ...

• Multi-cut and fail-soft by Matthew R. Brades, CCC, June 30, 2016 » Multi-Cut

External Links

• Fail soft from Wikipedia
• Soft Machine - Hazard Profile (1975), YouTube Video
  Lineup: Mike Ratledge, John Marshall, Karl Jenkins, Roy Babbington, Allan Holdsworth
  
• Ian Carr's Nucleus - Song For The Bearded Lady (1970), YouTube Video
  Lineup: Ray Russell, Ian Carr, Brian Smith, Karl Jenkins, John Marshall, Roy Babbington
  

References

1. ^ John Philip Fishburn (1983). Another optimization of alpha-beta search. SIGART Bulletin, Issue 84, pdf
2. ^ Jean-Christophe Weill (1991). Experiments With the NegaC* Search - An Alternative for Othello Endgame Search. Heuristic Programming in AI 2
3. ^ Chrilly Donninger (1993). Null Move and Deep Search: Selective-Search Heuristics for Obtuse Chess Programs. ICCA Journal, Vol. 16, No. 3

What links here?

Up one Level