A tree is "efficient" in the sense that it stops investigating branches as soon as it finds they contain an impossible/inconsistent solution (no point to keep checking). I initially said "tree may be inefficient" (see previous comment) because of past experience compared to other algorithms for given problem. To be fair,comparing to other methods mentioned here, I say it is quite efficient stricly as far as the number of times recursive find() is called. Still there may be something faster. Easily checcked via incrementing a global variable upon entry. Nice subject for a competition, in case monks get rasty rusty (oxidised, reacting with oxygen) ...