... by: Norman Powell
Unsolvable - by Deterministic methods.Similar opening, with a few more eliminations when reaching Forcing Networks:
Hidden Pair in Box 3 for values 7 8 in Cells B7 C9, eliminate other values in cells.
X-Cycles Value 7 in D3 has weak links to both ends of the chain - eliminate.
Forcing Networks: Eliminate 4 in B4, 8 in D6 and 1 in H9.
This is the end of the deterministic eliminations, the next step is a trial by error technique.
I have implemented a Bowman's Bingo, based on Forcing Networks which can solve any legitimate sudoku, so I use it as a last resort. It picks the first value of the first cell with a pair (or cell with the least number of possibilities if a pair is not available. Applies a Forcing Network until it runs out of values to switch on or off or a contradiction is found, then picks the first value in the cell the first cell with the least number of possibilities. It continues making trial guesses, until a contradiction is found, which it uses to inform the values on the previous level. It continues going up and levels making trial guesses until either a contradiction is established with the base guess, which it then eliminates. Or if the first guess happened to be part of the solution, no elimination is found, but the solution is return as part of the search.
In this case Bowman's Bingo returns an elimination of Value 1 in A9, after traversing 5 levels of guesses, making 19 trial guesses across these levels.
Value 2 in A9 is then solved, then 9 in B8.
Value 3 is confined in Row G in Box 8 eliminate 3 in 3 in G6.
The next Forcing Network after suggesting a number of eliminations finds the solution.
I have modified my Forcing Networks to allow it to run through all the possible ONs and OFFs of all the values before moving on to the next technique or returning to beginning if there are eliminations. Avoids search direction bias. I find that frequently the solution is within reach of a pass.