Solving Sudoku Programmatically


  • Boxes are individual squares
  • Units are complete rows, columns, and 3x3 squares
  • Peers are boxes that belong to the same unit
rows = 'ABCDEFGHI'
cols = '123456789'
def cross(a, b):
# output = []
# for s in a:
# for t in b:
# output.append(s+t)
# return output
return [s+t for s in a for t in b]
boxes = cross(rows, cols)
row_units = [cross(r, cols) for r in rows]
column_units = [cross(rows, c) for c in cols]
square_units = [cross(rs, cs) for rs in ('ABC','DEF','GHI') for cs in ('123','456','789')]
unitlist = row_units + column_units + square_units
units = dict((s, [u for u in unitlist if s in u]) for s in boxes)
peers = dict((s, set(sum(units[s],[]))-set([s])) for s in boxes)

For unsolved boxes, eliminate solved peer box values from possible values:

def eliminate(values):
for box, val in values.items():
if len(val) == 1:
for peer in peers[box]:
values[peer] = values[peer].replace(val, '')

For unsolved boxes in a unit, if a possible digit appears once in a box, then that box is the only choice for that digit:

def only_choice(values):
for unit in unitlist:
for digit in "123456789":
boxes = [box for box in unit if digit in values[box]]
if len(boxes) == 1:
values[boxes[0]] = digit

Repeatedly reduce possible solutions using these two methods, eliminate and only choice, until no progress is made:

def reduce_puzzle(values):
progress = True
while progress:
# Check how many boxes have a determined value
solved_values_before = len([box for box in values.keys() if len(values[box]) == 1])
# Check how many boxes have a determined value, to compare
solved_values_after = len([box for box in values.keys() if len(values[box]) == 1])
# If no new values were added, stop the loop.
progress = solved_values_before != solved_values_after

Once board is reduced as much as possible, get unsolved box with the least number of possible digits and try each digit until board is solved:

def invalid(values):
return len([box for box in values.keys() if len(values[box]) == 0]) != 0
def solved(values):
return len([box for box in values.keys() if len(values[box]) != 1]) == 0
def search(values):
if invalid(values):
return None
if solved(values):
return values
n,s = min((len(values[s]), s) for s in boxes if len(values[s]) > 1)
for value in values[s]:
tmp = values.copy()
tmp[s] = value
tmp = search(tmp)
if tmp:
return tmp

In Sudoku, by eliminating the possibilities in a given box we can eventually narrow down the board to the solution one by one. When solving Sudoku, mentally if there is a digit within a “peer” (horizontal row, vertical row, or 3x3 grid) we know that that digit cannot be repeated again. We also know that if amongst all the open possibilities in a peer, there is a unique digit, that digit must be the solution to that box (Example: If we have four boxes open and the possibilities for those boxes are 123, 234, 234, 24, then we know that from the first box’s possibilities of 123, that 1 is unique, so that can be the only possible solution to that box). By using these two techniques we can reduce the possibilities and partially solve the board.

After reducing the board down using those two techniques, we can further solve the board by trying a possibility for a box and seeing if that leads us down to a solution; if not, we must try the next possibility and so on.

The two parts to solving Sudoku can be referred to as constraint propagation and search.




disscepolo della sperientia

Love podcasts or audiobooks? Learn on the go with our new app.

Recommended from Medium

Building a distributed MIDI Controller with Android Things and Nearby API

FAQ | Simon Says Extension for Final Cut Pro X

60 Practices For Quality Engineering : Skills (Part 5)

was serving people and not himself’

Garbage Collection — Java How it works ?

Complete Guide to Whitelist Lottery & SolRazr IDO


Flipside’s Chainwalkers and The Redshift COPY Command

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Paul Yeo

Paul Yeo

disscepolo della sperientia

More from Medium

Breast Cancer Detection

Rectified Linear Activation Unit (ReLU)

Hard Drive Failure Detection using S.M.A.R.T attributes on Backblaze dataset.

How To Teach Machines To Learn Like Humans | Dataloop Blog