This repository contains a downloadable (in-progress) application for solvable Sudoku puzzle challenges.
Originally a class assignment designed to solve predefined Sudoku challenges as a CSP (Constraint Satisfaction Problem), this application has been transformed into a GUI designed for the enjoyment of others. The algorithm used to solve a given Sudoku board is called "backtracking" (see SudokuPuzzleSolver.py).
In Artifical Intellgence, "backtracking is a general algorithm for finding all (or some) solutions to some computational problems, notably constraint satisfaction problems, that incrementally builds candidates to the solutions, and abandons a candidate ("backtracks") as soon as it determines that the candidate cannot possibly be completed to a valid solution" - https://en.wikipedia.org/wiki/Backtracking
The implemented algorithm's accuracy has been verified by comparing its outputted solutions to various Sudoku solvers found online.
Puzzles included are those varying in difficulty provided by the professor as well as randomly generated ones.
Click any tile to make it active. Then enter a number of your choice. To clear a tile: right-click, or press DEL while tile is active.
Sudoku rules explained here.
TL;DR:
- A Sudoku board consists of 9x9 tiles
- All numbers in each row must be unique
- All numbers in each column must be unique
- Diagonal numbers can be identical
The installer is available under releases.
pip install https://github.com/pyinstaller/pyinstaller/archive/develop.tar.gzonto your machine. I strongly recommend consulting outside resources to absolutely ensure that you are doing this correctly.- Change directories to the root of this repository.
- Run the following magic line:
pyinstaller --onefile -w GUI.py
It will generate the .exe under the "dist" folder, so make sure to cut+paste it into the parent directory.
Just a fair warning- pyinstaller can be finicky. If using anaconda, install pyinstaller in the same anaconda virtual environment that you're using with this project.
Also note: include --hidden-import "matplotlib" to the above command if you are contributing and using matplotlib. It behaves very strangely, which may be in part due to it's several dependencies on other libraries.
Your computer requires very minimal processing power to run this application.
However, there may be noticeable lag when the program attempts to solve each puzzle once. As a vague benchmark: the easier puzzles takes roughly 1 second or less to solve on my PC. The hardest single puzzle takes about 5 seconds to solve. This area of performance should be considered moot nonetheless, as the algorithm is only run once and the duration may only last seconds.
- Windows 10: Tested and working
- Linux: Not tested
- MacOS: Not tested