The thesis explores different approaches and algorithms for generating, grading and solving the sudoku grid. The aim of the thesis is to determine which approaches and algorithms are more efficient and how they differ from each other. Mathematical puzzle sudoku is a kind of Latin square, features of which dictate the form of a solution and, consequently, the number of possible solutions to the problem. Generating grids can be done in two opposite ways: generating with filling an empty grid and generating with deleting values from a full grid. Generating a grid is from a computer's perspective similar to solving one. Determining the difficulty level of a grid is a delicate problem, since it is necessary to consider human techniques of solving sudoku.
With the help of listed literature and software solution I confirm, during the thesis, the findings on time complexity of solving sudoku grids and usefulness of grading sudoku grids by using information entropy.