22

Let us say there is a program such that if you give a partially filled Sudoku of any size it gives you corresponding completed Sudoku.

Can you treat this program as a black box and use this to solve TSP? I mean is there a way to represent TSP problem as partially filled Sudoku, so that if I give you the answer of that Sudoku, you can tell the solution for TSP in polynomial time?

If yes, how? how do you represent TSP as a partially filled Sudoku and interpret corresponding filled Sudoku for the result.

Rodrigo de Azevedo
  • 978
  • 9
  • 20

2 Answers2

30

For 9x9 Sudoku, no. It is finite so can be solved in $O(1)$ time.

But if you had a solver for $n^2 \times n^2$ Sudoku, that worked for all $n$ and all possible partial boards, and ran in polynomial time, then yes, that could be used to solve TSP in polynomial time, as completing a $n^2 \times n^2$ Sudoku is NP-complete.

The proof of NP-completeness works by reducing from some NP-complete problem R to Sudoku; then because R is NP-complete, you can reduce from TSP to R (that follows from the definition of NP-completeness); and chaining those reductions gives you a way to use the Sudoku solver to solve TSP.

D.W.
  • 167,959
  • 22
  • 232
  • 500
25

It is indeed possible to use a general Sudoku solver to solve instances of TSP, and if this solver takes polynomial time then the whole process will as well (in complexity terminology, there is a polynomial-time reduction from TSP to Sudoku). This is because Sudoku is NP-complete and TSP is in NP. But as is usually the case in this area, looking at the details of the reduction isn't particularly illuminating. If you want, you can piece it together by using the simple reduction from Latin square completion to Sudoku here, the reduction from triangulating uniform tripartite graphs to Latin square completion here, the reduction from 3SAT to triangulation here, and a formulation of TSP as a 3SAT problem. However, if you want to understand the idea behind reducing from Sudoku to TSP I think you would be better off studying Cook's theorem (showing that SAT is NP-complete) and a couple of simple reductions from 3SAT (e.g. to 3-dimensional matching) and being satisfied in the knowledge that the TSP-Sudoku reduction is just the same kind of thing but longer and more fiddly.

rlms
  • 520
  • 3
  • 7