To be continued

Max Cut Problem

Reference

Given an undirected graph G=(V,E) G=(V, E) where

  • VV is the set of vertices
  • EE is the set of edges with nonnegative edge weights wij=wjiw_{i j}=w_{j i}

Partition the set of vertices of a graph into two categories by setting values

  • If vertex ii belong to Sˉ\bar{S}, then s(i)=1s(i)=-1
  • If vertex ii belong to SS, then s(i)=1s(i)=1

Therefore

  • A cut occurs if two vertices of an edge disagrees, the edge is a cut s(i)s(j)=1s(i) s(j)=-1.
  • if two vertices belong to the same category, the edge is not a cut s(i)s(j)=1s(i) s(j)=1

1.1.1. Cost function

The corresponding Max-Cut Hamiltonian then reads: H^C=121i<jnwij(1sisj) \hat{H}_C= \frac{1}{2} \sum_{1 \leq i<j \leq n} w_{i j}\left(1-s_i s_j\right) The goal is to find a classification that allows the maximum number of cuts to be made maximize121i<jnwij(1sisj) \operatorname{maximize} \frac{1}{2} \sum_{1 \leq i<j \leq n} w_{i j}\left(1-s_i s_j\right) The eigenstate to this Hamiltonian H^C\hat{H}_C with the highest eigenvalue corresponds to the maximum cut. The formulation H^C\hat{H}_C is also called 'Cost function' C=ijCij=121i<jnwij(1sisj) \begin{aligned} C=&\sum_{\langle ij \rangle} C_{\langle ij \rangle}\\ =&\frac{1}{2} \sum_{1 \leq i<j \leq n} w_{i j}\left(1-s_i s_j\right) \end{aligned} Because

  1. if two vertices belong to Sˉ\bar{S}, then s(i)=s(j)=1s(i)=s(j)=-1

Cij=12wij(1sisj)=12wij(1(1)(1))=0 \begin{aligned} C_{\langle i j\rangle}=&\frac{1}{2}w_{i j}\left(1-s_i s_j\right)\\ =&\frac{1}{2}w_{i j}\left(1-(-1)(-1)\right)\\ =&0 \end{aligned}

  1. if two vertices belong to SS, then s(i)=s(j)=1s(i)=s(j)=1

Cij=12wij(1sisj)=12wij(11×1))=0 \begin{aligned} C_{\langle i j\rangle}=&\frac{1}{2}w_{i j}\left(1-s_i s_j\right)\\ =&\frac{1}{2}w_{i j}\left(1-1\times 1)\right)\\ =&0 \end{aligned}

  1. if two vertices belong to different categories

Cij=12wij(1sisj)=12wij(1(1)×1)=wij \begin{aligned} C_{\langle i j\rangle}=&\frac{1}{2}w_{i j}\left(1-s_i s_j\right)\\ =&\frac{1}{2}w_{i j}\left(1-(-1)\times 1\right)\\ =& w_{i j} \end{aligned}

Therefore. the Cost function CC is the sum of weights of cutting edge.

1.1.2. State

In this section, we will use a example of 4 vertices graph to show how to present states of system by Dirac notation.

The ket denote states of system ABCD|ABCD\rangle

  • If vortex AA belong to category SS, the place of AA in ket denote 0, that is 0BCD|0BCD\rangle
  • If vortex AA belong to category Sˉ\bar{S}, the place of AA in ket denote 1, that is 1BCD|1BCD\rangle

All possible cuts of this 4 vertices graph are

We can find the Max-Cut state are 0110|0110\rangle and 1001|1001\rangle and max-cut number is 4.

results matching ""

    No results matching ""