翻译一段英文算法1. Build the symmetric non-negative n x n matrix M asdescribed in Section 2.2. Let x\3 be the principal eigenvector ofM. Initialize thesolution vector x with the n x 1 zero vector. InitializeL with the set of all candidate ass

翻译一段英文算法1. Build the symmetric non-negative n x n matrix M asdescribed in Section 2.2. Let x\3 be the principal eigenvector ofM. Initialize thesolution vector x with the n x 1 zero vector. InitializeL with the set of all candidate ass
翻译一段英文算法
1. Build the symmetric non-negative n x n matrix M as
described in Section 2.
2. Let x\3 be the principal eigenvector ofM. Initialize the
solution vector x with the n x 1 zero vector. Initialize
L with the set of all candidate assignments.
3. Find a\3 = argmaxa2L(x\3(a)). If x\3(a\3) = 0 stop
and return the solution x. Otherwise set x(a\3) = 1
and remove a\3 from L.
4. Remove from L all potential assignments in conflict
with a\3 = (i, i0). These are assignments of the form
(i, k) and (q, i0) for one-to-one correspondence constraints
(they will be of the form (i, k) for one-to-many
constraints).
5. If L is empty return the solution x. Otherwise go back
to step 3.

翻译一段英文算法1. Build the symmetric non-negative n x n matrix M asdescribed in Section 2.2. Let x\3 be the principal eigenvector ofM. Initialize thesolution vector x with the n x 1 zero vector. InitializeL with the set of all candidate ass
1、根据第二节中的描述构建对称非负n×n阶矩阵M；
2、设x是M的主特征向量,并用nx1阶零向量将x初始化；用待处理作业数据初始化L；
3、查找a= argmaxa2L(x\3(a)).如果x\3(a\3) = 0停止并返回x,否则令x(a)=1并将a从L中移除.
4、将L中所有可能与a=(i,i0)冲突的作业数据从L中移除,作业数据中形式为(i,k) 和(q,i0)的数据与a是一一对应的约束条件.（而（i,k）这样的形式是一对多的约束条件）.
5、如果L为空,则返回结果x,否则转到第三步.