1: \begin{abstract}

2: A numerical method is developed to solve linear semi-infinite programming problem (LSIP) in which

3: the iterates produced by the algorithm are feasible for the original problem.

4: This is achieved by constructing a sequence of standard linear programming problems 

5: with respect to the successive discretization of the index set such that the approximate

6: regions are included in the original feasible region. 

7: The convergence of the approximate solutions to the solution of the original problem is proved 

8: and the associated optimal objective function values of the approximate problems are 

9: monotonically decreasing and converge to the optimal value of LSIP.

10: An adaptive refinement procedure is designed to discretize the index set and update the constraints

11: for the approximate problem. 

12: Numerical experiments demonstrate the performance of the proposed algorithm.

13: \end{abstract}