1: \begin{abstract}

2: %In the class of large-scale unconstrained optimization methods, the limited memory BFGS (L-

3: %BFGS) method has gained more significance over the last few decades. Recently, %Sugimoto and

4: %Yamashita proposed the regularized limited memory BFGS (Regularized L-BFGS) with non

5: %-monotone technique and reported that it is competitive to the standard L-BFGS method. L-

6: %BFGS uses line search while RL-BFGS does not uses line search and hence it fails to take

7: %longer step. In order to take longer step, we propose to use Wolfe condition as line search

8: %when the iteration of RL-BFGS is successful and search direction is sufficient descent 

9: %direction. We show that the proposed method has global convergence under usual assumptions.

10: %Finally, we compare numerical results of these algorithms on the set of CUTEst problems,

11: %which shows that RL-BFGS with proposed technique is robust and stable than the L-BFGS and

12: %RL-BFGS method.

13: %   \end{abstract}

14: