1: \begin{abstract}

2: The original simplicial method ({\sc OSM}), a variant of the classic Kelley's cutting plane method,

3: has been shown to converge to the minimizer of a composite convex and submodular objective,

4: though no rate of convergence for this method was known. Moreover, {\sc OSM} is required to solve subproblems in each iteration

5: whose size grows linearly in the number of iterations. We propose a limited memory version of Kelley's method ({\sc L-KM}) and of {\sc OSM} that requires

6: limited memory (at most $n+1$ constraints for an $n$-dimensional problem) independent of the iteration. We prove convergence for {\sc L-KM} when the convex part of the objective ($g$) is strongly convex

7: and show it converges linearly when $g$ is also smooth. Our analysis relies on duality between minimization of the composite objective and minimization of a convex function over the corresponding submodular base polytope. We introduce a limited memory version, {\sc L-FCFW},

8: of the Fully-Corrective Frank-Wolfe ({\sc FCFW}) method with approximate correction,

9: to solve the dual problem. We show that {\sc L-FCFW} and {\sc L-KM} are dual algorithms

10: that produce the same sequence of iterates;

11: hence both converge linearly (when $g$ is smooth and strongly convex) and with limited memory. We propose {\sc L-KM} to minimize composite convex and submodular objectives;

12: however, our results on {\sc L-FCFW} hold for general polytopes and may be of independent interest.

13: \end{abstract}

14: