A fundamental problem in linear programming, machine learning, and computational geometry is the Convex Hull Membership (CHM): Given a point p and a subset S of n points in Rm, is p ∈ conv(S)? The Triangle Algorithm (TA) computes p′ ∈ conv(S) so that, either ‖p−p‖ ≤ εR, R = max{‖p−v‖ : v ∈ S}; or p′ is a witness, i.e. the orthogonal bisector of pp′ separates p from conv(S). By the Spherical-CHM we mean a CHM, where p = 0, ‖v‖ = 1, ∀v ∈ S. First, we prove the equivalence of exact and approximate versions of CHM and Spherical-CHM. On the one hand, this makes it possible to state a simple O(1/ε) iteration TA, each taking O(n +m) time. On the other hand, using this iteration complexity we prove if for each p′ ∈ conv(S) with ‖p‖ > ε that is not a witness there is v ∈ S with ‖p − v‖ ≥ √ 1 + ε, the iteration complexity of TA reduces to O(1/ε). This matches complexity of Nesterov’s fast-gradient method. The analysis also suggests a strategy for when the property does not hold at an iterate. Lastly, as an application of TA, we show how to solve strict LP feasibility as a dual of CHM. In summary, TA and the Spherical-CHM provide a convenient geometric setting for efficient solution to large-scale CHM and related problems, such as computing all vertices of conv(S).
[1]
L. G. H. Cijan.
A polynomial algorithm in linear programming
,
1979
.
[2]
Peter Richtárik,et al.
Smooth minimization of nonsmooth functions with parallel coordinate descent methods
,
2013,
Modeling and Optimization: Theory and Applications.
[3]
Leonid Khachiyan,et al.
Diagonal Matrix Scaling and Linear Programming
,
1992,
SIAM J. Optim..
[4]
Narendra Karmarkar,et al.
A new polynomial-time algorithm for linear programming
,
1984,
STOC '84.
[5]
Bahman Kalantari,et al.
Robust vertex enumeration for convex hulls in high dimensions
,
2018,
AISTATS.
[6]
Bahman Kalantari,et al.
A procedure of Chvátal for testing feasibility in linear programming and matrix scaling
,
2006
.
[7]
Martin Jaggi,et al.
Coresets for polytope distance
,
2009,
SCG '09.
[8]
E. Gilbert.
An Iterative Procedure for Computing the Minimum of a Quadratic Form on a Convex Set
,
1966
.
[9]
Philip Wolfe,et al.
An algorithm for quadratic programming
,
1956
.
[10]
Bahman Kalantari,et al.
A characterization theorem and an algorithm for a convex hull problem
,
2012,
Ann. Oper. Res..
[11]
Bahman Kalantari,et al.
An Algorithmic Separating Hyperplane Theorem and Its Applications
,
2014,
Discret. Appl. Math..