IDEAS home Printed from https://ideas.repec.org/a/spr/annopr/v295y2020i1d10.1007_s10479-020-03557-0.html
   My bibliography  Save this article

Robust vertex enumeration for convex hulls in high dimensions

Author

Listed:
  • Pranjal Awasthi

    (Rutgers University)

  • Bahman Kalantari

    (Rutgers University)

  • Yikai Zhang

    (Rutgers University)

Abstract

The problem of computing the vertices of the convex hull of a given input set $$S= \{v_i \in \mathbb {R} ^m: i=1, \dots , n\}$$ S = { v i ∈ R m : i = 1 , ⋯ , n } is a classic and fundamental problem, studied in the context of computational geometry, linear and convex programming, machine learning and more. In this article we present All Vertex Triangle Algorithm (AVTA), a robust and efficient algorithm for this problem. On the one hand, without any assumptions, AVTA computes approximation to the subset $$\overline{S}$$ S ¯ of all K vertices of the convex hull of S so that the convex hull of the approximate subset of vertices is as close to conv(S) as desired. On the other hand, assuming a known lower bound $$\gamma $$ γ on the ratio $$\varGamma _*/R$$ Γ ∗ / R , where $$\varGamma _*$$ Γ ∗ the minimum of the distances from each vertex to the convex hull of the remaining vertices and R the diameter of S, AVTA can recover all of $$\overline{S}$$ S ¯ . Furthermore, assuming that instead of S the input is an $$\varepsilon $$ ε -perturbation of S, $$\overline{S}_\varepsilon $$ S ¯ ε , where $$\Vert v_i - v^{\varepsilon }_i \Vert \le \varepsilon R$$ ‖ v i - v i ε ‖ ≤ ε R , AVTA can compute approximation to $$conv(\overline{S}_\varepsilon )$$ c o n v ( S ¯ ε ) , to any prescribed accuracy. Also, given a lower bound to the ratio $$\varSigma _*/R$$ Σ ∗ / R , where $$\varSigma _*$$ Σ ∗ is the minimum of the distances from each vertex to the convex hull of the remaining point of S, AVTA can recover all of $$\overline{S}_\varepsilon $$ S ¯ ε . We show $$\varSigma _* \ge \rho _* \varGamma _*/R$$ Σ ∗ ≥ ρ ∗ Γ ∗ / R , where $$\rho _*$$ ρ ∗ is the minimum distance between distinct pair of points in S and prove the following main results: (1) Given any $$t \in (0,1)$$ t ∈ ( 0 , 1 ) , AVTA computes a subset $$\overline{S}^t$$ S ¯ t of $$\overline{S}$$ S ¯ of cardinality $$K^{(t)}$$ K ( t ) in $$O(n K^{(t)}(m+ t^{-2}))$$ O ( n K ( t ) ( m + t - 2 ) ) operations so that for any $$p \in conv(S)$$ p ∈ c o n v ( S ) its Euclidean distance to $$conv(\overline{S}^t)$$ c o n v ( S ¯ t ) is at most tR. (2) Given $$\gamma \le \gamma _* = \varGamma _*/R$$ γ ≤ γ ∗ = Γ ∗ / R , AVTA computes $$\overline{S}$$ S ¯ in $$O(nK(m+ \gamma ^{-2}))$$ O ( n K ( m + γ - 2 ) ) operations. (3) If K is known, the complexity of AVTA is $$O(nK(m+ \gamma _*^{-2}) \log (\gamma _*^{-1}))$$ O ( n K ( m + γ ∗ - 2 ) log ( γ ∗ - 1 ) ) . Assuming instead of S, its $$\varepsilon $$ ε -perturbation, $$S_\varepsilon $$ S ε is given, we prove (i) Given any $$t \in (0,1)$$ t ∈ ( 0 , 1 ) , AVTA computes a subset $$\overline{S}_\varepsilon ^t \subset \overline{S}_\varepsilon $$ S ¯ ε t ⊂ S ¯ ε of cardinality $$K^{(t)}_\varepsilon $$ K ε ( t ) in $$O(n K^{(t)}_\varepsilon (m+ t^{-2}))$$ O ( n K ε ( t ) ( m + t - 2 ) ) operations so that for any $$p \in conv(S)$$ p ∈ c o n v ( S ) its distance to $$conv(\overline{S}_\varepsilon ^t)$$ c o n v ( S ¯ ε t ) is at most $$(t+\varepsilon ) R$$ ( t + ε ) R . (ii) Given $$\sigma \in [4 \varepsilon , \sigma _*= \varGamma _*/R]$$ σ ∈ [ 4 ε , σ ∗ = Γ ∗ / R ] , AVTA computes $$\overline{S}_\varepsilon $$ S ¯ ε in $$O(nK_\varepsilon (m+ \sigma ^{-2}))$$ O ( n K ε ( m + σ - 2 ) ) operations, where $$K \le K_\varepsilon \le n$$ K ≤ K ε ≤ n . (iii) If $$\gamma \le \gamma _*=\varGamma _*/R$$ γ ≤ γ ∗ = Γ ∗ / R is known satisfying $$4 \varepsilon \le \gamma \rho _*/R$$ 4 ε ≤ γ ρ ∗ / R , AVTA computes $$\overline{S}_\varepsilon $$ S ¯ ε in $$O(nK_\varepsilon (m+ (\gamma \rho _*)^{-2}))$$ O ( n K ε ( m + ( γ ρ ∗ ) - 2 ) ) operations. (iv) Given $$\sigma \in [4 \varepsilon , \sigma _*]$$ σ ∈ [ 4 ε , σ ∗ ] , if K is known, AVTA computes $$\overline{S}_\varepsilon $$ S ¯ ε in $$O(nK(m+ \sigma _*^{-2}) \log (\sigma _*^{-1}))$$ O ( n K ( m + σ ∗ - 2 ) log ( σ ∗ - 1 ) ) operations. We also consider the application of AVTA in the recovery of vertices through the projection of S or $$S_\varepsilon $$ S ε under a Johnson–Lindenstrauss randomized linear projection $$L : \mathbb {R}^{m} \rightarrow \mathbb {R}^{m'}$$ L : R m → R m ′ . Denoting $$U=L(S)$$ U = L ( S ) and $$U_\varepsilon =L(S_\varepsilon )$$ U ε = L ( S ε ) , by relating the robustness parameters of conv(U) and $$conv(U_\varepsilon )$$ c o n v ( U ε ) to those of conv(S) and $$conv(S_\varepsilon )$$ c o n v ( S ε ) , we derive analogous complexity bounds for probabilistic computation of the vertex set of conv(U) or those of $$conv(U_\varepsilon )$$ c o n v ( U ε ) , or an approximation to them. Finally, we apply AVTA to design new practical algorithms for two popular machine learning problems: topic modeling and non-negative matrix factorization. For topic models, our new algorithm leads to significantly better reconstruction of the topic-word matrix than state of the art approaches of Arora et al. (International conference on machine learning, pp 280–288, 2013) and Bansal et al. (Advances in neural information processing systems, pp 1997–2005, 2014). Additionally, we provide a robust analysis of AVTA and empirically demonstrate that it can handle larger amounts of noise than existing methods. For non-negative matrix factorization we show that AVTA is competitive with existing methods that are specialized for this task in Arora et al. (Proceedings of the forty-fourth annual ACM symposium on theory of computing, ACM, pp 145–162, 2012a). We also contrast AVTA with Blum et al. (Proceedings of the twenty-seventh annual ACM-SIAM symposium on discrete algorithms, Society for Industrial and Applied Mathematics, pp 548–557, 2016) Greedy Clustering coreset algorithm for computing approximation to the set of vertices and argue that not only there are regimes where AVTA outperforms that algorithm but it can also be used as a pre-processing step to their algorithm. Thus the two algorithms in fact complement each other.

Suggested Citation

  • Pranjal Awasthi & Bahman Kalantari & Yikai Zhang, 2020. "Robust vertex enumeration for convex hulls in high dimensions," Annals of Operations Research, Springer, vol. 295(1), pages 37-73, December.
  • Handle: RePEc:spr:annopr:v:295:y:2020:i:1:d:10.1007_s10479-020-03557-0
    DOI: 10.1007/s10479-020-03557-0
    as

    Download full text from publisher

    File URL: http://link.springer.com/10.1007/s10479-020-03557-0
    File Function: Abstract
    Download Restriction: Access to the full text of the articles in this series is restricted.

    File URL: https://libkey.io/10.1007/s10479-020-03557-0?utm_source=ideas
    LibKey link: if access is restricted and if your library uses this service, LibKey will redirect you to where you can use your library subscription to access this item
    ---><---

    As the access to this document is restricted, you may want to search for a different version of it.

    References listed on IDEAS

    as
    1. Marguerite Frank & Philip Wolfe, 1956. "An algorithm for quadratic programming," Naval Research Logistics Quarterly, John Wiley & Sons, vol. 3(1‐2), pages 95-110, March.
    2. Bahman Kalantari, 2015. "A characterization theorem and an algorithm for a convex hull problem," Annals of Operations Research, Springer, vol. 226(1), pages 301-349, March.
    Full references (including those not matched with items on IDEAS)

    Citations

    Citations are extracted by the CitEc Project, subscribe to its RSS feed for this item.
    as


    Cited by:

    1. Filippozzi, Rafaela & Gonçalves, Douglas S. & Santos, Luiz-Rafael, 2023. "First-order methods for the convex hull membership problem," European Journal of Operational Research, Elsevier, vol. 306(1), pages 17-33.

    Most related items

    These are the items that most often cite the same works as this one and are cited by the same works as this one.
    1. Guillaume Sagnol & Edouard Pauwels, 2019. "An unexpected connection between Bayes A-optimal designs and the group lasso," Statistical Papers, Springer, vol. 60(2), pages 565-584, April.
    2. Abdelfettah Laouzai & Rachid Ouafi, 2022. "A prediction model for atmospheric pollution reduction from urban traffic," Environment and Planning B, , vol. 49(2), pages 566-584, February.
    3. Chou, Chang-Chi & Chiang, Wen-Chu & Chen, Albert Y., 2022. "Emergency medical response in mass casualty incidents considering the traffic congestions in proximity on-site and hospital delays," Transportation Research Part E: Logistics and Transportation Review, Elsevier, vol. 158(C).
    4. Francesco Rinaldi & Damiano Zeffiro, 2023. "Avoiding bad steps in Frank-Wolfe variants," Computational Optimization and Applications, Springer, vol. 84(1), pages 225-264, January.
    5. Beck, Yasmine & Ljubić, Ivana & Schmidt, Martin, 2023. "A survey on bilevel optimization under uncertainty," European Journal of Operational Research, Elsevier, vol. 311(2), pages 401-426.
    6. Friesz, Terry L. & Tourreilles, Francisco A. & Han, Anthony Fu-Wha, 1979. "Multi-Criteria Optimization Methods in Transport Project Evaluation: The Case of Rural Roads in Developing Countries," Transportation Research Forum Proceedings 1970s 318817, Transportation Research Forum.
    7. Fabiana R. Oliveira & Orizon P. Ferreira & Gilson N. Silva, 2019. "Newton’s method with feasible inexact projections for solving constrained generalized equations," Computational Optimization and Applications, Springer, vol. 72(1), pages 159-177, January.
    8. Ali Fattahi & Sriram Dasu & Reza Ahmadi, 2019. "Mass Customization and “Forecasting Options’ Penetration Rates Problem”," Operations Research, INFORMS, vol. 67(4), pages 1120-1134, July.
    9. Bo Jiang & Tianyi Lin & Shiqian Ma & Shuzhong Zhang, 2019. "Structured nonconvex and nonsmooth optimization: algorithms and iteration complexity analysis," Computational Optimization and Applications, Springer, vol. 72(1), pages 115-157, January.
    10. James Chok & Geoffrey M. Vasil, 2023. "Convex optimization over a probability simplex," Papers 2305.09046, arXiv.org.
    11. A. de Palma & Y. Nesterov, 2001. "Stationary Dynamic Solutions in Congested Transportation Networks: Summary and Perspectives," THEMA Working Papers 2001-19, THEMA (THéorie Economique, Modélisation et Applications), Université de Cergy-Pontoise.
    12. D. J. White, 1993. "A parametric‐based heuristic program for the quadratic assignment problem," Naval Research Logistics (NRL), John Wiley & Sons, vol. 40(4), pages 553-568, June.
    13. Kang, Seong-Cheol & Lee, Hoyoung, 2019. "Economic appraisal of implementing electric vehicle taxis in Seoul," Research in Transportation Economics, Elsevier, vol. 73(C), pages 45-52.
    14. Michael W. Levin & Melissa Duell & S. Travis Waller, 2020. "Arrival Time Reliability in Strategic User Equilibrium," Networks and Spatial Economics, Springer, vol. 20(3), pages 803-831, September.
    15. Yong Xia & Zi Xu, 2010. "An efficient Lagrangian smoothing heuristic for Max-Cut," Indian Journal of Pure and Applied Mathematics, Springer, vol. 41(5), pages 683-700, October.
    16. Arezou Rahimi & Luis A. Vale-Silva & Maria Fälth Savitski & Jovan Tanevski & Julio Saez-Rodriguez, 2024. "DOT: a flexible multi-objective optimization framework for transferring features across single-cell and spatial omics," Nature Communications, Nature, vol. 15(1), pages 1-15, December.
    17. P. B. Assunção & O. P. Ferreira & L. F. Prudente, 2021. "Conditional gradient method for multiobjective optimization," Computational Optimization and Applications, Springer, vol. 78(3), pages 741-768, April.
    18. Ya-Feng Liu & Xin Liu & Shiqian Ma, 2019. "On the Nonergodic Convergence Rate of an Inexact Augmented Lagrangian Framework for Composite Convex Programming," Mathematics of Operations Research, INFORMS, vol. 44(2), pages 632-650, May.
    19. Waddell, Paul & Boeing, Geoff & Gardner, Max & Porter, Emily, 2018. "An Integrated Pipeline Architecture for Modeling Urban Land Use, Travel Demand, and Traffic Assignment," SocArXiv 74zaw, Center for Open Science.
    20. Butyn, Emerson & Karas, Elizabeth W. & de Oliveira, Welington, 2022. "A derivative-free trust-region algorithm with copula-based models for probability maximization problems," European Journal of Operational Research, Elsevier, vol. 298(1), pages 59-75.

    Corrections

    All material on this site has been provided by the respective publishers and authors. You can help correct errors and omissions. When requesting a correction, please mention this item's handle: RePEc:spr:annopr:v:295:y:2020:i:1:d:10.1007_s10479-020-03557-0. See general information about how to correct material in RePEc.

    If you have authored this item and are not yet registered with RePEc, we encourage you to do it here. This allows to link your profile to this item. It also allows you to accept potential citations to this item that we are uncertain about.

    If CitEc recognized a bibliographic reference but did not link an item in RePEc to it, you can help with this form .

    If you know of missing items citing this one, you can help us creating those links by adding the relevant references in the same way as above, for each refering item. If you are a registered author of this item, you may also want to check the "citations" tab in your RePEc Author Service profile, as there may be some citations waiting for confirmation.

    For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: Sonal Shukla or Springer Nature Abstracting and Indexing (email available below). General contact details of provider: http://www.springer.com .

    Please note that corrections may take a couple of weeks to filter through the various RePEc services.

    IDEAS is a RePEc service. RePEc uses bibliographic data supplied by the respective publishers.