Curvature and complexity: Better lower bounds for geodesically convex optimization

被引:0
作者
Criscitiello, Christopher [1 ]
Boumal, Nicolas [1 ]
机构
[1] Ecole Polytech Fed Lausanne EPFL, Inst Math, Lausanne, Switzerland
来源
THIRTY SIXTH ANNUAL CONFERENCE ON LEARNING THEORY, VOL 195 | 2023年 / 195卷
关键词
geodesic convexity; Riemannian optimization; curvature; lower bounds; hyperbolic; WORST-CASE PERFORMANCE;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We study the query complexity of geodesically convex (g-convex) optimization on a manifold. To isolate the effect of that manifold's curvature, we primarily focus on hyperbolic spaces. In a variety of settings (smooth or not; strongly g-convex or not; high- or low-dimensional), known upper bounds worsen with curvature. It is natural to ask whether this is warranted, or an artifact. For many such settings, we propose a first set of lower bounds which indeed confirm that (negative) curvature is detrimental to complexity. To do so, we build on recent lower bounds (Hamilton and Moitra, 2021; Criscitiello and Boumal, 2022a) for the particular case of smooth, strongly g-convex optimization. Using a number of techniques, we also secure lower bounds which capture dependence on condition number and optimality gap, which was not previously the case. We suspect these bounds are not optimal. We conjecture optimal ones, and support them with a matching lower bound for a class of algorithms which includes subgradient descent, and a lower bound for a related game. Lastly, to pinpoint the difficulty of proving lower bounds, we study how negative curvature influences (and sometimes obstructs) interpolation with g-convex functions.
引用
收藏
页数:45
相关论文
共 50 条
  • [41] Information complexity of mixed-integer convex optimization
    Basu, Amitabh
    Jiang, Hongyi
    Kerger, Phillip
    Molinaro, Marco
    [J]. MATHEMATICAL PROGRAMMING, 2025, 210 (1-2) : 3 - 45
  • [42] Quantum SDP-Solvers: Better upper and lower bounds
    van Apeldoorn, Joran
    Gilyen, Andras
    Gribling, Sander
    de Wolf, Ronald
    [J]. 2017 IEEE 58TH ANNUAL SYMPOSIUM ON FOUNDATIONS OF COMPUTER SCIENCE (FOCS), 2017, : 403 - 414
  • [43] Better and Simpler Lower Bounds for Differentially Private Statistical Estimation
    Narayanan, Shyam
    [J]. IEEE TRANSACTIONS ON INFORMATION THEORY, 2025, 71 (02) : 1376 - 1388
  • [44] Lower Bounds on the Complexity of MSO1 Model-Checking
    Ganian, Robert
    Hlineny, Petr
    Langer, Alexander
    Obdrzalek, Jan
    Rossmanith, Peter
    Sikdar, Somnath
    [J]. 29TH INTERNATIONAL SYMPOSIUM ON THEORETICAL ASPECTS OF COMPUTER SCIENCE, (STACS 2012), 2012, 14 : 326 - 337
  • [45] Lower bounds for randomized and quantum query complexity using Kolmogorov arguments
    Laplante, Sophie
    Magniez, Frederic
    [J]. SIAM JOURNAL ON COMPUTING, 2008, 38 (01) : 46 - 62
  • [46] Functional Lower Bounds for Arithmetic Circuits and Connections to Boolean Circuit Complexity
    Forbes, Michael A.
    Kumar, Mrinal
    Saptharishi, Ramprasad
    [J]. 31ST CONFERENCE ON COMPUTATIONAL COMPLEXITY (CCC 2016), 2016, 50
  • [47] Depth-4 Lower Bounds, Determinantal Complexity: A Unified Approach
    Suryajith Chillara
    Partha Mukhopadhyay
    [J]. computational complexity, 2019, 28 : 545 - 572
  • [48] Distribution Testing Lower Bounds via Reductions from Communication Complexity
    Blais, Eric
    Canonne, Clement L.
    Gur, Tom
    [J]. 32ND COMPUTATIONAL COMPLEXITY CONFERENCE (CCC 2017), 2017, 79
  • [49] Depth-4 Lower Bounds, Determinantal Complexity: A Unified Approach
    Chillara, Suryajith
    Mukhopadhyay, Partha
    [J]. COMPUTATIONAL COMPLEXITY, 2019, 28 (04) : 545 - 572
  • [50] Distribution Testing Lower Bounds via Reductions from Communication Complexity
    Blais, Eric
    Canonne, Clement L.
    Gur, Tom
    [J]. ACM TRANSACTIONS ON COMPUTATION THEORY, 2019, 11 (02)