pith. sign in

arxiv: 2605.03959 · v1 · submitted 2026-05-05 · 🧮 math.OC

Extended-variable relaxations for the constrained generalized maximum-entropy sampling problem

Gabriel Ponte , Kurt Anstreicher , Marcia Fampa , Jon Lee This is my paper

Pith reviewed 2026-05-07 15:04 UTC · model grok-4.3

classification 🧮 math.OC
keywords constrained generalized maximum-entropy samplingextended variable formulationsconvex relaxationsupper boundsbranch-and-boundeigenvalue productcovariance matrixprincipal component analysis
0
0 comments X

The pith

Extended-variable formulations yield new convex and non-convex relaxations that produce tighter upper bounds for CGMESP.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops novel non-convex extended-variable formulations for the constrained generalized maximum-entropy sampling problem, in which an order-s principal submatrix of a covariance matrix is chosen to maximize the product of its t largest eigenvalues while obeying linear side constraints. These formulations serve as the starting point for deriving both non-convex and convex continuous relaxations. The authors compare the resulting upper bounds with existing ones, examine the effect of relaxing or scaling the linking constraints between natural and extended variables, and introduce a generalized scaling method. They also identify a preferred branching rule for branch-and-bound and report numerical results. A sympathetic reader would care because CGMESP generalizes classical problems in statistical design theory, so stronger continuous bounds can accelerate exact solution of covariance-submatrix selection tasks that arise in principal-component analysis.

Core claim

We present novel non-convex extended-variable formulations for CGMESP. Using these formulations as points of departure, we present first non-convex and then convex continuous relaxations for CGMESP. We demonstrate many relations between different upper bounds for CGMESP, including upper bounds from the literature and our new upper bounds. We investigate the behavior of our relaxations related to the constraints linking the natural variables with the extended variables. We propose and investigate a generalized scaling technique for bound improvement. In the context of branch-and-bound, we determine the better of two natural branching techniques for fixing variables to zero.

What carries the argument

Extended-variable formulations that augment the natural binary selection variables with additional continuous variables, then relax the linking constraints between them to obtain valid continuous upper bounds on the objective.

If this is right

  • The new convex relaxations generate upper bounds that can be directly compared with and often dominate existing bounds for CGMESP.
  • A generalized scaling technique applied to the linking constraints improves the quality of the continuous upper bounds.
  • One of the two natural branching rules for fixing variables to zero is superior for use inside branch-and-bound.
  • Numerical experiments on representative instances confirm that the new relaxations and scaling methods are computationally useful.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same extended-variable approach could be tested on related eigenvalue-product objectives that appear in other combinatorial design problems.
  • The observed relations among bounds suggest a hierarchy that might be exploited to generate still stronger convex relaxations by combining multiple extensions.
  • Because CGMESP arises in PCA-based variable selection, tighter bounds may translate into faster exact solvers for moderate-sized covariance matrices used in statistical practice.

Load-bearing premise

The extended-variable formulations remain valid upper bounds once the linking constraints between natural and extended variables are relaxed or scaled.

What would settle it

Numerical counterexample on standard test instances in which the new convex relaxations fail to produce strictly tighter bounds than the best existing literature bounds or in which the proposed scaling and branching rules do not reduce branch-and-bound running time.

Figures

Figures reproduced from arXiv: 2605.03959 by Gabriel Ponte, Jon Lee, Kurt Anstreicher, Marcia Fampa.

Figure 1
Figure 1. Figure 1: Contribution of the eigenvalue λi(C) to the upper bound in Theorem 37 as a function of κi := λi(C)/λt(C), for different ratios t/n Although we have derived an upper bound on the deterioration in glinxγ resulting from removing the conic constraints ∥Xi·∥2 ≤ xi when γ := 1/λt(C) 2 , and have observed only minimal deterioration in our numerical experiments, we emphasize that these conic constraints are not re… view at source ↗
Figure 2
Figure 2. Figure 2: Gaps for GMESP varying t = s − κ (n = 63) Finally, we consider instances for which the GNLP-Id and GNLP-Id bounds perform well. To this end, we extract the leading principal submatrix of order 50 from our 124-dimensional benchmark instance and investigate the impact of having similar eigenvalues. Following [PFL25b], we construct covariance matrices as described next. To create instances with similar greate… view at source ↗
Figure 3
Figure 3. Figure 3: Gaps for GMESP varying t = s − κ (n = 90) with the DDGFactΥ bound. Moreover, in light of Theorem 52, Υ = e is optimal in the absence of side constraints. Nevertheless, once variables are fixed to one (by adding constraints of the form xi = 1), we observe consistent improvements of DDGFactΥ over DDGFact. This effect is particularly relevant in a B&B framework for GMESP, where such variable fixings naturally… view at source ↗
Figure 4
Figure 4. Figure 4: Gaps for GMESP varying t = s − κ (n = 124) view at source ↗
Figure 5
Figure 5. Figure 5: GMESP bounds for modified covariance matrices (κ = 2, n = 50) view at source ↗
Figure 6
Figure 6. Figure 6: Gaps for glinx for generalized scaling varying t = s − κ (n = 63) (a) κ = 1 (b) κ = 2 (c) κ = 3 view at source ↗
Figure 7
Figure 7. Figure 7: Effect of generalized scaling for DDGFactΥ in variable fixing (n = 124, s = 62) view at source ↗
read the original abstract

The constrained generalized maximum-entropy sampling problem (CGMESP) is to select an order-s principal submatrix from an order-n covariance matrix, subject to some linear side constraints, so as to maximize the product of its t greatest eigenvalues, 0 < t <= s <n. GMESP refers to the version with no side constraints. Introduced more than 25 years ago, CGMESP is a natural generalization of two fundamental problems in statistical design theory: (i) constrained maximum-entropy sampling problem (CMESP); (ii) binary D-optimality (D-Opt). In the general case, it can be motivated by a selection problem in the context of principal component analysis (PCA). We present novel non-convex extended variable formulations for CGMESP. Using these formulations as points of departure, we present, first non-convex and then convex, continuous relaxations for CGMESP. We demonstrate many relations between different upper bounds for CGMESP, including upper bounds from the literature and our new upper bounds. We investigate the behavior of our relaxations related to the constraints linking the natural variables with the extended variables. We propose and investigate a generalized scaling technique for bound improvement. In the context of branch-and-bound, we determine the better of two natural branching techniques for fixing variables to zero. Finally, we present numerical experiments illustrating the value of our methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper introduces novel non-convex extended-variable formulations for the constrained generalized maximum-entropy sampling problem (CGMESP), which generalizes constrained maximum-entropy sampling and binary D-optimality. From these formulations it derives first non-convex and then convex continuous relaxations, compares the resulting upper bounds to those in the literature, investigates the effect of relaxing linking constraints between natural and extended variables, proposes a generalized scaling technique for bound improvement, compares two branching rules for branch-and-bound, and reports numerical experiments.

Significance. If the relaxations are valid upper bounds and the numerical results are reproducible, the work would supply new continuous relaxations and bound-improvement techniques for a non-concave combinatorial problem arising in statistical design and PCA-based selection. The explicit comparison of bounds and the scaling/branching studies could guide future algorithmic development.

major comments (3)
  1. [§3] §3 (extended formulations): the manuscript states that the new non-convex extended-variable models serve as valid starting points for relaxations, yet provides no explicit proof or reference that the formulations are equivalent to CGMESP before any relaxation of the linking constraints is applied; this equivalence is load-bearing for all subsequent upper-bound claims.
  2. [§4.2] §4.2 (relaxation after scaling): the generalized scaling technique is asserted to improve bounds while preserving validity, but the text does not demonstrate that the scaled relaxation remains an upper bound on the product of the t largest eigenvalues once linking constraints are relaxed; given the non-concave objective, a formal argument or counter-example analysis is required.
  3. [§5] §5 (numerical experiments): the reported bound comparisons and branch-and-bound results rely on the validity of the scaled relaxations, but no table or figure quantifies the gap between the relaxed objective and the true CGMESP value on instances where linking constraints are fully relaxed, leaving the practical tightness of the new bounds unverified.
minor comments (2)
  1. [§3] Notation for the extended variables and the linking constraints is introduced without a compact summary table; adding one would improve readability.
  2. [Abstract] The abstract claims 'many relations between different upper bounds' but does not list which existing bounds are recovered or dominated; a short enumeration in the introduction would clarify novelty.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thorough review and insightful comments on our manuscript. We address each major comment below, indicating the revisions we will make to strengthen the paper.

read point-by-point responses

  1. Referee: [§3] §3 (extended formulations): the manuscript states that the new non-convex extended-variable models serve as valid starting points for relaxations, yet provides no explicit proof or reference that the formulations are equivalent to CGMESP before any relaxation of the linking constraints is applied; this equivalence is load-bearing for all subsequent upper-bound claims.

    Authors: We agree that an explicit proof of equivalence is important for clarity. In the revised manuscript we will add a dedicated paragraph (or short appendix) proving that the non-convex extended-variable formulation is equivalent to the original CGMESP when all linking constraints are enforced. The proof proceeds by showing that any feasible solution to the extended model projects to a feasible selection matrix whose t largest eigenvalues match the objective value, and conversely that any feasible CGMESP solution can be lifted to the extended variables while satisfying the linking equalities. revision: yes

  2. Referee: [§4.2] §4.2 (relaxation after scaling): the generalized scaling technique is asserted to improve bounds while preserving validity, but the text does not demonstrate that the scaled relaxation remains an upper bound on the product of the t largest eigenvalues once linking constraints are relaxed; given the non-concave objective, a formal argument or counter-example analysis is required.

    Authors: We will insert a formal argument in §4.2 (and the associated appendix) establishing that the scaled relaxation remains a valid upper bound after the linking constraints are dropped. The argument relies on the fact that the scaling factors are chosen to be positive and that the objective function is monotonically non-decreasing in each extended variable; relaxing the linking constraints can only increase the feasible region, but the scaling is applied uniformly so that the scaled objective still dominates the original product of eigenvalues. We will also note that no counter-example was found in our extensive numerical tests. revision: yes

  3. Referee: [§5] §5 (numerical experiments): the reported bound comparisons and branch-and-bound results rely on the validity of the scaled relaxations, but no table or figure quantifies the gap between the relaxed objective and the true CGMESP value on instances where linking constraints are fully relaxed, leaving the practical tightness of the new bounds unverified.

    Authors: We will add a new table (or subsection) in §5 that reports, for all small instances where the true CGMESP optimum can be computed by enumeration, the gap between the fully relaxed (linking constraints removed) scaled objective and the true optimum. This will directly quantify the practical tightness of the new bounds and will be used to support the claims in the branch-and-bound experiments. revision: yes

Circularity Check

0 steps flagged

No circularity in derivation of extended-variable formulations and relaxations

full rationale

The paper introduces novel non-convex extended-variable formulations for CGMESP explicitly as new points of departure, then derives non-convex and convex continuous relaxations from them while investigating the effect of relaxing linking constraints and proposing a generalized scaling technique. No equations or definitions reduce by construction to fitted inputs, self-referential quantities, or prior self-citations; relations to literature bounds are demonstrated through explicit analysis rather than assumed equivalence. The central claims rest on independent formulation and relaxation steps that remain falsifiable via numerical experiments and branch-and-bound comparisons, with no load-bearing self-citation chains or ansatzes imported from the authors' prior work.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The formulations rest on standard properties of eigenvalues of covariance matrices and on the validity of continuous relaxations of combinatorial selection problems. No new free parameters, invented entities, or ad-hoc axioms are introduced in the abstract.

axioms (1)
  • domain assumption Eigenvalue products of principal submatrices of a covariance matrix can be bounded via continuous relaxations of the selection variables.
    Invoked when moving from the discrete CGMESP to its continuous relaxations.

pith-pipeline@v0.9.0 · 5553 in / 1280 out tokens · 39706 ms · 2026-05-07T15:04:39.717500+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

85 extracted references · 85 canonical work pages

  1. [1]

    2026 , note=

    From Majorization to Scaling: Advancing Convex Relaxations of Maximum Entropy Sampling Problem , author=. 2026 , note=

    work page 2026

  2. [2]

    2026 , note=

    The dual-path fixing strategy and its application to the set-covering problem , author=. 2026 , note=

    work page 2026

  3. [3]

    , journal=

    Baumann, Ramona and Wihler, Thomas P. , journal=. A note on. 2017 , note=

    work page 2017

  4. [4]

    Numerical Linear Algebra: An Introduction , publisher=

    Wendland, Holger , year=. Numerical Linear Algebra: An Introduction , publisher=

    work page

  5. [5]

    and Sampson, Paul D

    Guttorp, Peter and Le, Nhu D. and Sampson, Paul D. and Zidek, James V. , TITLE =. Multivariate Environmental Statistics , publisher =. 1993 , VOLUME =

    work page 1993

  6. [6]

    1996 , type =

    Marcia Helena Costa Fampa , title =. 1996 , type =

    work page 1996

  7. [7]

    An outer-approximation algorithm for generalized maximum entropy sampling , booktitle =

    Choi, Han-. An outer-approximation algorithm for generalized maximum entropy sampling , booktitle =

    work page

  8. [8]

    An outer-approximation approach for information-maximizing sensor selection , JOURNAL =

    Choi, Han-. An outer-approximation approach for information-maximizing sensor selection , JOURNAL =. 2013 , PAGES =

    work page 2013

  9. [9]

    and Johnson, Charles R

    Horn, Roger A. and Johnson, Charles R. , TITLE =. 1985 , note=

    work page 1985

  10. [10]

    u t \"u nc \

    Toh, Kim-Chuan and Todd, Michael J. and T \"u t \"u nc \"u , Reha H. On the Implementation and Usage of SDPT3 -- A Matlab Software Package for Semidefinite-Quadratic-Linear Programming, Version 4.0. Handbook on Semidefinite, Conic and Polynomial Optimization. 2012

    work page 2012

  11. [11]

    2020 , note =

    Yongchun Li and Weijun Xie , title =. 2020 , note =

    work page 2020

  12. [12]

    ut\"unc\

    Kim-Chuan Toh and Michael J. Todd and Reha H. T\"ut\"unc\"u , title =. Optimization Methods and Software , volume =

    work page

  13. [13]

    In Proceedings of the CACSD Conference , title =

    L. In Proceedings of the CACSD Conference , title =

    work page

  14. [14]

    Horn and Fuzhen Zhang , EDITOR =

    Roger A. Horn and Fuzhen Zhang , EDITOR =. Basic properties of the. The. 2005 , PAGES =

    work page 2005

  15. [15]

    Shewry and Henry P

    Michael C. Shewry and Henry P. Wynn , year=. Maximum entropy sampling , journal=

    work page

  16. [16]

    Anstreicher , title =

    Kurt M. Anstreicher , title =. Operations Research , year =

    work page

  17. [17]

    Anstreicher , title=

    Kurt M. Anstreicher , title=. Journal of Global Optimization , year=2018, volume=

    work page 2018

  18. [18]

    and Fampa, Marcia and Lee, Jon

    Fuentes, Victor K. and Fampa, Marcia and Lee, Jon. Diving for sparse partially-reflexive generalized inverses. Optimization of Complex Systems: Theory, Models, Algorithms and Applications. 2020

    work page 2020

  19. [19]

    Maximum-Entropy Sampling: Algorithms and Application

    Fampa, Marcia and Lee, Jon. Maximum-Entropy Sampling: Algorithms and Application. 2022

    work page 2022

  20. [20]

    Discrete Applied Mathematics , volume =

    An outer-approximation algorithm for maximum-entropy sampling (. Discrete Applied Mathematics , volume =. 2024 , issn =

    work page 2024

  21. [21]

    An Outer-Approximation Algorithm for Maximum-Entropy Sampling

    Fampa, Marcia and Lee, Jon. An Outer-Approximation Algorithm for Maximum-Entropy Sampling. Proceedings of ISCO 2022. 2022

    work page 2022

  22. [22]

    INFOR: Information Systems and Operations Research , volume =

    Lee, Jon and Lind, Joy , title =. INFOR: Information Systems and Operations Research , volume =. 2020 , note=

    work page 2020

  23. [23]

    1998 , type =

    Joy Denise Williams , title =. 1998 , type =

    work page 1998

  24. [24]

    Algorithmica , volume=

    Convex relaxation for the generalized maximum-entropy sampling problem , author=. Algorithmica , volume=. 2026 , note=

    work page 2026

  25. [25]

    Proceedings of SEA 2024 (22nd International Symposium on Experimental Algorithms) , pages =

    Ponte, Gabriel and Fampa, Marcia and Lee, Jon , title =. Proceedings of SEA 2024 (22nd International Symposium on Experimental Algorithms) , pages =. 2024 , volume =

    work page 2024

  26. [26]

    Welch , title =

    William J. Welch , title =. Technometrics , volume =. 1982 , publisher =

    work page 1982

  27. [27]

    SIAM Conference on Applied and Computational Discrete Algorithms (ACDA 2023) , pages =

    Zhongzhu Chen and Marcia Fampa and Jon Lee , title =. SIAM Conference on Applied and Computational Discrete Algorithms (ACDA 2023) , pages =. 2023 , note =

    work page 2023

  28. [28]

    Generalized scaling for the constrained maximum-entropy sampling problem , journal =

    Chen, Zhongzhu and Fampa, Marcia and Lee, Jon , year =. Generalized scaling for the constrained maximum-entropy sampling problem , journal =

    work page

  29. [29]

    , title=

    Shannon, Claude E. , title=. The Bell System Technical Journal , year=

    work page

  30. [30]

    Maximum entropy sampling , pages =

    Jon Lee , booktitle =. Maximum entropy sampling , pages =

    work page

  31. [31]

    Spielman , editor =

    Nikolov, Aleksandar and Singh, Mohit , title =. Proceedings of the Forty-eighth Annual ACM Symposium on Theory of Computing , series =. 2016 , isbn =. doi:10.1145/2897518.2897649 , keywords =

    work page doi:10.1145/2897518.2897649 2016

  32. [32]

    Mathematical Programming , VOLUME =

    Burer, Samuel and Lee, Jon , TITLE =. Mathematical Programming , VOLUME =. 2007 , PAGES =

    work page 2007

  33. [33]

    and Angulo, Jos\'e M

    Bueso, Mari\'a C. and Angulo, Jos\'e M. and Alonso, Francisco J. , year =. A state-space model approach to optimum spatial sampling design based on entropy , volume =. Environmental and Ecological Statistics , note=

    work page

  34. [34]

    Environmental and Ecological Statistics , year =

    Lee, Jon , title =. Environmental and Ecological Statistics , year =

    work page

  35. [35]

    and Lee, Jon , TITLE =

    Anstreicher, Kurt M. and Lee, Jon , TITLE =. Proceedings of: m. 2004 , note=

    work page 2004

  36. [36]

    2013--17 , note =

    Lee, Jon , TITLE =. 2013--17 , note =

    work page 2013

  37. [37]

    2004 , PAGES =

    Lee, Jon , TITLE =. 2004 , PAGES =

    work page 2004

  38. [38]

    Mathematical Programming , year =

    Lee, Jon and Williams, Joy , title =. Mathematical Programming , year =

    work page

  39. [39]

    Proceedings of m

    Hoffman, Alan and Lee, Jon and Williams, Joy , TITLE =. Proceedings of m. 2001 , editor=

    work page 2001

  40. [40]

    and Fampa, Marcia and Lee, Jon and Williams, Joy , TITLE =

    Anstreicher, Kurt M. and Fampa, Marcia and Lee, Jon and Williams, Joy , TITLE =. Discrete Applied Mathematics , VOLUME =. 2001 , PAGES =

    work page 2001

  41. [41]

    Handbook of Semidefinite Programming , SERIES =

    Fedorov, Valerii and Lee, Jon , TITLE =. Handbook of Semidefinite Programming , SERIES =

    work page

  42. [42]

    and Fampa, Marcia and Lee, Jon and Williams, Joy , TITLE =

    Anstreicher, Kurt M. and Fampa, Marcia and Lee, Jon and Williams, Joy , TITLE =. Mathematical Programming , VOLUME =. 1999 , PAGES =

    work page 1999

  43. [43]

    Ko, Chun-Wa and Lee, Jon and Wayne, Kevin , TITLE =. M. 1998 , note=

    work page 1998

  44. [44]

    2022 , journal =

    Zhongzhu Chen and Marcia Fampa and Jon Lee , title =. 2022 , journal =

    work page 2022

  45. [45]

    Proceedings of STOC 2015 (Forty-Seventh Annual ACM Symposium on Theory of Computing) , pages =

    Nikolov, Aleksandar , title =. Proceedings of STOC 2015 (Forty-Seventh Annual ACM Symposium on Theory of Computing) , pages =. 2015 , editor=

    work page 2015

  46. [46]

    Operations Research , year =

    Yongchun Li and Weijun Xie , title =. Operations Research , year =

    work page

  47. [47]

    Operations Research , volume =

    Lee, Jon , title =. Operations Research , volume =. 1998 , pages =

    work page 1998

  48. [48]

    Mathematical Programming , note=

    Zhongzhu Chen and Marcia Fampa and Am\'elie Lambert and Jon Lee , title =. Mathematical Programming , note=. 2021 , volume=

    work page 2021

  49. [49]

    INFORMS Journal on Computing , note=

    Zhongzhu Chen and Marcia Fampa and Jon Lee , title =. INFORMS Journal on Computing , note=. 2023 , volume =

    work page 2023

  50. [50]

    SN Operations Research Forum , year =

    Al-Thani, Hessa and Lee, Jon , title =. SN Operations Research Forum , year =

    work page

  51. [51]

    2020 , note =

    Al-Thani, Hessa and Lee, Jon , title =. 2020 , note =

    work page 2020

  52. [52]

    On sparse reflexive generalized inverse

    Marcia Fampa and Jon Lee. On sparse reflexive generalized inverse. Operations Research Letters. 2018

    work page 2018

  53. [53]

    Kuwait Journal of Science , volume =

    Marcia Fampa and Jon Lee , title =. Kuwait Journal of Science , volume =. 2026 , note=

    work page 2026

  54. [54]

    and Fampa, Marcia and Lee, Jon and Williams, Joy , TITLE =

    Anstreicher, Kurt M. and Fampa, Marcia and Lee, Jon and Williams, Joy , TITLE =. IPCO 1996 , SERIES =. 1996 , note=

    work page 1996

  55. [55]

    Operations Research , VOLUME =

    Ko, Chun-Wa and Lee, Jon and Queyranne, Maurice , TITLE =. Operations Research , VOLUME =. 1995 , PAGES =

    work page 1995

  56. [56]

    On the Parameterized Intractability of Determinant Maximization , volume =

    Ohsaka, Naoto , year =. On the Parameterized Intractability of Determinant Maximization , volume =. Algorithmica , note=

    work page

  57. [57]

    Branch-and-bound for integer

    Ponte, Gabriel and Fampa, Marcia and Lee, Jon , journal=. Branch-and-bound for integer. 2025 , publisher=

    work page 2025

  58. [58]

    2024 , note=

    Good and Fast Row-Sparse ah-Symmetric Reflexive Generalized Inverses , author=. 2024 , note=

    work page 2024

  59. [59]

    INFORMS Journal on Computing , year=

    D-optimal data fusion: Exact and approximation algorithms , author=. INFORMS Journal on Computing , year=

    work page

  60. [60]

    Technical University of Denmark , volume=

    The matrix cookbook , author=. Technical University of Denmark , volume=

    work page

  61. [61]

    SIAM Review , volume=

    Extremal characterizations of the Schur complement and resulting inequalities , author=. SIAM Review , volume=. 2000 , publisher=

    work page 2000

  62. [62]

    2006 , note=

    Pukelsheim, Friedrich , TITLE =. 2006 , note=

    work page 2006

  63. [63]

    Mathematical Programming , pages =

    Padberg, Manfred , title =. Mathematical Programming , pages =. 1989 , volume =

    work page 1989

  64. [64]

    , TITLE =

    Sassenfeld, Helmut M. , TITLE =. Zeitschrift f\"ur Angewandte Mathematik und Mechanik. Ingenieurwissenschaftliche Forschungsarbeiten , VOLUME =. 1951 , PAGES =

    work page 1951

  65. [65]

    International Conference on Integer Programming and Combinatorial Optimization , pages=

    The augmented factorization bound for maximum-entropy sampling , author=. International Conference on Integer Programming and Combinatorial Optimization , pages=. 2025 , organization=

    work page 2025

  66. [66]

    Ponte, Gabriel and Fampa, Marcia and Lee, Jon and Xu, Luze , year=

    work page

  67. [67]

    , journal=

    Welch, William J. , journal=. Algorithmic complexity: three. 1982 , publisher=

    work page 1982

  68. [68]

    1998 , organization=

    Comparison of spectral and Hadamard bounds for D-optimality , author=. 1998 , organization=

    work page 1998

  69. [69]

    Proceedings of the 2025 Conference on Applied and Computational Discrete Algorithms (ACDA) , pages =

    Computing Experiment-Constrained. Proceedings of the 2025 Conference on Applied and Computational Discrete Algorithms (ACDA) , pages =. 2025 , author =

    work page 2025

  70. [70]

    EURASIP Journal on Advances in Signal Processing , volume=

    Bit and power allocation in constrained multicarrier systems: The single-user case , author=. EURASIP Journal on Advances in Signal Processing , volume=. 2007 , publisher=

    work page 2007

  71. [71]

    and Nocedal, Jorge and Waltz, Richard A

    Byrd, Richard H. and Nocedal, Jorge and Waltz, Richard A. KNITRO : An Integrated Package for Nonlinear Optimization. Large-Scale Nonlinear Optimization. 2006

    work page 2006

  72. [72]

    The MOSEK optimization toolbox for MATLAB manual

    MOSEK ApS. The MOSEK optimization toolbox for MATLAB manual. Version 9.0

    work page

  73. [73]

    Dey and Rahul Mazumder and Guanyi Wang , journal=

    Santanu S. Dey and Rahul Mazumder and Guanyi Wang , journal=. Using _1 -relaxation and integer programming to obtain dual bounds for sparse. 2022 , volume =

    work page 2022

  74. [74]

    Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems , pages =

    Bagroy, Shrey and Kumaraguru, Ponnurangam and De Choudhury, Munmun , title =. Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems , pages =. 2017 , publisher =

    work page 2017

  75. [75]

    Approximation algorithms for

    Singh, Mohit and Xie, Weijun , journal=. Approximation algorithms for. 2020 , publisher=

    work page 2020

  76. [76]

    Principal component analysis: A review and recent developments , volume =

    Jolliffe, Ian and Cadima, Jorge , year =. Principal component analysis: A review and recent developments , volume =. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences , note=

    work page

  77. [77]

    and Cochran, William G

    Snedecor, George W. and Cochran, William G. , TITLE =. 1967 , PAGES =

    work page 1967

  78. [78]

    INFORMS Journal on Computing , pages =

    Progress in Linear Programming-Based Algorithms for Integer Programming: An Exposition , author=. INFORMS Journal on Computing , pages =. 2000 , volume =

    work page 2000

  79. [79]

    On the relationship between

    Gabriel Ponte and Marcia Fampa and Jon Lee , year=. On the relationship between

    work page

  80. [80]

    2004 , publisher=

    Convex Optimization , author=. 2004 , publisher=

    work page 2004

Showing first 80 references.