Pith. sign in

REVIEW 3 major objections 4 minor 45 references

Improving the Robustness of the Projected Gradient Descent Method for Nonlinear Constrained Optimization Problems in Topology Optimization

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A projected gradient descent method with Schur-complement bound handling and a constraint-aware update split matches or beats MMA on 3D heat-sink topology problems with minimal tuning, and its active-set projection provably reaches a KKT…

desk verdict A useful engineering contribution to PGD for topology optimization, but the formal convergence claim is narrower than the abstract suggests. read the letter →

arxiv 2412.07634 v1 pith:PLFHY4ZQ submitted 2024-12-10 math.OC math-phmath.MP

Lucka Barbeau , Marc-Étienne Lamarche-Gagnon , Florin Ilinca This is my paper · ORCID
classification math.OCmath-phmath.MP MSC 90C3090C0665K05
keywords inertialprojectedgradientdescentinfeasibleactivesetbulkconstraintsmanipulationSchurcomplementdecompositionstepnonlineartopologyoptimizationheatsinkdesign
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper is trying to establish that projected gradient descent, a simple first-order method that scales well to large design spaces, can be made robust enough for topology optimization problems with nonlinear constraints, a regime where the method is often considered fragile. Two mechanisms are added: bound constraints are folded directly into the projection step through a Schur complement inside an active-set solver that adds and removes constraints in bulk, and the update step is split into a component lying in the constraint directions and a component orthogonal to them, so the orthogonal part can be rescaled using an approximation of the Lagrangian and damped when constraints are broken. Tested on a three-dimensional heat-sink design with a volume constraint and a nonlinear overhang constraint, the enhanced PGD reaches final cost values comparable to or slightly better than the Method of Moving Asymptotes while requiring almost no parameter tuning. The interest, if the claim holds, is that a robust first-order optimizer could replace tuning-heavy workhorse methods like MMA on very large problems where second-order methods become impractical.

What carries the argument

The machinery has three pieces. The first is a Schur complement decomposition of the projection system, separating a small block of active global constraints from a diagonal block of univariate constraints, so the projection requires solving only an $m^* \times m^*$ linear system rather than a system involving every bound. The second is an infeasible active-set algorithm that adds all violated constraints in bulk, removes constraints with negative Lagrange multipliers in bulk, and reverts to one-at-a-time constraint manipulation when the cost-monotonicity check fails; Theorem 1 proves it converges to a stable active set satisfying the KKT conditions of the projection subproblem. The third is a post-projection split of the update vector, computed by a Gram-Schmidt process, into a component $\Delta\boldsymbol{\phi}^\parallel$ within the span of the constraint gradients and a component $\Delta\boldsymbol{\phi}^\perp$ in the null space of those gradients; only the null-space component is scaled by $\gamma_n/\alpha_n$ (with $\gamma_n$ built from the variation of an approximate Lagrangian gradient) and damped by $\mu^h$ when constraints have been broken for $h$ consecutive steps, which keeps the cost function moving while actively correcting constraint violations.

What would settle it

Solve a small nonconvex test problem in which the local Lipschitz estimate of Equation (55) is systematically too small, for instance a quadratic cost whose curvature jumps sharply when a continuation parameter changes, and check whether the implemented algorithm, which uses $\gamma_n \neq \alpha_n$ and the infeasible branch of Equation (42), ever produces an iteration with increased cost or a constraint violation that persists instead of decaying; the paper reports that no such backtracking was needed in its heat-sink tests, so a single instance that requires it would show the robustness claim is case-dependent.

Watch

Extended reading notes

Core claim

In the authors' terms, the central claim is that two modifications to the inertial projected gradient descent method make it reliable for nonlinear constrained topology optimization: first, univariate constraints such as design-variable bounds are incorporated directly into the projection step using a Schur complement that separates the diagonal block of univariate constraints from the block of global constraints, inside an active-set algorithm that manipulates constraints in bulk and only falls back to single-constraint handling when a monotonicity check fails; second, the projected update is decomposed, relative to the constraint gradients, into a component in the constraint variation space and a component in its orthogonal complement, and only the orthogonal component is scaled by a factor derived from the variation of an approximate Lagrangian, with an additional relaxation factor applied when constraints have been broken for several consecutive steps. Theorem 1 states that the active-set iteration converges to a stable active set satisfying the KKT conditions of the linearized projection subproblem. Empirically, on a 3D heat-conduction topology optimization problem, with and without the overhang constraint, the default-parameter PGD converges to final cost values comparable to or lower than those of the Method of Moving Asymptotes, converges faster after the nonlinear constraint is broken, and shows only mild sensitivity to its tunable parameters.

Load-bearing premise

The algorithm's convergence is proven only for the case where its two step-size parameters are equal and every iterate stays inside the feasible region; the version actually run uses unequal parameters and deliberately steps outside feasibility, so the reliability of the results rests on empirical behavior rather than on the proof.

Editorial extensions

If this is right

  • With default settings ($\hat{\beta}=0.2$, $\mu=0.95$) the proposed PGD reaches final cost values comparable to or lower than MMA on both tested heat-sink cases, while MMA required its asymptote move limit to be tuned for the problem.
  • The overhang constraint, which is strongly nonlinear, is recovered faster after being broken when the Lagrangian-based $\gamma$ correction is active, shortening the intervals during which the constraint is violated inside continuation loops.
  • Because the Schur complement reduces the projection to a small $m^* \times m^*$ linear solve, the cost per iteration stays near-constant as the number of design variables grows, which is what makes the method practical for large design spaces.
  • Theorem 1 guarantees that every projection subproblem terminates at a stable active set satisfying the KKT conditions of the linearized problem, so the inner iteration is well-defined before the outer step is taken.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The infeasible branch of the update acts like an implicit penalty-recovery mechanism, and a natural extension is to test whether the same robustness carries over to other nonlinear constraint families, such as stress or buckling constraints in structural design.
  • Replacing the fixed relaxation factor $\mu$ with one that scales with the actual magnitude of constraint violation, which the authors list as future work, would likely remove the remaining sensitivity to constraint jumps when the continuation parameters change between optimization loops.
  • The Schur-complement active-set projection is a general device for quadratic-programming projections with many separable bound constraints and few global constraints, so other first-order methods could borrow it independently of the PGD update scheme.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes two enhancements to an inertial projected gradient descent (PGD) method for large-scale constrained optimization, targeting topology optimization. First, univariate bound constraints are incorporated directly into the projection step through a Schur complement within an active-set method that manipulates constraints in bulk. Second, the update step is decomposed into components parallel and orthogonal to the constraint gradients, with the orthogonal component scaled by a factor derived from an approximation of the Lagrangian and by a relaxation factor activated when constraints are broken. The authors prove convergence of the active-set projection subproblem (Theorem 1), give sufficient descent conditions for the modified outer iteration, and report numerical results on a 3D heat-sink topology optimization problem with a volume constraint and a nonlinear overhang constraint, comparing against MMA and GCMMA. The paper includes a Python reference implementation in the supplementary material and a parameter-sensitivity study.

Significance. If the claims are substantiated, the work is a useful contribution to first-order methods for topology optimization: it offers a practical way to handle many univariate constraints inside the projection and improves robustness for nonlinear constraints while retaining the scalability of PGD. The paper's strengths include detailed algorithmic presentation, a supplementary Python implementation, a careful ablation study (traditional, intermediary, and proposed variants), and an explicit discussion of parameter sensitivity and limitations. The main weakness is that the formal convergence analysis does not cover the algorithm as actually implemented: the descent conditions require feasible iterates and α_n = γ_n, while the implemented update admits an infeasible branch and the paper states that γ_n ≠ α_n in most cases. The numerical evidence is also limited to one proprietary-solver test case, so the 'comparable to or exceeding MMA' claim is suggestive rather than definitive.

major comments (3)
  1. [Section 2.3, Eqs. (42)–(54)] The convergence analysis does not cover the implemented algorithm. The sufficient descent conditions in Eqs. (49)–(51) are derived under the assumptions that the iterates remain feasible and that min(1, γ_n/α_n) is applied to both components of the update. Immediately after Eq. (54) the paper states that these bounds 'only guarantee convergence when using α_n = γ_n' and that 'in practice, γ_n ≠ α_n in most cases.' The implemented update, Eq. (42), also contains an infeasible branch in which the parallel component is not scaled by γ_n and the orthogonal component is damped by μ^h. No descent inequality is proved for this branch, yet Figures 6 and 8 show constraint breakage precisely in the loops where the reported speedups occur. Therefore Theorem 1 establishes convergence of the active-set projection subproblem, not of the outer inertial PGD iteration, and the abstract's claim that the algorithm 'converges to a stable active set that satisfies the KKT criteria' is overstated for the full method. This is a load-bearing gap that should be fixed either by extending the analysis to the implemented branch or by explicitly presenting the method as heuristic with empirical support.
  2. [Section 2.3, Eqs. (55)–(58)] The practical step-size rules are not shown to satisfy the sufficient descent conditions. The paper defines α_n = 1/L using the local estimate (55), γ_n by (57), and β_n by (58) with a user-chosen β_hat, and then states that 'in practice, the choice of Lipschitz constant does not guarantee convergence, as it is only a local approximation that can carry a non-negligible error.' Since the descent conditions (49)–(51) require an accurate global or sufficiently large local Lipschitz constant, and the implementation uses a heuristic estimate, the formal guarantee does not apply to the parameter settings used in the numerical experiments. The authors should either add a safeguard (for example, a backtracking or restart rule that provably restores descent when the cost increases or constraints are violated) or clearly label the step-size selection as heuristic and separate it from the formal convergence statements.
  3. [Section 2.1.2, Theorem 1 proof] The proof of Theorem 1 has a gap in the fallback mechanism. Step 6.d.ii reintroduces the 'most binding' broken constraint and the proof asserts that repeated single-constraint manipulation converges by citing classical active-set literature, but no direct argument is given that the fallback loop terminates without cycling, particularly after multiple constraints have been removed from the stored copy of J_{o-1}. The lemmas rely on convexity and self-compatibility, but the proof does not show that the algorithm cannot cycle between the reintroduction and removal steps. Since Theorem 1 is a central formal claim, the proof should be made rigorous for this fallback path, or the theorem should be restricted to the cases where the fallback is not needed.
minor comments (4)
  1. [Section 2.1.2] The active-set description is difficult to follow because the internal steps 6, 6a–6d and the flowchart in Figure 1 use inconsistent notation for the 'cost function': in Lemmas 1 and 2 and in the proof of Theorem 1, the cost is the distance ||Δφ_tilde - Δφ||, but step 6 is described as checking whether the 'cost function increased' while Eq. (25) is written as a distance inequality. Please clarify that the merit function is the projection distance, not the original objective C.
  2. [Section 3.3] The numerical comparison is based on a single heat-sink test case with a proprietary finite element solver (DFEM). The claim that the method performs 'comparable to or exceeding MMA' would be strengthened by testing on at least one additional problem class or by reporting results with an open-source solver, especially since the MMA move limit was tuned for this problem (values 0.1 and 0.05 are used in the two cases).
  3. [Section 3.4] Default parameter values (β_hat = 0.2, μ = 0.95, ζ = 0.5, ε_i = 0.02 a_i) appear to be selected using the same test cases that are later used to demonstrate robustness. This is not circular in a mathematical sense, but the paper should acknowledge that the parameter-sensitivity study does not fully validate the 'minimal parameter tuning' claim on independent problems.
  4. [Appendix A] Tables A1–A4 report total iteration counts and counts of fallback events, but they do not directly report the 'frequency' of step 6 or step 6c failures as claimed in the text. Please state the fallback rates as fractions of total iterations, which would make the empirical claim about the rarity of the fallback easier to assess.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the PGD modifications are constructed from gradient and Lagrangian estimates, not from the target solutions, and the performance claims are benchmarked against the external MMA solver; the main gap is an unproven parameter regime, which is a correctness limitation, not circularity.

full rationale

The paper's derivation chain is not circular. Step sizes and relaxation factors are defined from local Lipschitz and Lagrangian-gradient estimates (Eqs. 55-58), not from the optimized solutions or cost values used to evaluate performance. The active-set projection convergence (Theorem 1, Section 2.1) is proved for the projection subproblem with explicit KKT conditions, and the outer-iteration descent analysis in Section 2.3 states its assumptions (Lipschitz gradient, feasible iterates, alpha_n = gamma_n) rather than hiding them. The paper itself flags the gap: 'These bounds only guarantee convergence when using alpha_n = gamma_n ... In practice, gamma_n != alpha_n in most cases' (after Eq. 54), and also notes 'the choice of Lipchsitz constant does not guarantee convergence' and that the active-set loop 'can still get stuck ... if too many constraints are broken.' These are honest limitations of the formal result, not circular reasoning. Self-citations (DFEM solver, overhang constraint formulation, adjoint implementation) support implementation details and are not load-bearing for the claimed convergence or for the comparison against MMA, which is an external benchmark. Default parameter values (mu = 0.95, beta_hat = 0.2) are described as adequate through heuristics and sensitivity studies on the same test cases, which may weaken the 'minimal tuning' claim empirically, but no fitted parameter is renamed as a prediction and no equation-level reduction of the result to its inputs is present. Therefore the appropriate circularity score is 0.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The central claim rests on standard assumptions (Lipschitz smoothness, linear independence) plus an unverified assumption that the chosen step sizes satisfy the sufficient descent conditions despite the proof gap. The algorithm introduces several user-chosen hyperparameters that are tuned on the test case, which supports the 'minimal tuning' claim only loosely.

free parameters (5)
  • beta_hat (inertia parameter) = 0.2 (default, range [0,1])
    Controls the inertial momentum term in Equation (58); user-chosen and tested in Section 3.4.3; not derived from first principles.
  • mu (constraint breakage relaxation) = 0.95 (default, in (0,1])
    Relaxes the orthogonal update when constraints are broken (Equation (42)); user-chosen; parameter study in Section 3.4.2 finds 0.9 to 0.95 optimal.
  • epsilon_i (constraint tolerance) = 0.02 * a_i (default)
    User tolerance defining when a constraint counts as broken; affects convergence speed and constraint satisfaction (Section 3.4.1).
  • zeta (active-set oversize relaxation) = 0.5
    Relaxes the initial step when the active set exceeds the design dimension (step 4 of the active set algorithm); chosen ad hoc.
  • initial step factor = 0.1
    Scales the initial step when no previous iterate exists (Equation (59)); heuristic choice without prior-step information.
assumptions (6)
  • domain assumption The cost function gradient is Lipschitz continuous.
    Assumed at the start of Section 2.3 (Equation (43)) to derive the descent inequality.
  • domain assumption All active constraints are linearly independent and the feasible set Q is nonempty.
    Assumed before Theorem 1 to prove active-set convergence; stated in Section 2.1.2.
  • domain assumption Constraints are differentiable and are linearized around the current iterate for the projection step.
    Used throughout Section 2.1; standard practice but an approximation for nonlinear constraints and a source of iterative error.
  • ad hoc to paper The step-size choices satisfy the sufficient descent conditions (49) to (51) in practice.
    The convergence proof only covers alpha_n = gamma_n and feasible iterates; the paper acknowledges gamma_n differs from alpha_n in practice, so this is an unverified assumption.
  • ad hoc to paper The relaxation parameters mu and epsilon_i preserve descent and do not destabilize the iteration.
    No formal analysis of the infeasible branch of Equation (42); behavior is demonstrated empirically in Section 3.4.
  • domain assumption The three-field filtering and projection (Helmholtz filter and smoothed Heaviside) yield a well-posed relaxed topology optimization problem.
    Standard in topology optimization; used in Section 3.2 and relied on for the numerical comparisons.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Improving the Robustness of the Projected Gradient Descent Method for Nonlinear Constrained Optimization Problems in Topology Optimization." pith.science (2026). https://pith.science/paper/PLFHY4ZQ

@misc{pith2026241207634,
  author       = {Pith},
  title        = {Pith review of: Improving the Robustness of the Projected Gradient Descent Method for Nonlinear Constrained Optimization Problems in Topology Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PLFHY4ZQ}},
  note         = {Machine review of arXiv:2412.07634}
}
read the original abstract

The Projected Gradient Descent (PGD) algorithm is a widely used and efficient first-order method for solving constrained optimization problems due to its simplicity and scalability in large design spaces. Building on recent advancements in the PGD algorithm where an inertial step component has been introduced to improve efficiency in solving constrained optimization problems this study introduces two key enhancements to further improve the algorithm's performance and adaptability in large-scale design spaces. First, univariate constraints (such as design variable bounds constraints) are directly incorporated into the projection step via the Schur complement and an improved active set algorithm with bulk constraints manipulation, avoiding issues with min-max clipping. Second, the update step is decomposed relative to the constraint vector space, enabling a post-projection adjustment based on the state of the constraints and an approximation of the Lagrangian, significantly improving the algorithm's robustness for problems with nonlinear constraints. Applied to a topology optimization problem for heat sink design, the proposed PGD algorithm demonstrates performance comparable to or exceeding that of the Method of Moving Asymptotes (MMA), with minimal parameter tuning. These results position the enhanced PGD as a robust tool for complex optimization problems with large variable space, such as topology optimization problems.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 38 canonical work pages

  1. [1]

    A review of the design methods of complex topology structures for 3D printing

    Feng J, Fu J, Lin Z, Shang C, Li B. A review of the design methods of complex topology structures for 3D printing. Visual Computing for Industry, Biomedicine, and Art. 2018;1:1–16. doi: 10.1186/s42492-018-0004-3

    work page doi:10.1186/s42492-018-0004-3 2018

  2. [2]

    Topology Optimization of Continuum Structures: A Review.Applied Mechanics Reviews

    Eschenauer HA, Olhoff N. Topology Optimization of Continuum Structures: A Review.Applied Mechanics Reviews. 2001;54(4):331–390. doi: 10.1115/1.1388075

    work page doi:10.1115/1.1388075 2001

  3. [3]

    A review about the engineering design of optimal heat transfer systems using topology optimization

    Dbouk T. A review about the engineering design of optimal heat transfer systems using topology optimization. Applied Thermal Engineering. 2017;112:841–854. doi: 10.1016/j.applthermaleng.2016.10.134

    work page doi:10.1016/j.applthermaleng.2016.10.134 2017

  4. [4]

    A Review of Topology Optimisation for Fluid-Based Problems.Fluids

    Alexandersen J, Andreasen CS. A Review of Topology Optimisation for Fluid-Based Problems.Fluids. 2020;5(1):29. doi: 10.3390/fluids5010029

    doi:10.3390/fluids5010029 2020

  5. [5]

    The method of moving asymptotes—a new method for structural optimization

    Svanberg K. The method of moving asymptotes—a new method for structural optimization. International Journal for Numerical Methods in Engineering. 1987;24(2):359–373. doi: 10.1002/nme.1620240207

    doi:10.1002/nme.1620240207 1987

  6. [6]

    An SLP algorithm and its application to topology optimization

    Gomes FA, Senne TA. An SLP algorithm and its application to topology optimization. Computational & Applied Mathematics. 2011;30:53–89. doi: 10.1590/S1807-03022011000100003

    work page doi:10.1590/s1807-03022011000100003 2011

  7. [7]

    An efficient second-order SQP method for structural topology optimization

    Rojas-Labanda S, Stolpe M. An efficient second-order SQP method for structural topology optimization. Structural and Multidisciplinary Optimization. 2016;53(6):1315–1333. doi: 10.1007/s00158-016-1471-7

    work page doi:10.1007/s00158-016-1471-7 2016

  8. [8]

    A review of homogenization and topology optimization III—topology optimization using optimality criteria

    Hassani B, Hinton E. A review of homogenization and topology optimization III—topology optimization using optimality criteria. Computers & Structures. 1998;69(6):739–756. doi: 10.1016/S0045-7949(98)00131-6

    work page doi:10.1016/s0045-7949(98)00131-6 1998

Show all 45 references
  1. [9]

    Primal-dual Newton-type interior-point method for topology optimization.Journal of Optimization Theory and Applications

    Hoppe RH, Petrova SI, Schulz V. Primal-dual Newton-type interior-point method for topology optimization.Journal of Optimization Theory and Applications. 2002;114:545–571. doi: 10.1023/A:1014072503048

    2002 doi

  2. [10]

    Multimaterial topology optimization by volume constrained Allen–Cahn system and regularized projected steepest descent method

    Tavakoli R. Multimaterial topology optimization by volume constrained Allen–Cahn system and regularized projected steepest descent method. Computer Methods in Applied Mechanics and Engineering. 2014;276:534–565. doi: 10.1016/j.cma.2014.04.009

    2014 doi

  3. [11]

    Inertial projected gradient method for large-scale topology optimization.Japan Journal of Industrial and Applied Mathematics

    Nishioka A, Kanno Y. Inertial projected gradient method for large-scale topology optimization.Japan Journal of Industrial and Applied Mathematics. 2023;40(2):877–905. doi: https://doi.org/10.1007/s13160-023-00563-0

    2023 doi

  4. [12]

    Parameter-free projected gradient descent

    Chzhen E, Giraud C, Stoltz G. Parameter-free projected gradient descent. arXiv preprint arXiv:2305.19605. 2023

    2023

  5. [13]

    Boosting the robustness of neural networks with M-PGD

    He C, Zhou L, Zhang K, et al. Boosting the robustness of neural networks with M-PGD. In: Springer. 2022:562–573

    2022

  6. [14]

    A random active set method for strictly convex quadratic problem with simple bounds.Mathematics of Computation

    Gu R, Gao B. A random active set method for strictly convex quadratic problem with simple bounds.Mathematics of Computation. 2024. doi: 10.1090/mcom/3647

    2024 doi

  7. [15]

    Minimizing quadratic functions subject to bound constraints with the rate of convergence and finite termination.Computational Optimization and Applications

    Dostal Z, Schoberl J. Minimizing quadratic functions subject to bound constraints with the rate of convergence and finite termination.Computational Optimization and Applications. 2005;30:23–43. doi: 10.1007/s10589-005-4584-7

    2005 doi

  8. [16]

    An infeasible active set method for quadratic problems with simple bounds.SIAM Journal on Optimization

    Kunisch K, Rendl F. An infeasible active set method for quadratic problems with simple bounds.SIAM Journal on Optimization. 2003;14(1):35–52. doi: 10.1137/S1052623401382206

    2003 doi

  9. [17]

    A feasible active set method for strictly convex quadratic problems with simple bounds.SIAM Journal on Optimization

    Hungerlander P, Rendl F. A feasible active set method for strictly convex quadratic problems with simple bounds.SIAM Journal on Optimization. 2015;25(3):1633–1659. doi: 10.1137/140961676

    2015 doi

  10. [18]

    Primal and dual active-set methods for convex quadratic programming.Mathematical Programming.2016;159(1):469–

    Forsgren A, Gill PE, Wong E. Primal and dual active-set methods for convex quadratic programming.Mathematical Programming.2016;159(1):469–

    2016

  11. [19]

    An active set strategy for solving optimization problems with up to 200,000,000 nonlinear constraints

    Schittkowski K. An active set strategy for solving optimization problems with up to 200,000,000 nonlinear constraints. Applied Numerical Mathematics. 2009;59(12):2999–3007. doi: 10.1016/j.apnum.2009.07.009

    2009 doi

  12. [20]

    A two-stage active-set algorithm for bound-constrained optimization

    Cristofari A, De Santis M, Lucidi S, Rinaldi F. A two-stage active-set algorithm for bound-constrained optimization. Journal of Optimization Theory and Applications. 2017;172:369–401. doi: 10.1007/s10957-016-1024-9

    2017 doi

  13. [21]

    Topology optimization for additive manufacturing: considering maximum overhang constraint

    Gaynor AT, Meisel NA, Williams CB, Guest JK. Topology optimization for additive manufacturing: considering maximum overhang constraint. In: 2014:2036

    2014

  14. [22]

    Undercut and overhang angle control in topology optimization: A density gradient based integral approach

    Qian X. Undercut and overhang angle control in topology optimization: A density gradient based integral approach. International Journal for Numerical Methods in Engineering. 2017;111(3):247–272. doi: 10.1002/nme.5461

    2017 doi

  15. [23]

    Additively manufactured conformal cooling channels through topology optimization

    Lamarche-Gagnon M ´E, Molavi-Zarandi M, Raymond V, Ilinca F. Additively manufactured conformal cooling channels through topology optimization. Structural and Multidisciplinary Optimization. 2024;67(8):1–19. doi: 10.1007/s00158-024-03846-3

    2024 doi

  16. [24]

    Topology optimization considering overhang constraints: Eliminating sacrificial support material in additive manufacturing through design

    Gaynor AT, Guest JK. Topology optimization considering overhang constraints: Eliminating sacrificial support material in additive manufacturing through design. Structural and Multidisciplinary Optimization. 2016;54(5):1157–1172. doi: 110.1007/s00158-016-1551-x

    2016

  17. [25]

    First-order methods in optimization

    Beck A. First-order methods in optimization. SIAM, 2017 Improving the Robustness of the Projected Gradient Descent Method for Nonlinear Constrained Optimization Problems in Topology Optimization 27

    2017

  18. [26]

    Improved analysis of clipping algorithms for non-convex optimization.Advances in Neural Information Processing Systems

    Zhang B, Jin J, Fang C, Wang L. Improved analysis of clipping algorithms for non-convex optimization.Advances in Neural Information Processing Systems. 2020;33:15511–15521

    2020

  19. [27]

    QPSchur: a dual, active-set, Schur-complement method for large-scale and structured convex quadratic programming

    Bartlett RA, Biegler LT. QPSchur: a dual, active-set, Schur-complement method for large-scale and structured convex quadratic programming. Optimization and Engineering. 2006;7:5–32. doi: 10.1007/s11081-006-6588-z

    2006 doi

  20. [28]

    Active-set methods for quadratic programming

    Wong E. Active-set methods for quadratic programming. University of California, San Diego, 2011

    2011

  21. [29]

    DFEMwork: A Parallel Computing Framework for Material Processing

    Audet M, H ´etu JF, Ilinca F. DFEMwork: A Parallel Computing Framework for Material Processing. National Research Council Canada Publications Archive. 2008:1–10. Collection: NRC Publications Archive / Archives des publications du CNRC, Record ID: c971447f-d5ef-4206- a15f-c206abce54b9

    2008

  22. [30]

    Development of a Topology Optimization Framework For Cooling Channel Design in Die Casting Molds

    Navah F, Lamarche-Gagnon M´E, Ilinca F, Audet M, Molavi-Zarandi M, Raymond V. Development of a Topology Optimization Framework For Cooling Channel Design in Die Casting Molds. In: . 85666. American Society of Mechanical Engineers. 2021:V010T10A054

    2021

  23. [31]

    A Projected Gradient and Constraint Linearization Method for Nonlinear Model Predictive Control

    Torrisi G, Grammatico S, Smith RS, Morari M. A Projected Gradient and Constraint Linearization Method for Nonlinear Model Predictive Control. SIAM Journal on Control and Optimization. 2018;56(3):1968-1999. doi: 10.1137/16M1098103

    2018 doi

  24. [32]

    Constrained optimization and Lagrange multiplier methods

    Bertsekas DP. Constrained optimization and Lagrange multiplier methods. Academic press, 2014

    2014

  25. [33]

    Numerical optimization

    Nocedal J, Wright SJ. Numerical optimization. Springer, 1999

    1999

  26. [34]

    Linear Algebra and Its Applications

    Strang G. Linear Algebra and Its Applications. Cengage Learning, 2012

    2012

  27. [35]

    Convex Optimization

    Boyd S, Vandenberghe L. Convex Optimization. Cambridge University Press, 2004

    2004

  28. [36]

    dSJ, et al

    Harris CR, Millman KJ, Walt v. dSJ, et al. Array programming with NumPy. Nature. 2020;585(7825):357–362. doi: 10.1038/s41586-020-2649-2

    2020 doi

  29. [37]

    Thermofluid Topology Optimization for Cooling Channel Design

    Navah F, Lamarche-Gagnon M ´E, Ilinca F. Thermofluid Topology Optimization for Cooling Channel Design. Applied Thermal Engineering. 2024;236:121317. doi: 10.1016/j.applthermaleng.2023.121317

    2024

  30. [38]

    Filters in Topology Optimization Based on Helmholtz-type Differential Equations.International Journal for Numerical Methods in Engineering

    Lazarov BS, Sigmund O. Filters in Topology Optimization Based on Helmholtz-type Differential Equations.International Journal for Numerical Methods in Engineering. 2011;86(6):765–781. doi: 10.1002/nme.3072

    2011 doi

  31. [39]

    On projection methods, convergence and robust formulations in topology optimization

    Wang F, Lazarov BS, Sigmund O. On projection methods, convergence and robust formulations in topology optimization. Structural and Multidisciplinary Optimization. 2011;43:767–784. doi: 10.1007/s00158-010-0602-y

    2011 doi

  32. [40]

    Generating optimal topologies in structural design using a homogenization method

    Bendsøe MP, Kikuchi N. Generating optimal topologies in structural design using a homogenization method. Computer Methods in Applied Mechanics and Engineering. 1988;71(2):197-224. doi: 10.1016/0045-7825(88)90086-2

    1988 doi

  33. [41]

    Volume 11: Heat Transfer and Thermal Engineering of ASME International Mechanical Engineering Congress and Exposition; : 2021

    A Comparative Study Between a Sharp and a Diffuse Topology Optimization Method for Thermal Problems. Volume 11: Heat Transfer and Thermal Engineering of ASME International Mechanical Engineering Congress and Exposition; : 2021

    2021

  34. [42]

    Gmsh.; 2020

    Geuzaine C, Remacle JF. Gmsh.; 2020. A three-dimensional finite element mesh generator with built-in pre- and post-processing facilities

    2020

  35. [43]

    The NLopt Nonlinear-Optimization Package

    Johnson SG. The NLopt Nonlinear-Optimization Package. https://github.com/stevengj/nlopt; 2007. An open-source library for nonlinear optimization

    2007

  36. [44]

    Attacking Large Language Models with Projected Gradient Descent.arXiv preprint arXiv:2402.09154

    Geisler S, Wollschl¨ager T, Abdalla M, Gasteiger J, G¨ unnemann S. Attacking Large Language Models with Projected Gradient Descent.arXiv preprint arXiv:2402.09154. 2024. How to cite this article: Barbeau L., Lamarche-Gagnon M.- ´E., and Ilinca F.. Improving the Robustness of t...

    2024 arXiv

  37. [508]

    doi: 10.1007/s10107-015-0966-2

    doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.