Pith. sign in

REVIEW 5 major objections 6 minor 62 references

Tight Constraint Prediction of Six-Degree-of-Freedom Transformer-based Powered Descent Guidance

T0 review · 5 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Transformer-based Successive Convexification predicts which constraints bind at the optimum, then solves only those to cut 6-DoF Mars powered descent guidance solve time from 14.61 s to 4.98 s.

desk verdict A genuinely useful empirical extension of T-PDG to 6-DoF SCvx with a clever rotation augmentation and honest benchmarks, but the feasibility guarantee is not supported by the algorithm as written. read the letter →

arxiv 2501.00930 v1 pith:7DURBQEG submitted 2025-01-01 math.OC cs.LGcs.ROcs.SYeess.SY

Julia Briden , Trey Gurga , Breanna Johnson , Abhishek Cauligi , Richard Linares This is my paper · ORCID
classification math.OCcs.LGcs.ROcs.SYeess.SY MSC 90C30
keywords T-SCvxtightconstraintpredictiontransformerneuralnetworksuccessiveconvexificationsix-degree-of-freedompowereddescentguidancerotation-invariantdataaugmentationMarslandingreal-timetrajectoryoptimization
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 claims that a transformer network can learn which constraints are active at the optimum of a 6-DoF powered descent guidance problem, and that solving only those constraints plus a learned warm start cuts solve time dramatically. On a 50-timestep free-final-time Mars landing case, the resulting algorithm, T-SCvx, reduces mean solve time from 14.61 s to 4.98 s, a 66% reduction, while keeping the full-problem penalty check that guarantees feasibility of returned trajectories. The paper also shows that rotating the training data about the vertical axis preserves the active-constraint pattern, so fewer than 2,000 raw samples suffice for over 96% binary accuracy in constraint prediction. The significance is that fast, reliable onboard trajectory generation for high-fidelity 6-DoF descent becomes plausible on flight-class hardware.

What carries the argument

The load-bearing machinery is a transformer neural network that maps a 17-dimensional problem-parameter vector (initial position, velocity, attitude, angular velocity, mass, glideslope angle, pitch angle, and SCvx iteration number) to a binary vector marking which discretized inequality constraints are tight, plus a second transformer that predicts the full state, control, and final time. Rotation-invariant data augmentation around the Up axis inflates 1,592 raw samples to 11,634, exploiting the symmetry that the active-constraint pattern is unchanged under such rotations. The reduced subproblem is solved with only the predicted tight constraints, and its solution warm-starts the full problem so that feasibility is checked against the original constraints.

What would settle it

Run T-SCvx and plain SCvx on a batch of initial conditions deliberately drawn outside the training ranges; if any T-SCvx trajectory violates the glideslope, thrust-magnitude, gimbal, or attitude constraints, or has measurably higher fuel use, while the SCvx trajectory is feasible, the reduced-problem surrogate claim is falsified. A sharper check is to compare the network's predicted tight-constraint set with the true active set at the converged solution for those out-of-distribution cases and show the mismatch rate grows with Mahalanobis distance from the training data.

Watch

Extended reading notes

Core claim

The central claim is that the optimal solution of the full nonconvex problem is recovered by the reduced problem formed from equality constraints plus the predicted tight inequality constraints, so a neural network can replace most of the repeated convexification work. T-SCvx predicts both the tight-constraint set and a full state, control, and final-time warm start at each successive-convexification iteration, scales the trust region by the fraction of constraints that changed, and accepts a solution only when the full-problem penalty cost shows no improvement. In the 6-DoF Mars powered landing test with 50 timesteps and free final time, this reduces mean solve time from 14.61 s to 4.98 s, a 66% mean reduction and a 70% median reduction, with feasibility enforced by warm-starting the full problem.

Load-bearing premise

The load-bearing premise is that the predicted tight-constraint set is a faithful surrogate for the full nonconvex problem, so solving the reduced subproblem and warm-starting the full problem returns a feasible, locally optimal trajectory; if the prediction is wrong on out-of-distribution inputs, the returned trajectory can violate constraints or be suboptimal.

Editorial extensions

If this is right

  • T-SCvx reaches a mean 4.98 s solve time on the 50-timestep free-final-time 6-DoF Mars PDG benchmark, versus 14.61 s for SCvx, a 66% mean and 70% median reduction.
  • Training requires fewer than 2,000 raw samples: rotation augmentation gives 11,634 samples, and the constraint-prediction network reaches 96.45% binary accuracy on held-out test cases.
  • Because the full-problem penalty is retained in the convergence check, returned trajectories are asserted feasible and locally optimal under SCvx's convergence conditions.
  • T-SCvx beats linear-interpolation lookup tables by more than 99% in inference time and memory usage, and generalizes better to out-of-distribution inputs than k-d-tree lookup.
  • The same active-set prediction and warm-starting recipe can be carried into active-set-based nonlinear programming solvers, as the paper identifies as future work.

Reading between the lines

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

  • If the tight-constraint surrogate holds in general, the active-set-plus-warm-start recipe could apply to other sequential convex programming problems beyond powered descent, since the reduction is formulation-agnostic.
  • The rotation-invariant augmentation suggests a broader principle: any symmetry of the dynamics that leaves the active set invariant can be encoded cheaply in the training data rather than in the network architecture, potentially lowering sample counts in other guidance problems.
  • The reported speedups are on a particular convex solver; flight hardware may show different gains, and the paper itself lists verification on radiation-hardened processors as future work.
  • A testable extension is to use the predicted tight-constraint set to sparsify the linear algebra inside the convex solver, which would compound the wall-clock savings by reducing per-iteration factorization cost.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. This paper proposes T-SCvx, a learning-based warm-start and problem-reduction method for the 6-DoF powered descent guidance problem. A transformer predicts the set of tight/active constraints at each SCvx iteration and also predicts a full state/control/final-time initial guess. The reduced subproblem defined by the predicted tight constraints is then solved, and the paper claims this preserves feasibility and local optimality while reducing mean solve time by 66% (from 14.61 s to 4.98 s) on 545 held-out test cases. A rotation-invariant data augmentation scheme is used to expand a base dataset of roughly 1,592 samples to 11,634 samples. The method is benchmarked against SCvx and against linear-interpolation and kd-tree lookup-table approaches.

Significance. Fast onboard 6-DoF powered descent guidance is an important and active problem, and the idea of learning active sets to reduce successive convexification subproblems is attractive. The rotation-augmentation contribution is concrete and potentially useful beyond this specific application, and the paper provides a substantial empirical evaluation against SCvx and lookup-table baselines. However, the central feasibility guarantee is not supported by the algorithms as written, and the data-splitting procedure may compromise the reported generalization results. These issues are significant but appear fixable within the scope of a revision.

major comments (5)
  1. [Section III.C, Claim and Proof] The proof of the Claim is incomplete and internally unclear. It assumes a feasible sequence with a descent direction d satisfying d^T grad f(x*) < 0 and then concludes that the active set suffices, but it never shows that a solution to the reduced problem satisfies the omitted inequality constraints of the original nonconvex problem. The displayed inequality 'f(z_k) < f(x*) + (1/2)||z_k - x*|| d^T grad f(x*)' is also problematic: since d^T grad f(x*) < 0, the right-hand side is less than f(x*), so the inequality would contradict local optimality rather than establish it. As written, the proof does not support the load-bearing assertion that solving the reduced problem recovers the full optimal solution.
  2. [Section III.E, Algorithm 3] Algorithm 3 returns the output of Reduced-Solve directly and contains no full-problem feasibility check, no penalty-residual verification, and no warm-started full SCvx continuation. The statement in Section III.E that 'the full problem is used for the evaluation cost function' is not reflected in the algorithm, because the cost returned by Reduced-Solve is computed on the reduced subproblem. This is a load-bearing gap: Table 3 shows the constraint classifier achieves only 96.45% binary accuracy versus 95.95% for a predict-only-zeros baseline, so roughly 4% of tight-constraint predictions are wrong on the test set, and nothing in Algorithm 3 checks whether those errors are benign. Even in Algorithm 2, the stopping test Delta J = 0 does not by itself certify feasibility, since the exact penalty J can be positive at a stationary point of the penalty; the cited SCvx theory only guarantees a KKT point when the limit point is feasible. The claimed feasibility guarantee therefore requires either an explicit full-problem verification step in Algorithm 3, a warm-started full SCvx solve, or a proof that wrong predictions cannot cause constraint violation.
  3. [Section III.E.3, Training, Validation, and Testing] The data split is performed after rotation augmentation: the text states 'All sampled data, including the rotated samples, were split into 80% training and validation data and 20% test data.' This allows rotated copies of the same base sample to appear in both the training and test sets, because the rotation is applied before the split and no grouping by base sample is described. As a result, the reported 96.45% binary accuracy and the 66% solve-time reduction may reflect memorization of augmented variants rather than generalization to unseen initial conditions. The split should be performed on the 1,592 base samples before augmentation, or the authors should verify that no test sample shares a base sample with any training sample.
  4. [Table 3, Constraint NN Accuracy] The reported constraint-prediction accuracy is only 0.5 percentage points above the predict-only-zeros baseline (96.45% versus 95.95%). Because tight-constraint labels are presumably sparse, binary accuracy is dominated by the zero class, and a predictor that never predicts a tight constraint can already achieve 95.95%. The paper's runtime and feasibility arguments depend on the quality of the positive (tight) predictions, so precision, recall, and F1 for the tight class should be reported, together with the effect of false negatives and false positives on the returned trajectories.
  5. [Section III.D, Algorithm 2, Trust-Region Modification] Algorithm 2 modifies the SCvx update by scaling the trust region with the predicted constraint-change fraction tau_r and by solving subproblems built only from predicted tight constraints, but no convergence or optimality analysis is provided for this modified algorithm. The SCvx guarantees cited from [19] apply to the full convex subproblem; it is not immediate that they carry over when constraints are dropped at each iteration and the trust-region radius is scaled by a learned quantity. The paper should state explicitly which convergence and feasibility guarantees T-SCvx inherits, or should reframe the relevant claims as empirical rather than as consequences of the SCvx theory.
minor comments (6)
  1. [Section IV.B] The sentence 'the standard deviation for T-SCvx is slightly higher than SCvx, at 5.24 seconds' should specify which distribution the 5.24 s standard deviation refers to, and the following sentence 'one standard deviation remains below the SCvx mean of 14.61' is ambiguous because a standard deviation is not a bound on the mean.
  2. [Section VI, Conclusion] The conclusion states that the networks achieve 'test accuracies of over 96% and under 1 MSE,' but Table 3 reports a solution NN test MSE of 1.040, which is not under 1. Please correct the inconsistency.
  3. [Section III.B.1] The text refers to 'Equation (21)' and 'Eq. (22)' for the trust-region update, but the relevant equations in the manuscript appear to be (14) and (15); the equation numbering in the surrounding text is inconsistent and should be harmonized.
  4. [Nomenclature] The nomenclature lists 'omega = penalty coefficients,' but the penalty weights in the formulation are denoted lambda and tau; please correct this entry.
  5. [Section IV.A] The text says 'less than 1,600 samples' and later gives '1,592 samples'; please use one consistent number throughout.
  6. [Abstract and Section V] The abstract and conclusion state that T-SCvx 'enables onboard computation,' but Section V notes that verification on flight-grade radiation-hardened hardware remains future work. Please soften the wording to avoid overclaiming.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; minor self-citation of T-PDG is not load-bearing.

full rationale

The paper's evaluation is a standard supervised-learning benchmark: the tight-constraint and solution networks are trained on SCvx-generated data and tested on a held-out 20% split (Section III.E.3). The reported 66% mean solve-time reduction is measured on 545 test samples (Section IV.B), not derived from the training fit, so the central speedup claim is not a fitted input renamed as a prediction. The theoretical Claim in Section III.C is not a circular derivation: it asserts that if one knows the active set at a local solution, the reduced problem recovers that solution; although the proof is incomplete and the statement is not generally true without additional constraint qualifications, this is a correctness gap rather than a definitional equivalence between input and output. The paper self-cites T-PDG [5] for the transformer architecture and the tight-constraint prediction idea, but that citation is background and is not load-bearing: the networks are retrained on 6-DoF SCvx data and benchmarked against SCvx and lookup tables. Algorithm 3's feasibility guarantee is under-supported because Reduced-Solve is returned without a full-problem verification step, but that is an algorithmic/completeness issue, not circularity. No equation in the paper reduces to its own inputs by construction.

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

The central claim rests on SCvx convergence theory imported from prior work, on an unproved active-set reduction claim, and on several hand-chosen hyperparameters and sampling choices. No new physical entities are introduced.

free parameters (5)
  • reduced-problem trust region initialization eta_reduced,init = 0.01
    Chosen by hand, smaller than the full-problem value, based on the assumption that the predicted warm start is close to the solution; affects convergence behavior and reported solve times.
  • SCvx penalty weight lambda = 500
    Hand-selected penalty weight in the exact penalty function; directly affects the number of SCvx iterations and the reported runtime comparison.
  • neural network hyperparameters = constraint NN: 384-dim, 2 heads, 4 layers, dropout 0.1; solution NN: 768-dim, 2 heads, 4 layers, dropout 0.1; batch…
    Chosen by the authors without a reported sensitivity study; these choices affect the reported accuracies and the resulting warm-start quality.
  • rotation augmentation angles = 0, 45, 90, 135, 180, 225, 270, 315 degrees
    Hand-selected set of eight rotations; assumes these discrete rotations sufficiently cover the continuous symmetry about the Up axis.
  • maximum SCvx iterations itermax = 20
    Problems not converged within 20 iterations are labeled infeasible; this cutoff shapes the training dataset and the reported success rate.
assumptions (6)
  • standard math LICQ holds at local optima of the sampled 6-DoF problems
    Invoked from the cited SCvx theory [19] to guarantee convergence to a stationary point; not verified per sample.
  • domain assumption KL property and Lipschitz continuous gradients hold for the 6-DoF problem
    Required for strong convergence and uniqueness of SCvx solutions according to [19]; not checked for this problem instance.
  • domain assumption Tight constraint sets are invariant under rotations about the Up axis
    Section III.E.2 asserts all induced constraints are symmetric about the Up axis, enabling rotation-based data augmentation; no proof is given.
  • ad hoc to paper The reduced problem defined by predicted active constraints recovers the full optimal solution
    The Claim in Section III.C is meant to justify the method, but its proof is incomplete and the statement is not generally true for nonconvex problems without additional regularity.
  • domain assumption The trained networks generalize from the training distribution to test and out-of-distribution problems
    Standard machine-learning assumption; only a Mahalanobis-distance-based OOD proxy is used, not a deployment-style stress test.
  • domain assumption The N=50, FOH discretization is a sufficiently accurate approximation of the continuous-time problem
    The introduction acknowledges the discretization scheme heavily influences solution validity; the paper does not validate discretization fidelity for the reported trajectories.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tight Constraint Prediction of Six-Degree-of-Freedom Transformer-based Powered Descent Guidance." pith.science (2026). https://pith.science/paper/7DURBQEG

@misc{pith2026250100930,
  author       = {Pith},
  title        = {Pith review of: Tight Constraint Prediction of Six-Degree-of-Freedom Transformer-based Powered Descent Guidance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7DURBQEG}},
  note         = {Machine review of arXiv:2501.00930}
}
read the original abstract

This work introduces Transformer-based Successive Convexification (T-SCvx), an extension of Transformer-based Powered Descent Guidance (T-PDG), generalizable for efficient six-degree-of-freedom (DoF) fuel-optimal powered descent trajectory generation. Our approach significantly enhances the sample efficiency and solution quality for nonconvex-powered descent guidance by employing a rotation invariant transformation of the sampled dataset. T-PDG was previously applied to the 3-DoF minimum fuel powered descent guidance problem, improving solution times by up to an order of magnitude compared to lossless convexification (LCvx). By learning to predict the set of tight or active constraints at the optimal control problem's solution, Transformer-based Successive Convexification (T-SCvx) creates the minimal reduced-size problem initialized with only the tight constraints, then uses the solution of this reduced problem to warm-start the direct optimization solver. 6-DoF powered descent guidance is known to be challenging to solve quickly and reliably due to the nonlinear and non-convex nature of the problem, the discretization scheme heavily influencing solution validity, and reference trajectory initialization determining algorithm convergence or divergence. Our contributions in this work address these challenges by extending T-PDG to learn the set of tight constraints for the successive convexification (SCvx) formulation of the 6-DoF powered descent guidance problem. In addition to reducing the problem size, feasible and locally optimal reference trajectories are also learned to facilitate convergence from the initial guess. T-SCvx enables onboard computation of real-time guidance trajectories, demonstrated by a 6-DoF Mars powered landing application problem.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 60 canonical work pages

  1. [19]

    Successive Convexification of Non-Convex Optimal Control Problems and Its Convergence Properties,

    Mao, Y., Szmuk, M., and Açikmeşe, B., “Successive Convexification of Non-Convex Optimal Control Problems and Its Convergence Properties,”Proc. IEEE Conf. on Decision and Control, 2016

    work page 2016

  2. [1]

    A survey of attitude representations,

    Shuster, M. D., “A survey of attitude representations,”Journal of the Astronautical Sciences, Vol. 41, No. 4, 1993, pp. 439 – 517

    work page 1993

  3. [2]

    Successive Convexification for Real-Time Six-Degree-of-Freedom Powered Descent Guidance with State-Triggered Constraints,

    Szmuk, M., Reynolds, T. P., and Açıkmeşe, B., “Successive Convexification for Real-Time Six-Degree-of-Freedom Powered Descent Guidance with State-Triggered Constraints,”AIAA Journal of Guidance, Control, and Dynamics, Vol. 43, No. 8, 2020, pp. 1399–1413

    work page 2020

  4. [3]

    A survey on artificial intelligence trends in spacecraft guidance dynamics and control,

    Izzo, D., Märtens, M., and Pan, B., “A survey on artificial intelligence trends in spacecraft guidance dynamics and control,” Astrodynamics, Vol. 3, 2019, pp. 287–299

    work page 2019

  5. [4]

    Transformers for Trajectory Optimization with Application to Spacecraft Rendezvous,

    Guffanti, T., Gammelli, D., D’Amico, S., and Pavone, M., “Transformers for Trajectory Optimization with Application to Spacecraft Rendezvous,”IEEE Aerospace Conference, 2024

    work page 2024

  6. [5]

    Improving Computational Efficiency for Powered Descent Guidance via Transformer-based Tight Constraint Prediction,

    Briden, J., Gurga, T., Johnson, B. J., Cauligi, A., and Linares, R., “Improving Computational Efficiency for Powered Descent Guidance via Transformer-based Tight Constraint Prediction,”AIAA Scitech Forum, 2024

    work page 2024

  7. [6]

    A Versatile Powered Guidance Algorithm,

    Lu, P., Forbes, S., and Baldwin, M., “A Versatile Powered Guidance Algorithm,”AIAA Conf. on Guidance, Navigation and Control, 2012

    work page 2012

  8. [7]

    Propellant-Optimal Powered Descent Guidance,

    Lu, P., “Propellant-Optimal Powered Descent Guidance,”AIAA Journal of Guidance, Control, and Dynamics, Vol. 41, No. 4, 2018, pp. 813–826

    work page 2018

Show all 62 references
  1. [8]

    E.,Optimal control theory: an introduction, Courier Corporation, 2012

    Kirk, D. E.,Optimal control theory: an introduction, Courier Corporation, 2012

    2012

  2. [9]

    Apollo Lunar Descent Guidance,

    Klumpp, A. R., “Apollo Lunar Descent Guidance,”Automatica, Vol. 10, No. 2, 1974, pp. 133–146. 19

    1974

  3. [10]

    Computationally Inexpensive Guidance Algorithm for Fuel-Efficient Terminal Descent,

    Najson, F., and Mease, K. D., “Computationally Inexpensive Guidance Algorithm for Fuel-Efficient Terminal Descent,”AIAA Journal of Guidance, Control, and Dynamics, Vol. 29, No. 4, 2006, pp. 955–964

    2006

  4. [11]

    Minimum-Fuel Powered Descent for Mars Pinpoint Landing,

    Topcu, U., Casoliva, J., and Mease, K. D., “Minimum-Fuel Powered Descent for Mars Pinpoint Landing,”AIAA Journal of Spacecraft and Rockets, Vol. 44, No. 2, 2007, pp. 324–331

    2007

  5. [12]

    Powered descent guidance methods for the Moon and Mars,

    Sostaric, R. R., and Rea, J. R., “Powered descent guidance methods for the Moon and Mars,”AIAA Conf. on Guidance, Navigation and Control, 2005

    2005

  6. [13]

    An Algorithm for Nonlinear Optimization Using Linear Programming and Equality Constrained Subproblems,

    Byrd, R. H., Gould, N. I. M., Nocedal, J., and Waltz, R. A., “An Algorithm for Nonlinear Optimization Using Linear Programming and Equality Constrained Subproblems,”Mathematical Programming, Vol. 100, 2003, pp. 27–48

    2003

  7. [14]

    Nonlinear Optimization by Successive Linear Programming,

    Palacios-Gomez, F., Lasdon, L., and Engquist, M., “Nonlinear Optimization by Successive Linear Programming,”Management Science, Vol. 28, No. 10, 1982, pp. 1106–1120

    1982

  8. [15]

    DC programming: Overview,

    Horst, R., and Thoai, N. V., “DC programming: Overview,”Journal of Optimization Theory & Applications, Vol. 103, 1999, pp. 1–43

    1999

  9. [16]

    The Concave-Convex Procedure (CCCP),

    Yuille, A., and Rangarajan, A., “The Concave-Convex Procedure (CCCP),”Conf. on Neural Information Processing Systems, 2001

    2001

  10. [17]

    Solving Nonconvex Optimal Control Problems by Convex Optimization,

    Liu, X., and Lu, P., “Solving Nonconvex Optimal Control Problems by Convex Optimization,”AIAA Journal of Guidance, Control, and Dynamics, Vol. 37, No. 3, 2014, pp. 750 – 765

    2014

  11. [18]

    Entry trajectory optimization by second-order cone programming,

    Liu, X., Shen, Z., and Lu, P., “Entry trajectory optimization by second-order cone programming,”AIAA Journal of Guidance, Control, and Dynamics, Vol. 39, No. 2, 2015, pp. 227–241

    2015

  12. [20]

    Constrained Trajectory Optimization for Planetary Entry via Sequential Convex Programming,

    Wang, Z., and Grant, M. J., “Constrained Trajectory Optimization for Planetary Entry via Sequential Convex Programming,” AIAA Journal of Guidance, Control, and Dynamics, Vol. 40, No. 10, 2017, pp. 2603 – 2615

    2017

  13. [21]

    Convex Optimization for Trajectory Generation: A Tutorial on Generating Dynamically Feasible Trajectories Reliably and Efficiently,

    Malyuta, D., Reynolds, T. P., Szmuk, M., Lew, T., Bonalli, R., Pavone, M., and Açıkmeşe, B., “Convex Optimization for Trajectory Generation: A Tutorial on Generating Dynamically Feasible Trajectories Reliably and Efficiently,”IEEE Control Systems Letters, Vol. 42, No. 5, 2022,...

    2022

  14. [22]

    A Globally Convergent Method for Nonlinear Programming,

    Han, S. P., “A Globally Convergent Method for Nonlinear Programming,”Journal of Optimization Theory & Applications, Vol. 22, No. 3, 1977, pp. 297–309

    1977

  15. [23]

    Algorithms for Nonlinear Constraints that use Lagrangian Functions,

    Powell, M. J., “Algorithms for Nonlinear Constraints that use Lagrangian Functions,”Mathematical Programming, Vol. 14, No. 1, 1978, pp. 224–248

    1978

  16. [24]

    A Recursive Quadratic Programming Algorithm that uses Differentiable Exact Penalty Functions,

    Powell, M. J., and Yuan, Y., “A Recursive Quadratic Programming Algorithm that uses Differentiable Exact Penalty Functions,” Mathematical Programming, Vol. 35, 1986, pp. 265–278

    1986

  17. [25]

    A Strategy for Global Convergence in a Sequential Quadratic Programming Algorithm,

    Boggs, P. T., and Tolle, W. J., “A Strategy for Global Convergence in a Sequential Quadratic Programming Algorithm,”SIAM Journal on Numerical Analysis, Vol. 26, No. 3, 1989, pp. 600–623

    1989

  18. [26]

    Sequential Quadratic Programming,

    Boggs, P. T., and Tolle, W. J., “Sequential Quadratic Programming,”Acta Numerica, Vol. 4, 1995, pp. 1–52

    1995

  19. [27]

    A Successive Quadratic Programming Algorithm with Global and Superlinear Convergence Properties,

    Fukushima, M., “A Successive Quadratic Programming Algorithm with Global and Superlinear Convergence Properties,” Mathematical Programming, Vol. 35, 1986, pp. 253–264

    1986

  20. [28]

    SNOPT: An SQP algorithm for large-scale constrained optimization,

    Gill, P. E., Murray, W., and Saunders, M. A., “SNOPT: An SQP algorithm for large-scale constrained optimization,”SIAM Review, Vol. 47, No. 1, 2005, pp. 99–131

    2005

  21. [29]

    Path-Constrained Trajectory Optimization using Sparse Sequential Quadratic Programming,

    Betts, J. T., and Huffman, W. P., “Path-Constrained Trajectory Optimization using Sparse Sequential Quadratic Programming,” AIAA Journal of Guidance, Control, and Dynamics, Vol. 16, No. 1, 1993, pp. 59–68

    1993

  22. [30]

    Lossless Convexification of a Class of Optimal Control Problems with Non-Convex Control Constraints,

    Açıkmeşe, B., and Blackmore, L., “Lossless Convexification of a Class of Optimal Control Problems with Non-Convex Control Constraints,”Automatica, Vol. 47, No. 2, 2011, pp. 341–347

    2011

  23. [31]

    LosslessConvexification of Non-convex Optimal ControlProblems for StateConstrained Linear Systems,

    Harris, M., and Açıkmeşe, B., “LosslessConvexification of Non-convex Optimal ControlProblems for StateConstrained Linear Systems,”Automatica, Vol. 50, No. 9, 2014, pp. 2304–2311. 20

    2014

  24. [32]

    Lossless convexfication of control constraints for a class of nonlinear optimal control problems,

    Blackmore, L., Açıkmeşe, B., and Carson, J. M., “Lossless convexfication of control constraints for a class of nonlinear optimal control problems,”Systems & Control Letters, Vol. 61, No. 4, 2012, pp. 863–871

    2012

  25. [33]

    A powered descent guidance algorithm for Mars pinpoint landing,

    Açıkmeşe, B., and Ploen, S. R., “A powered descent guidance algorithm for Mars pinpoint landing,”AIAA Conf. on Guidance, Navigation and Control, 2005

    2005

  26. [34]

    Convex Programming Approach to Powered Descent Guidance for Mars Landing,

    Açıkmeşe, B., and Ploen, S. R., “Convex Programming Approach to Powered Descent Guidance for Mars Landing,”AIAA Journal of Guidance, Control, and Dynamics, Vol. 30, No. 5, 2007, pp. 1353–1366

    2007

  27. [35]

    Minimum Landing Error Powered Descent Guidance for Mars Landing using Convex Optimization,

    Blackmore, L., Açıkmeşe, B., and Scharf, D. P., “Minimum Landing Error Powered Descent Guidance for Mars Landing using Convex Optimization,”AIAA Journal of Guidance, Control, and Dynamics, Vol. 33, No. 4, 2010

    2010

  28. [36]

    Lossless Convexification of Non-convex Control Bound and Pointing Constraints of the Soft Landing Optimal Control Problem,

    Açıkmeşe, B., Carson, J., and Blackmore, L., “Lossless Convexification of Non-convex Control Bound and Pointing Constraints of the Soft Landing Optimal Control Problem,”IEEE Transactions on Control Systems Technology, Vol. 21, No. 6, 2013, pp. 2104–2113

    2013

  29. [37]

    MaximumDivertforPlanetaryLandingUsingConvexOptimization,

    Harris,M.W.,andAçıkmeşe,B.,“MaximumDivertforPlanetaryLandingUsingConvexOptimization,” JournalofOptimization Theory & Applications, 2013, pp. 1–21

    2013

  30. [38]

    Automated Custom Code Generation for Embedded, Real-time Second Order Cone Programming,

    Dueri, D., Zhang, J., and Açıkmeşe, B., “Automated Custom Code Generation for Embedded, Real-time Second Order Cone Programming,”IFAC World Congress, 2014

    2014

  31. [39]

    Customized Real-Time Interior-Point Methods for Onboard Powered-Descent Guidance,

    Dueri, D., Açıkmeşe, B., Scharf, D. P., and Harris, M. W., “Customized Real-Time Interior-Point Methods for Onboard Powered-Descent Guidance,”AIAA Journal of Guidance, Control, and Dynamics, Vol. 40, No. 2, 2017, pp. 197–212

    2017

  32. [40]

    Press, 2004

    Boyd, S., and Vandenberghe, L.,Convex Optimization, Cambridge Univ. Press, 2004

    2004

  33. [41]

    J.,Numerical Optimization, 2nd ed., Springer, 2006

    Nocedal, J., and Wright, S. J.,Numerical Optimization, 2nd ed., Springer, 2006

    2006

  34. [42]

    Optimal Power Descent Guidance with 6-DoF Line of Sight Constraints via Unit Dual Quaternions,

    Lee, U., and Mesbahi, M., “Optimal Power Descent Guidance with 6-DoF Line of Sight Constraints via Unit Dual Quaternions,” AIAA Scitech Forum, 2015

    2015

  35. [43]

    Constrained Autonomous Precision Landing via Dual Quaternions and Model Predictive Control,

    Lee, U., and Mesbahi, M., “Constrained Autonomous Precision Landing via Dual Quaternions and Model Predictive Control,” AIAA Journal of Guidance, Control, and Dynamics, 2016

    2016

  36. [44]

    Learning Mixed-Integer Convex Optimization Strategies for Robot Planning and Control,

    Cauligi, A., Culbertson, P., Stellato, B., Bertsimas, D., Schwager, M., and Pavone, M., “Learning Mixed-Integer Convex Optimization Strategies for Robot Planning and Control,”Proc. IEEE Conf. on Decision and Control, 2020

    2020

  37. [45]

    CoCo: Online Mixed-Integer Control via Supervised Learning,

    Cauligi, A., Culbertson, P., Schmerling, E., Schwager, M., Stellato, B., and Pavone, M., “CoCo: Online Mixed-Integer Control via Supervised Learning,”IEEE Robotics and Automation Letters, Vol. 7, No. 2, 2022, pp. 1447–1454

    2022

  38. [46]

    PRISM: Recurrent neural networks and presolve methods for fast mixed-integer optimal control,

    Cauligi, A., Chakrabarty, A., Di Cairano, S., and Quirynen, R., “PRISM: Recurrent neural networks and presolve methods for fast mixed-integer optimal control,”Learning for Dynamics & Control, 2022

    2022

  39. [47]

    Successive Convexification for 6-DoF Mars Rocket Powered Landing with Free-Final-Time,

    Szmuk, M., , and Açıkmeşe, B., “Successive Convexification for 6-DoF Mars Rocket Powered Landing with Free-Final-Time,” AIAA Scitech Forum, 2018

    2018

  40. [48]

    State-Dependent Trust Region for Successive Convex Programming for Autonomous Spacecraft,

    Bernardini, N., Baresi, N., and Armellin, R., “State-Dependent Trust Region for Successive Convex Programming for Autonomous Spacecraft,”Astrodynamics, Vol. 8, 2024, pp. 553–575

    2024

  41. [49]

    Convergence of Descent Methods for Semi-Algebraic and Tame Problems: Proximal Algorithms, Forward–Backward Splitting, and Regularized Gauss–Seidel Methods,

    Attouch, H., Bolte, J., and Svaiter, B. F., “Convergence of Descent Methods for Semi-Algebraic and Tame Problems: Proximal Algorithms, Forward–Backward Splitting, and Regularized Gauss–Seidel Methods,”Mathematical Programming, Vol. 137, No. 1-2, 2013, pp. 91–129

    2013

  42. [50]

    SuccessiveConvexificationforTrajectoryOptimization with Continuous-Time Constraint Satisfaction,

    Elango,P.,Luo,D.,Kamath,A.G.,Uzun,S.,Kim,T.,andAçıkmeşe,B.,“SuccessiveConvexificationforTrajectoryOptimization with Continuous-Time Constraint Satisfaction,”ArXiv Preprint, 2024. Available athttps://arxiv.org/abs/2404.16826

    2024 arXiv

  43. [51]

    Random Convex Programs,

    Calafiore, G. C., “Random Convex Programs,”SIAM Journal on Optimization, Vol. 20, No. 6, 2010

    2010

  44. [52]

    Automatic differentiation in PyTorch,

    Paszke, A., Gross, S., Chintala, S., Chanan, G., Yang, E., DeVito, Z., Lin, Z., Desmaison, A., Antiga, L., and Lerer, A., “Automatic differentiation in PyTorch,”Conf. on Neural Information Processing Systems - Autodiff Workshop, 2017

    2017

  45. [53]

    ECOS: An SOCP solver for embedded systems,

    Domahidi, A., Chu, E., and Boyd, S., “ECOS: An SOCP solver for embedded systems,”European Control Conference, 2013

    2013

  46. [54]

    Revisiting Data Augmentation for Rotational Invariance in Convolutional Neural Networks,

    Quiroga, F., Ronchetti, F., Lanzarini, L., and Bariviera, A. F., “Revisiting Data Augmentation for Rotational Invariance in Convolutional Neural Networks,”Int. Conf. on Modelling and Simulation in Management Sciences, 2019. 21

    2019

  47. [55]

    Enhancements on the Convex Programming Based Powered Descent Guidance Algorithm for Mars Landing,

    Açikmeşe, B., Blackmore, L., Scharf, D. P., and Wolf, A., “Enhancements on the Convex Programming Based Powered Descent Guidance Algorithm for Mars Landing,”AIAA/AAS Astrodynamics Specialist Conference, 2008

    2008

  48. [56]

    Interpolation-Enhanced Powered Descent Guidance for Onboard Nominal, Off-Nominal, and Multi-X Scenarios,

    Scharf, D. P., Ploen, S. R., and Açıkmeşe, B., “Interpolation-Enhanced Powered Descent Guidance for Onboard Nominal, Off-Nominal, and Multi-X Scenarios,”AIAA Conf. on Guidance, Navigation and Control, 2015

    2015

  49. [57]

    Multidimensional binary search trees used for associative searching,

    Bentley, J. L., “Multidimensional binary search trees used for associative searching,”Communications of the ACM, Vol. 18, No. 9, 1975, pp. 509–517

    1975

  50. [58]

    On Lines and Planes of Closest Fit to Systems of Points in Space,

    Pearson, K., “On Lines and Planes of Closest Fit to Systems of Points in Space,”Philosophical Magazine, Vol. 2, No. 11, 1901, pp. 559–572

    1901

  51. [59]

    Scikit-learn: Machine Learning in Python,

    Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., and Duchesnay, E., “Scikit-learn: Machine Learning in Python,”Journal of...

    2011

  52. [60]

    The Mahalanobis distance,

    De Maesschalck, R., Jouan-Rimbaud, D., and Massart, D. L., “The Mahalanobis distance,”Chemometrics and Intelligent Laboratory Systems, Vol. 50, No. 1, 2000, pp. 1–18

    2000

  53. [61]

    RAD750™SpaceWire-Enabled Flight Computer for Lunar Reconnaissance Orbiter,

    Berger, R., Dennis, A., Eckhardt, D., Miller, S., Robertson, J., Saridakis, D., Stanley, D., Vancampen, M., and Nguyen, Q., “RAD750™SpaceWire-Enabled Flight Computer for Lunar Reconnaissance Orbiter,”International SpaceWire Conference, 2007

    2007

  54. [62]

    Learning to Actively Reduce Memory Requirements for Robot Control Tasks,

    Booker, M., and Majumdar, A., “Learning to Actively Reduce Memory Requirements for Robot Control Tasks,”Learning for Dynamics & Control, 2021. 22

    2021

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.