Distributed Gradient Descent: Nonconvergence to Saddle Points and the Stable-Manifold Theorem

Brian Swenson; H. Vincent Poor; Ryan Murray; Soummya Kar

REVIEW 2 major objections 4 minor 28 references

Distributed Gradient Descent: Nonconvergence to Saddle Points and the Stable-Manifold Theorem

T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read Continuous-time distributed gradient descent almost always avoids nondegenerate saddle points and converges to local minima, because convergence to each saddle occurs only from a low-dimensional stable manifold.

desk verdict A genuinely new stable-manifold theorem for continuous-time DGD, but the global iff is not proved: the constructed stable set is only a local patch. read the letter →

arxiv 1908.02747 v2 pith:QA3DFBVR submitted 2019-08-07 math.OC cs.MA

Brian Swenson , Ryan Murray , H. Vincent Poor , Soummya Kar This is my paper · ORCID

classification math.OCcs.MA MSC 37D1090C2693A14

keywords distributedgradientdescentsaddlepointsstable-manifoldtheoremnonconvexoptimizationconsensusandinnovationsmulti-agentsystemscontinuous-timedynamics

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

The paper addresses a gap in nonconvex distributed optimization: in centralized gradient descent, the classical stable-manifold theorem guarantees that convergence to saddle points only happens from rare initial conditions, but no such theorem existed for distributed gradient descent. The authors prove that continuous-time distributed gradient descent (DGD), where agents combine local gradients with a consensus term, has the same property. For any nondegenerate saddle point, the set of initial conditions that lead to it is a lower-dimensional manifold, so almost every initialization avoids saddles. Combined with a convergence-to-critical-points result, this implies DGD typically converges to local minima in smooth nonconvex problems.

What carries the argument

The machinery is a non-autonomous stable-manifold construction for the penalized dynamics $\dot{x} = -\nabla h(x) - \beta_t Q x$, with $Q = L \otimes I_d$ in the DGD case. The proof recenters the dynamics at the moving saddle $g(\beta_t)$ of the penalized objective, diagonalizes the time-varying Hessian by a unitary $U(t)$, splits coordinates into stable and unstable subspaces, and defines the stable manifold through an integral equation solved by contraction on a Banach space. The stable and unstable evolution operators $V^s$ and $V^u$ carry the forward/backward estimates that control the contraction, and the manifold is realized as the graph of a continuous function over the stable coordinates, giving dimension $Nd - p$.

What would settle it

For a smooth coercive objective satisfying the assumptions, compute the exact stable set near a nondegenerate saddle (analytically in a quadratic example or numerically in a general one) and check whether every trajectory that converges to the saddle starts on the proof's locally constructed manifold; a trajectory that converges while leaving the local stable-coordinate ball would falsify the stated if-and-only-if characterization.

Watch

Extended reading notes

Core claim

The central discovery is Theorem 2: under the paper's assumptions, if $x^*$ is a nondegenerate saddle point of the summed objective with $p$ negative Hessian eigenvalues, then for all sufficiently large initial times $t_0$ there exists a manifold $S$ of dimension $Nd - p$ such that a solution to the DGD dynamics converges to $x^*$ if and only if the entire network state at $t_0$ lies on $S$. Because $p$ is at least 1, this stable manifold has codimension at least one, so the union of such manifolds over all saddle points has measure zero. The paper proves this within a more general subspace-constrained optimization framework (Theorem 6), and Theorem 1 establishes that agents reach consensus and converge to the set of critical points. Together, the results say that for coercive objectives with nondegenerate saddles, DGD starting outside this countable union of low-dimensional sets converges to local minima rather than saddle points.

Load-bearing premise

The load-bearing premise is that the low-dimensional stable set built locally over small stable coordinates is actually the complete set of initial conditions that converge to the saddle, even though the proof only establishes this within a small ball in those coordinates.

Editorial extensions

If this is right

For almost every initial condition in Lebesgue measure, continuous-time DGD does not converge to any nondegenerate saddle point, under the paper's assumptions.
If the objective is coercive and all saddle points are nondegenerate, DGD converges to local minima for almost every initialization, not merely to the critical point set.
The result holds for any connected undirected communication graph and for consensus and innovation weight sequences with $0 \le \tau_\beta < \tau_\alpha \le 1$, so no centralized coordination is needed to escape saddle points.
The stable manifold dimension $Nd-p$ implies the set of bad initial conditions has codimension at least one: in a network with $N$ agents each holding a $d$-dimensional state, generic assignments of initial states avoid any given saddle.

Reading between the lines

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

The measure-zero conclusion suggests that discrete-time DGD with decaying step sizes may inherit similar saddle avoidance, but the paper only treats continuous time; verifying this would require showing the discrete map tracks the continuous flow near the saddle closely enough.
Even if the local stable set constructed in the proof does not capture every trajectory that eventually reaches the saddle, the almost-sure avoidance claim could still hold; the load-bearing statistical fact is only that the converging set has measure zero, not the exact if-and-only-if identification.
The dimension count $Nd-p$ means that adding more agents or dimensions makes saddle manifolds relatively rarer, so random initialization in larger networks should be even safer against saddle convergence.
One could test the proof's mechanism on a quadratic objective, where the true stable set is an explicit subspace; matching that subspace with the contraction-based construction would directly validate the local stable-manifold argument.

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.

Desk Editor's Note

read the letter

Two things to know: the paper proves a stable-manifold theorem for continuous-time distributed gradient descent (DGD) that is genuinely new, and the theorem as stated is not supported by the proof. The gap is not a demonstrated counterexample but it is load-bearing.

The positive side: the formulations are clean. The reduction of (2) to the subspace-constrained flow (10) is useful, and the Lyapunov-Perron construction is standard and mostly executed carefully. Theorem 1's convergence to critical points is a solid result. Theorem 2 is the first stable-manifold result for DGD: saddles have stable manifolds of codimension p, so typical initializations should avoid them. If correct, it fills a real gap in the distributed optimization literature.

Soft spots:

1. The global 'if and only if' in Theorem 2 is not proved. Section IV builds S as a graph over |z_s^0| < r/3. Lemma 8 only says that if a trajectory converges to 0 and its stable coordinate at t0 lies below r/3, then it is on S. Nothing forces a converging trajectory to enter that ball at t0. In the linear example h(x,y) = x^2 - y^2 with Q=0, the true stable manifold is the line y=0, unbounded in x, while the construction gives only the segment |x| < r/3. So the converse direction of Theorem 2 is unsupported.

2. In Theorem 1, Sard's theorem is invoked for a C^2 function on R^d. Sard requires differentiability order at least d for scalar-valued maps on R^d (C^1 when d=1). For d≥3, C^2 does not suffice. This is a standard oversight; the convergence-to-critical-points statement can likely be obtained another way, but the proof as written has a hole.

3. The recentering step asserts a global C^1 branch g(β) of critical points of the penalized function and a differentiable diagonalizing frame U(t). That needs more than the implicit function theorem gives you; a local branch near 0 is fine, and smooth eigenvalue/eigenvector selection is possible for large t with separated spectra, but the paper does not justify the global statement.

None of this makes me think the main claim is false. It makes me think the paper is a solid idea in need of an honest major revision: either prove the global stable manifold, or restate Theorem 2 as a local statement and show the typicality conclusion by a different argument (e.g., countable union of local graphs).

I'd send it to review. The reviewers who know nonautonomous stable-manifold theory will be able to tell whether the gap can be closed. For reading group: maybe, if the group works on distributed nonconvex optimization.

Referee Report

2 major / 4 minor

Summary. The paper studies continuous-time distributed gradient descent (2) for minimizing a sum of smooth, nonconvex functions over a connected network. It first proves (Theorem 1) that all trajectories achieve consensus and converge to the set of critical points of the global objective. The main result (Theorem 2) is a stable-manifold theorem asserting that, near a nondegenerate saddle point x*, the set of initial conditions from which the dynamics converge to x* is a manifold of codimension p (the number of negative Hessian eigenvalues), so that convergence to saddle points is non-generic. The proof is obtained by embedding the dynamics into a more general subspace-constrained gradient flow (10) and proving a stable-manifold theorem (Theorem 6) via a Lyapunov-Perron integral-equation argument.

Significance. The intended conclusion - that distributed gradient descent almost always avoids saddle points and converges to local minima - is of clear interest for nonconvex distributed optimization, and the extension of stable-manifold ideas to non-autonomous, consensus-driven dynamics is a valuable direction. The paper is self-contained and does not rely on circular parameter fitting: the proof of Theorem 2 is built from explicit contraction estimates and classical machinery. If the missing globalization step described below can be supplied, the result would be a solid contribution. As it stands, the central theorem is not fully proven, although the underlying claim is plausible and the gap appears repairable.

major comments (2)

[Section IV.B, items 3-4 and Lemma 8; Theorem 6; Theorem 2] The constructed manifold S is a local graph over B_{r/3}(0) in stable coordinates, and Lemma 8 only proves that a solution converging to 0 with |z_s(t0)| < r/3 starts in S. The proof does not show that every solution converging to 0 must have |z_s(t0)| < r/3, nor does it globalize the local graph by backward flow. Consequently, the if-and-only-if in Theorem 6 (and hence Theorem 2) is unsupported: initial conditions with large stable coordinate and zero unstable coordinate converge to 0 without belonging to the constructed S. A concrete illustration is the linear model h(x,y)=x^2-y^2 with Q=0, where the stable manifold is the entire line y=0 but the construction yields only the segment |x|<r/3. The typicality conclusion requires the full set of converging initial conditions, not merely a local patch, to have measure zero, so a globalization argument is needed.
[Section V and the paragraph after Theorem 2] The typicality conclusion is drawn from the union over all saddle points of the manifolds S_{x*}. Even if the local graph construction were accepted, the proof does not establish that S_{x*} contains all initial conditions that converge to x*, so the claimed codimension bound on the full stable set is not available. A measure-zero conclusion requires a bound on the full stable set, not just on a local patch; the paper should either prove the global statement or explicitly restrict the theorem to a local stable patch and proportionally adjust the conclusion.

minor comments (4)

[Section IV.B.1] The implicit function theorem is invoked to obtain a branch g(beta) of critical points of h(x)+beta x^T Q x for each beta >= 0 with g(beta) -> 0; for a saddle of h|C that is not a critical point of h, such a branch generally exists only for sufficiently large beta, so the statement should be adjusted.
[Lemma 8, Eq. (19)] The symbol V(t,tau) in the variation-of-constants formula is undefined; the formula should explicitly use V^s(t,tau) and V^u(t,tau) in the respective integral terms.
[Section IV.B, item 4] The notation S := {(t,z_s^0, psi(t,z_s^0)), t >= T, z_s^0 in R^k cap B_{r/3}(0)} is imprecise; it should say z_s^0 in B_{r/3}(0) subset R^k, and the dependence of the manifold on the initial time t0 should be made explicit in Theorem 6.
[Theorem 6] Theorem 6 is stated for arbitrary C^1 weights beta_t with beta_t -> infinity, but the proof's assertion that c(t) in (18) can be made uniformly small for large t0 requires a decay condition on |g'(beta_t) dot beta_t| that is not implied by the stated hypotheses. The DGD potentials satisfy such a condition, but the theorem should either state it explicitly or be restricted accordingly.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the stable-manifold theorem is proved self-containedly via a contraction-mapping construction, with prior self-citations only contextual.

full rationale

The paper's derivation chain is self-contained. Theorem 2 is reduced to the generalized dynamics (10), recentered and diagonalized in Section IV.B, and the stable set S is constructed as the graph of a fixed point of the contraction-mapping integral equation (17), with Lemmas 7 and 8 supplying the contraction and the inclusion of stable initializations. No free parameter is fitted to data, and no target result is assumed under another name. The authors' own earlier works (e.g., refs [5], [16], [19]) appear only in motivational or contextual roles and are not used to justify a load-bearing theorem. The local-patch issue noted by the reader — S is constructed only over |z_s^0| < r/3, while Theorem 2 asserts a global 'if and only if' — is a potential proof gap concerning globalization, not a circularity: nothing in the proof defines S in terms of the set of converging initial conditions, and the asserted equivalence is intended to follow from the contraction argument rather than from a fitted or assumed input. Therefore no step in the derivation reduces to its own inputs, and the appropriate circularity score is 0.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

No free parameters are fitted to data or chosen ad hoc; the potentials alpha_t and beta_t are fixed by Assumption 5 with a stated relation. The mathematical inputs are standard ODE and invariant-manifold tools plus the paper's domain assumptions. The two questionable items are the false Sard-style assertion used in Theorem 1 and the implicit globalization of the local stable manifold in Theorem 2.

assumptions (7)

standard math Existence and uniqueness of solutions to (2) under Assumption 3
Invoked in Section II; standard Picard-Lindelof theory for ODEs with Lipschitz vector fields.
standard math Asymptotic pseudotrajectory theorem of Benaim, Hofbauer, and Sorin (Theorem 4)
Used to prove Theorem 1; requires a Lyapunov function V with V(Lambda) having empty interior.
ad hoc to paper The set of critical values f(Lambda) of a C2 function f:R^d to R has empty interior
Assumed in the proof of Theorem 1 via Sard's theorem; Sard's theorem requires higher regularity when d is at least 3, so this can fail and is not a standard consequence.
standard math Implicit function theorem supplies a C1 branch g(beta) of critical points of h(x)+beta x^T Q x for all beta in [0, infinity)
Used in Section IV.B.1 to recenter the dynamics; the theorem is standard, but the global-in-time formulation is asserted rather than fully demonstrated.
domain assumption Domain assumptions: connected undirected graph, C2 coercive local objectives with Lipschitz gradients, decay exponents satisfying 0 <= tau_beta < tau_alpha <= 1
Assumptions 1 to 5; the whole result is conditional on these modeling choices.
domain assumption Nondegenerate saddle assumption (Hessian nonsingular) for every saddle point
Assumption 6; needed for the spectral gap and for the implicit function theorem in Theorem 6.
ad hoc to paper The local stable manifold constructed over B_{r/3}(0) can be globalized to the full set of converging initial conditions
Implicitly assumed by the iff statement in Theorem 2; Lemma 8 only handles |a_s| < r/3, and no backward-flow globalization is supplied.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Distributed Gradient Descent: Nonconvergence to Saddle Points and the Stable-Manifold Theorem." pith.science (2026). https://pith.science/paper/QA3DFBVR

@misc{pith2026190802747,
  author       = {Pith},
  title        = {Pith review of: Distributed Gradient Descent: Nonconvergence to Saddle Points and the Stable-Manifold Theorem},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QA3DFBVR}},
  note         = {Machine review of arXiv:1908.02747}
}

read the original abstract

The paper studies a distributed gradient descent (DGD) process and considers the problem of showing that in nonconvex optimization problems, DGD typically converges to local minima rather than saddle points. The paper considers unconstrained minimization of a smooth objective function. In centralized settings, the problem of demonstrating nonconvergence to saddle points of gradient descent (and variants) is typically handled by way of the stable-manifold theorem from classical dynamical systems theory. However, the classical stable-manifold theorem is not applicable in distributed settings. The paper develops an appropriate stable-manifold theorem for DGD showing that convergence to saddle points may only occur from a low-dimensional stable manifold. Under appropriate assumptions (e.g., coercivity), this result implies that DGD typically converges to local minima and not to saddle points.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 23 canonical work pages

[1]

Distributed optimization in sen sor networks,

M. Rabbat and R. Nowak, “Distributed optimization in sen sor networks,” in Proceedings of the 3rd International Symposium on Informat ion Processing in Sensor Networks , 2004, pp. 20–27

work page 2004
[2]

Diffusion adaptation strategie s for distributed optimization and learning over networks,

J. Chen and A. H. Sayed, “Diffusion adaptation strategie s for distributed optimization and learning over networks,” IEEE Transactions on Signal Processing, vol. 60, no. 8, pp. 4289–4305, 2012

work page 2012
[3]

An overview of recent pr ogress in the study of distributed multi-agent coordination,

Y . Cao, W. Y u, W. Ren, and G. Chen, “An overview of recent pr ogress in the study of distributed multi-agent coordination,” IEEE Transactions on Industrial Informatics , vol. 9, no. 1, pp. 427–438, 2012

work page 2012
[4]

D istributed optimization and statistical learning via the alternating direction method of multipliers,

S. Boyd, N. Parikh, E. Chu, B. Peleato, and J. Eckstein, “D istributed optimization and statistical learning via the alternating direction method of multipliers,” F oundations and Trends R© in Machine learning , vol. 3, no. 1, pp. 1–122, 2011

work page 2011
[5]

Clustering with distributed data ,

S. Kar and B. Swenson, “Clustering with distributed data ,” 2019, submitted for publication. Online: https://arxiv.org/ab s/1901.00214

arXiv 2019
[6]

Distributed subgradient met hods for multi- agent optimization,

A. Nedi´ c and A. Ozdaglar, “Distributed subgradient met hods for multi- agent optimization,” IEEE Transactions on Automatic Control , vol. 54, no. 1, p. 48, 2009

work page 2009
[7]

A distributed quasi-n ewton algo- rithm for empirical risk minimization with nonsmooth regul arization,

C. Lee, C. H. Lim, and S. J. Wright, “A distributed quasi-n ewton algo- rithm for empirical risk minimization with nonsmooth regul arization,” in Proceedings of the 24th ACM SIGKDD International Conferenc e on Knowledge Discovery & Data Mining , 2018, pp. 1646–1655

work page 2018
[8]

Distributed nonconvex opt imization over networks,

P . Di Lorenzo and G. Scutari, “Distributed nonconvex opt imization over networks,” in Proceedings of the 6th International W orkshop on Com- putational Advances in Multi-Sensor Adaptive Processing ( CAMSAP), 2015, pp. 229–232

work page 2015

Show all 28 references

[9]

Distributed nonconv ex multiagent optimization over time-varying networks,

Y . Sun, G. Scutari, and D. Palomar, “Distributed nonconv ex multiagent optimization over time-varying networks,” in 2016 50th Asilomar Con- ference on Signals, Systems and Computers . IEEE, 2016, pp. 788–794

2016
[10]

Distributed non-con vex optimization of multi-agent systems using boosting functions to escape l ocal optima,

S. Welikala and C. G. Cassandras, “Distributed non-con vex optimization of multi-agent systems using boosting functions to escape l ocal optima,” arXiv preprint arXiv:1903.04133 , 2019

1903 arXiv
[11]

Convergence of a multi-a gent projected stochastic gradient algorithm for non-convex optimizatio n,

P . Bianchi and J. Jakubowicz, “Convergence of a multi-a gent projected stochastic gradient algorithm for non-convex optimizatio n,” IEEE Trans- actions on Automatic Control , vol. 58, no. 2, pp. 391–405, 2012

2012
[12]

Next: In-network nonconv ex optimiza- tion,

P . Di Lorenzo and G. Scutari, “Next: In-network nonconv ex optimiza- tion,” IEEE Transactions on Signal and Information Processing ove r Networks, vol. 2, no. 2, pp. 120–136, 2016

2016
[13]

Identifying and attacking the saddle point prob lem in high- dimensional non-convex optimization,

Y . N. Dauphin, R. Pascanu, C. Gulcehre, K. Cho, S. Gangul i, and Y . Bengio, “Identifying and attacking the saddle point prob lem in high- dimensional non-convex optimization,” in Advances in neural informa- tion processing systems , 2014, pp. 2933–2941

2014
[14]

Gra dient descent only converges to minimizers,

J. D. Lee, M. Simchowitz, M. I. Jordan, and B. Recht, “Gra dient descent only converges to minimizers,” in Conference on learning theory , 2016, pp. 1246–1257

2016
[15]

First-order methods almost always avoid strict s addle points,

J. D. Lee, I. Panageas, G. Piliouras, M. Simchowitz, M. I . Jordan, and B. Recht, “First-order methods almost always avoid strict s addle points,” Mathematical Programming, pp. 1–27, 2019

2019
[16]

Revisiting normaliz ed gradient descent: Fast evasion of saddle points,

R. Murray, B. Swenson, and S. Kar, “Revisiting normaliz ed gradient descent: Fast evasion of saddle points,” IEEE Transactions on Automatic Control, vol. PP , pp. 1–1, 2019

2019
[17]

Shub, Global stability of dynamical systems

M. Shub, Global stability of dynamical systems . Springer Science & Business Media, 2013

2013
[18]

E. A. Coddington and N. Levinson, Theory of ordinary differential equations. Tata McGraw-Hill Education, 1955

1955
[19]

Anneal ing for distributed global optimization,

B. Swenson, S. Kar, H. V . Poor, and J. M. F. Moura, “Anneal ing for distributed global optimization,” 2019, to appear in Proce edings of IEEE Conference on Decision and Control

2019
[20]

Non-convex distributed opt imization,

T. Tatarenko and B. Touri, “Non-convex distributed opt imization,” IEEE Transactions on Automatic Control, vol. 62, no. 8, pp. 3744–3757, 2017

2017
[21]

Second- order guarantees of distributed gradient algorithms,

A. Daneshmand, G. Scutari, and V . Kungurtsev, “Second- order guarantees of distributed gradient algorithms,” arXiv preprint arXiv:1809.08694, 2018

2018 arXiv
[22]

Distributed parame ter estimation in sensor networks: Nonlinear observation models and imper fect com- munication,

S. Kar, J. M. Moura, and K. Ramanan, “Distributed parame ter estimation in sensor networks: Nonlinear observation models and imper fect com- munication,” IEEE Transactions on Information Theory , vol. 58, no. 6, pp. 3575–3605, 2012

2012
[23]

Constrained consensus and optimization in multi-agent networks,

A. Nedi´ c, A. Ozdaglar, and P . A. Parrilo, “Constrained consensus and optimization in multi-agent networks,” IEEE Transactions on Automatic Control, vol. 55, no. 4, pp. 922–938, 2010

2010
[24]

Distributed optimization over time-varying directed graphs,

A. Nedi´ c and A. Olshevsky, “Distributed optimization over time-varying directed graphs,” IEEE Transactions on Automatic Control , vol. 60, no. 3, pp. 601–615, 2014

2014
[25]

Gradient primal- dual al- gorithm converges to second-order stationary solutions fo r nonconvex distributed optimization,

M. Hong, J. D. Lee, and M. Razaviyayn, “Gradient primal- dual al- gorithm converges to second-order stationary solutions fo r nonconvex distributed optimization,” in Proceedings of the 35th International Conference on Machine Learning, PMLR 80 , 2018

2018
[26]

F. R. Chung and F. C. Graham, Spectral graph theory . American Mathematical Soc., 1997, no. 92

1997
[27]

Stochastic appr oximations and differential inclusions,

M. Bena¨ ım, J. Hofbauer, and S. Sorin, “Stochastic appr oximations and differential inclusions,” SIAM Journal on Control and Optimization , vol. 44, no. 1, pp. 328–348, 2005

2005
[28]

Lakshmikantham and S

V . Lakshmikantham and S. Leela, Differential and Integral Inequalities: Theory and Applications: V olume I: Ordinary Differential E quations. Academic press, 1969

1969

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Distributed optimization in sen sor networks,

M. Rabbat and R. Nowak, “Distributed optimization in sen sor networks,” in Proceedings of the 3rd International Symposium on Informat ion Processing in Sensor Networks , 2004, pp. 20–27

work page 2004

[2] [2]

Diffusion adaptation strategie s for distributed optimization and learning over networks,

J. Chen and A. H. Sayed, “Diffusion adaptation strategie s for distributed optimization and learning over networks,” IEEE Transactions on Signal Processing, vol. 60, no. 8, pp. 4289–4305, 2012

work page 2012

[3] [3]

An overview of recent pr ogress in the study of distributed multi-agent coordination,

Y . Cao, W. Y u, W. Ren, and G. Chen, “An overview of recent pr ogress in the study of distributed multi-agent coordination,” IEEE Transactions on Industrial Informatics , vol. 9, no. 1, pp. 427–438, 2012

work page 2012

[4] [4]

D istributed optimization and statistical learning via the alternating direction method of multipliers,

S. Boyd, N. Parikh, E. Chu, B. Peleato, and J. Eckstein, “D istributed optimization and statistical learning via the alternating direction method of multipliers,” F oundations and Trends R© in Machine learning , vol. 3, no. 1, pp. 1–122, 2011

work page 2011

[5] [5]

Clustering with distributed data ,

S. Kar and B. Swenson, “Clustering with distributed data ,” 2019, submitted for publication. Online: https://arxiv.org/ab s/1901.00214

arXiv 2019

[6] [6]

Distributed subgradient met hods for multi- agent optimization,

A. Nedi´ c and A. Ozdaglar, “Distributed subgradient met hods for multi- agent optimization,” IEEE Transactions on Automatic Control , vol. 54, no. 1, p. 48, 2009

work page 2009

[7] [7]

A distributed quasi-n ewton algo- rithm for empirical risk minimization with nonsmooth regul arization,

C. Lee, C. H. Lim, and S. J. Wright, “A distributed quasi-n ewton algo- rithm for empirical risk minimization with nonsmooth regul arization,” in Proceedings of the 24th ACM SIGKDD International Conferenc e on Knowledge Discovery & Data Mining , 2018, pp. 1646–1655

work page 2018

[8] [8]

Distributed nonconvex opt imization over networks,

P . Di Lorenzo and G. Scutari, “Distributed nonconvex opt imization over networks,” in Proceedings of the 6th International W orkshop on Com- putational Advances in Multi-Sensor Adaptive Processing ( CAMSAP), 2015, pp. 229–232

work page 2015