Pith. sign in

REVIEW 2 major objections 4 minor 1 cited by

Decentralized projected Riemannian stochastic recursive momentum method for nonconvex optimization

T0 review · 2 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read This paper proposes a decentralized single-loop stochastic recursive momentum method for smooth nonconvex optimization over compact submanifolds and proves an oracle complexity of $O(\epsilon^{-3/2})$ per node in the online setting.

desk verdict DPRSRM is a sensible combination of known ingredients, but the advertised O(ε^{-3/2}) oracle complexity is unsupported: a dropped term from Lemma 4.5 leaves a K^{-1/3} error, and Assumption 3.4 is too weak for several lemmas. read the letter →

arxiv 2412.02382 v2 pith:FRO7E45K submitted 2024-12-03 math.OC

classification math.OC MSC 90C0690C2290C2690C56
keywords decentralizedoptimizationRiemannianmanifoldstochasticrecursivemomentumvariancereductiongradienttrackingoraclecomplexitynonconvexconsensus
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 proposes DPRSRM, a decentralized algorithm for minimizing a sum of smooth nonconvex functions over a compact submanifold in the online setting, where each node observes only stochastic gradients of streaming data. The central claim is that DPRSRM reaches an $\epsilon$-stationary point with $O(\epsilon^{-3/2})$ stochastic gradient oracle calls per node, improving on the $O(\epsilon^{-2})$ complexity of earlier decentralized Riemannian stochastic methods. The method is single-loop, uses $O(1)$ stochastic gradient evaluations per iteration per node, and requires one consensus communication step per round, making it suitable for real-time streaming data. The analysis handles the nonconvex manifold constraint by projecting iterates onto the manifold and exploiting proximal smoothness, the property that the nearest-point projection is unique and Lipschitz in a tubular neighborhood.

What carries the argument

The central object is the DPRSRM update, which combines three components. The hybrid estimator $q_{i,k} = \operatorname{grad} f_i(x_{i,k}, \xi_{i,k}) + (1-\tau)(d_{i,k-1} - \operatorname{grad} f_i(x_{i,k-1}, \xi_{i,k}))$ is a recursive momentum variance-reduction step; it is clipped to norm $B$ to form $d_{i,k}$. The gradient tracker $s_{i,k} = \sum_j W_{ij} s_{j,k-1} + d_{i,k} - d_{i,k-1}$ propagates information about the global gradient through the network. The iterate is updated by tangent-space projection of $s$ onto the tangent bundle and then manifold projection: $x_{i,k+1} = P_{\mathcal M}(\sum_j W_{ij}x_{j,k} - \alpha v_{i,k})$. Proximal smoothness of the compact submanifold supplies the Lipschitz property of the projection operator $P_{\mathcal M}$ that controls the consensus error.

What would settle it

Run DPRSRM on a compact submanifold with a smooth nonconvex objective and a stochastic gradient oracle that is unbiased and has bounded variance but unbounded support, such as Gaussian noise, so that the printed Assumption 3.4 holds but the almost-sure bound fails; if the consensus error or the gradient-norm bound in Corollary 3.8 is violated, or if the clipping constant must grow with $1/\epsilon$, then the claimed $O(\epsilon^{-3/2})$ rate depends on the unstated stronger assumption.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that combining a hybrid stochastic recursive momentum estimator with gradient tracking and projection onto a compact submanifold yields an oracle complexity of $O(\epsilon^{-3/2})$ for decentralized online nonconvex manifold optimization. Corollary 3.8 states that with step size $\alpha = K^{-1/3}$ and momentum parameter $\tau = K^{-2/3}$, after $K = O(\max\{(C+\Gamma_1)^{3/2}\epsilon^{-3/2},\ \Gamma_2\epsilon^{-1},\ (6\rho_2\nu^2)^{3/4}\epsilon^{-3/4}\})$ iterations the iterates are $\epsilon$-stationary; the dominant term is the $\epsilon^{-3/2}$ oracle complexity. The result applies to general compact submanifolds, not just the Stiefel manifold, and needs only one consensus step per iteration, whereas earlier projection and retraction methods required $O(\log(1/\sigma_2))$ consensus rounds. The paper further shows numerically on decentralized principal component analysis and low-rank matrix completion that DPRSRM beats the stochastic decentralized Riemannian gradient descent and projected gradient baselines considered.

Load-bearing premise

The load-bearing premise is that every stochastic gradient is almost surely bounded by a fixed constant and that the average squared difference of two stochastic gradients grows at most linearly with the distance between their inputs, while the assumptions stated in the paper only guarantee bounded variance and a first-moment Lipschitz condition; if the stronger premise is not added, the consensus and optimality bounds do not follow.

Editorial extensions

If this is right

  • Each node reaches an $\epsilon$-stationary point with $O(\epsilon^{-3/2})$ stochastic gradient oracle calls, up to the smaller $\epsilon^{-1}$ and $\epsilon^{-3/4}$ terms in Corollary 3.8.
  • Only one consensus communication round is needed per iteration, removing the $O(\log(1/\sigma_2))$ consensus-factor inflation of earlier projection and retraction methods on the Stiefel manifold.
  • The method is single-loop and online: no restarting with a large batch is required, and each node uses only $O(1)$ gradient evaluations per iteration.
  • The convergence guarantee covers general compact submanifolds, using projection $P_{\mathcal M}$, not just Stiefel and orthogonality constraints.
  • Numerical tests on decentralized PCA and low-rank matrix completion show faster decrease of consensus error, objective gap, and gradient norm than the DRSGD and DRPGD baselines tested.

Reading between the lines

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

  • The $O(\epsilon^{-3/2})$ rate matches the known single-machine recursive-momentum rate, which suggests that the decentralized manifold constraint and one-round consensus need not add oracle overhead in this parameter regime.
  • Because the proof uses only projection Lipschitzness and a quadratic upper bound on the local objectives, the same estimator-and-consensus scheme should transfer to other proximally smooth constraint sets, such as sphere products or rank-constrained sets.
  • The clipping constant $B$ is a free user parameter; a data-dependent or adaptive choice of $B$ might preserve the rate while allowing larger practical steps, though this is not analyzed in the paper.
  • Since the method uses one communication round per iteration, it is a natural candidate for combining with gradient compression or quantization; whether the $O(\epsilon^{-3/2})$ oracle complexity survives communication compression is left open.
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

2 major / 4 minor

Summary. The paper proposes DPRSRM, a decentralized projected Riemannian stochastic recursive momentum method for minimizing the average of smooth nonconvex functions over a compact submanifold in an online setting. The algorithm combines a clipped hybrid stochastic gradient estimator (3.6)-(3.8), gradient tracking (3.9), tangent-space projection, and a single consensus step. The main theoretical claims are Theorem 3.6 (consensus error), Theorem 3.7 (optimality error), and Corollary 3.8, which asserts that with alpha = K^{-1/3} and tau = K^{-2/3} the method reaches an epsilon-stationary point (Definition 2.1) with oracle complexity O(epsilon^{-3/2}) per node, improving on existing O(epsilon^{-2}) rates. Numerical experiments on decentralized PCA and low-rank matrix completion are reported.

Significance. If the O(epsilon^{-3/2}) rate were valid, this would be a meaningful advance for online decentralized manifold optimization: the method is single-loop, uses O(1) stochastic gradients per iteration per node, has no large-batch restart, and uses one consensus step for general compact submanifolds. The paper is clearly structured and the appendix contains a detailed proof attempt. However, the central rate is not established by the submitted proof. The proof of Theorem 3.7 drops a term of order nu^2/(tau K), which with the claimed parameter choices is K^{-1/3}, and the proof relies on bounded stochastic gradients and a second-moment Lipschitz property that are not present in Assumption 3.4. Because the main contribution is precisely the improved oracle complexity, these issues are load-bearing.

major comments (2)
  1. [3.3, Theorem 3.7 and Corollary 3.8; proof in Section 4.2 and Lemma 4.5] Substituting Lemma 4.5 into the inequality (4.7) and dividing by alpha K / 4 yields an additional term 6(1+rho2 tau^2) nu^2 / (tau K). The displayed Theorem 3.7 instead contains 12/n (1+rho2 tau^2)(nu^2 tau + rho3 n alpha^4/tau), which replaces nu^2/(tau K) with nu^2 tau and introduces an unexplained factor 1/n. With the Corollary 3.8 choices alpha = K^{-1/3} and tau = K^{-2/3}, the omitted term is 6(1+rho2 tau^2) nu^2 K^{-1/3}, not O(K^{-2/3}). Therefore the bound in (4.8), the O(epsilon^{-3/2}) oracle complexity in (3.18), and Corollary 3.8 are not supported by the supplied proof. Making the omitted term below epsilon would require K = O(epsilon^{-3}), not K = O(epsilon^{-3/2}). A repair would need a time-varying tau or an initial large batch, neither of which is present in Algorithm 3.1.
  2. [3.1, Assumption 3.4 and Lemmas 4.1, 4.3, 4.5] The proof requires stronger stochastic assumptions than Assumption 3.4 provides. Lemma 4.1 uses ||d_0|| <= sqrt(n) L for the unclipped initial stochastic gradient; Lemma 4.3 uses deterministic bounds such as ||grad f(bar x_k)|| <= L; and Lemma 4.5, specifically equation (B.24), requires a second-moment Lipschitz condition E||grad f(x_k, xi_k) - grad f(x_{k-1}, xi_k)||^2 <= L^2 ||x_k - x_{k-1}||^2. Assumption 3.4 only contains the unbiasedness condition (3.3), the variance bound (3.4), and the first-moment condition (3.5). Moreover, (3.5) is labeled 'mean-squared' but is written with first moments. As written, the consensus and optimality bounds are unsupported. The proof also uses ||d_k|| <= sqrt(n) L for the clipped estimator, whereas Algorithm 3.1 clips at B >= L, so the bound should involve B unless B = L is imposed.
minor comments (4)
  1. [3.3, Corollary 3.8] The stated condition K >= max{4L, sqrt(n)D/delta, R sqrt(n)D / ((R(1-sigma2)-2delta)delta)} does not imply the step-size condition (3.13) when alpha = K^{-1/3}; the correct lower bounds would be of the form K >= (4L)^3, K >= (sqrt(n)D/delta)^3, and K >= (R sqrt(n)D / ((R(1-sigma2)-2delta)delta))^3.
  2. [3.1, Assumption 3.4] If the intended condition in (3.5) is indeed a mean-squared Lipschitz property, it should be written as E||grad f_i(x,xi) - grad f_i(y,xi)||^2 <= Lbar^2 ||x-y||^2 (or with squared norms on both sides), and this strengthening should appear in the assumptions used by Theorems 3.7 and 3.8.
  3. [Title and throughout] There are typographical errors, including the title 'OPTIMIZA TION' and 'Lispchitz' in the proof of Theorem 3.6, and the figure legends in Figures 1-3 render the step size as '=0.1' without displaying the symbol alpha.
  4. [6, Conclusions and Limitations] The section is titled 'Conclusions and Limitations' but contains no limitations statement; the paper should either add a limitations paragraph or rename the section.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the O(epsilon^{-3/2}) rate is derived from explicit assumptions and parameter choices; self-citations to the authors' earlier work are technical geometry/smoothness lemmas that do not presuppose the target oracle complexity.

full rationale

The claimed O(epsilon^{-3/2}) oracle complexity in Corollary 3.8 is obtained by substituting alpha = K^{-1/3} and tau = K^{-2/3} into Theorem 3.7 and then inverting the resulting epsilon-bounds (K1 proportional to epsilon^{-3/2}, K2 proportional to epsilon^{-1}, K3 proportional to epsilon^{-3/4}). No displayed equation assumes the conclusion: Lemmas 4.3-4.5 and Theorem 3.6 are derived from Assumptions 3.1-3.4 (unbiased stochastic gradients, bounded variance, mean-squared Lipschitz stochastic gradients, mixing matrix assumptions) through standard algebraic inequalities (Lemmas A.1 and A.2). The cited results from the authors' earlier work, e.g. [12, Lemma 2] and Lemmas A.3-A.4, are parameter-free geometry and smoothness lemmas about projections onto compact submanifolds and Riemannian gradients; their stated assumptions do not include the target O(epsilon^{-3/2}) rate, so citing them is not load-bearing circularity in the sense of presupposing the main result. The numerical experiments compare against external baselines and do not fit any parameter to the claimed rate. The apparent algebraic inconsistency in Theorem 3.7 (the displayed bound omits the nu^2/(tau K) term carried from Lemma 4.5, which with alpha = K^{-1/3} and tau = K^{-2/3} contributes K^{-1/3}) is a correctness concern about whether the stated rate is actually proven, not a circularity: the derivation chain does not reduce the conclusion to its own inputs.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

No new physical or mathematical entities are introduced. The algorithm is a combination of known components, and the ledger is dominated by standard domain assumptions for decentralized manifold optimization plus one unstated strengthening of the stochastic-gradient model.

free parameters (4)
  • Step size α = K^{-1/3}
    Chosen in Corollary 3.8 to balance the error terms; not fitted to data but a user-selected algorithm parameter.
  • Momentum parameter τ = K^{-2/3}
    Chosen in Corollary 3.8 to balance variance reduction and bias; user-selected.
  • Clipping constant B = B ≥ L, e.g., 10^8 in experiments
    User-defined clipping threshold; the proof requires B ≥ L, and experiments use a large value.
  • Experimental step-size coefficient β̂ = Grid-searched per algorithm and dataset
    Used only in the numerical experiments in Section 5, not in the theoretical rate; it is tuned for best performance.
assumptions (6)
  • domain assumption M is an R-proximally smooth compact submanifold of Euclidean space.
    Assumption 3.1 supplies the unique projection and the Lipschitz property (2.1) used throughout the consensus analysis.
  • domain assumption Each local function f_i is gradient Lipschitz on conv(M) and f has a finite optimal value f* over M^n.
    Provides the smoothness constants L_g and L_G used in the quadratic upper bound (3.1) and in the descent and tracking inequalities.
  • domain assumption The communication graph is connected and W is doubly stochastic with second singular value σ2 in [0,1).
    Controls consensus contraction and appears in the definitions of D, C, and the step-size restrictions in Theorem 3.6.
  • domain assumption Stochastic gradients are unbiased with bounded variance, as in (3.3)-(3.4).
    This is the stated oracle model, but it is insufficient for the proof as written.
  • ad hoc to paper The proof also needs stochastic gradients to be almost surely bounded and to satisfy a second-moment Lipschitz condition.
    Lemma 4.1 uses ||d_0|| ≤ √n L, and Lemma 4.5 uses E||g(x,ξ)-g(y,ξ)||² ≤ L²||x-y||², which are not stated in Assumption 3.4.
  • standard math Standard projection and recursion lemmas from [12, 36] are valid.
    The appendix relies on Lemma A.2 from [36] and projection Lipschitz/quadratic lemmas from [12] without reproving them.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Decentralized projected Riemannian stochastic recursive momentum method for nonconvex optimization." pith.science (2026). https://pith.science/paper/FRO7E45K

@misc{pith2026241202382,
  author       = {Pith},
  title        = {Pith review of: Decentralized projected Riemannian stochastic recursive momentum method for nonconvex optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FRO7E45K}},
  note         = {Machine review of arXiv:2412.02382}
}
abstract

This paper studies decentralized optimization over a compact submanifold within a communication network of $n$ nodes, where each node possesses a smooth non-convex local cost function, and the goal is to jointly minimize the sum of these local costs. We focus particularly on the online setting, where local data is processed in real-time as it streams in, without the need for full data storage. We propose a decentralized projected Riemannian stochastic recursive momentum (DPRSRM) method that employs local hybrid stochastic gradient estimators and uses the network to track the global gradient. DPRSRM achieves an oracle complexity of \(\mathcal{O}(\epsilon^{-\frac{3}{2}})\), outperforming existing methods that have at most \(\mathcal{O}(\epsilon^{-2})\) complexity. Our method requires only $\mathcal{O}(1)$ gradient evaluations per iteration for each local node and does not require restarting with a large batch gradient. Furthermore, we demonstrate the effectiveness of our proposed methods compared to state-of-the-art ones through numerical experiments on principal component analysis problems and low-rank matrix completion.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Distributed Stochastic Proximal Algorithm on Riemannian Submanifolds for Weakly-convex Functions

    math.OC 2025-10 conditional novelty 6.0 of 10

    A retraction-based distributed stochastic proximal framework reaches consensus and a nearly stationary point at rate O((1+κ_g)/√k) for weakly-convex costs on compact embedded submanifolds.

Reference graph

Works this paper leans on

41 extracted references · 35 canonical work pages · cited by 1 Pith paper

  1. [1]

    R. K. Ando, T. Zhang, and P. Bartlett, A framework for learning predictive structures from multiple tasks and unlabeled data , Journal of Machine Learning Research, 6 (2005), pp. 1871– 1853

  2. [2]

    Balashov and R

    M. Balashov and R. Kamalov, The gradient projection method with Armijo’s step size on man- ifolds, Computational Mathematics and Mathematical Physics, 61 (2021), pp. 1776–1786

  3. [3]

    Bianchi and J

    P. Bianchi and J. Jakubowicz, Convergence of a multi-agent projected stochastic gradient al- gorithm for non-convex optimization , IEEE Transactions on Automatic Control, 58 (2012), pp. 391–405

  4. [4]

    Boumal and P.-A

    N. Boumal and P.-A. Absil, Low-rank matrix completion via preconditioned optimization on the Grassmann manifold, Linear Algebra and its Applications, 475 (2015), pp. 200–239

  5. [5]

    Cardot and D

    H. Cardot and D. Degras, Online principal component analysis in high dimension: Which algo- rithm to choose? , International Statistical Review, 86 (2018), pp. 29–50

  6. [6]

    J. Chen, H. Ye, M. W ang, T. Huang, G. Dai, I. Tsang, and Y. Liu, Decentralized riemannian conjugate gradient method on the stiefel manifold , in The Twelfth International Conference on Learning Representations, 2024, https://openreview.net/forum?id=PQbFUMKLFp

  7. [7]

    S. Chen, A. Garcia, M. Hong, and S. Shahrampour, Decentralized Riemannian gradient de- scent on the Stiefel manifold , in International Conference on Machine Learning, PMLR, 2021, pp. 1594–1605

  8. [8]

    Cho and J

    M. Cho and J. Lee, Riemannian approach to batch normalization, Advances in Neural Information Processing Systems, 30 (2017)

Show all 41 references
  1. [9]

    F. H. Clarke, R. J. Stern, and P. R. Wolenski, Proximal smoothness and the lower-C2 property, Journal of Convex Analysis, 2 (1995), pp. 117–144

  2. [10]

    Cutkosky and F

    A. Cutkosky and F. Orabona, Momentum-based variance reduction in non-convex sgd, Advances in neural information processing systems, 32 (2019)

  3. [11]

    Davis, D

    D. Davis, D. Drusvyatskiy, and Z. Shi, Stochastic optimization over proximally smooth sets , arXiv:2002.06309, (2020)

  4. [12]

    Deng and J

    K. Deng and J. Hu, Decentralized projected riemannian gradient method for smooth optimization on compact submanifolds , arXiv preprint arXiv:2304.08241, (2023)

  5. [13]

    K. Deng, J. Hu, and H. W ang, Decentralized douglas-rachford splitting methods for smooth opti- mization over compact submanifolds , arXiv preprint arXiv:2311.16399, (2023). 16 KANGKANG DENG, JIANG HU

  6. [14]

    K. Deng, J. Hu, and Z. Wen, Oracle complexity of augmented lagrangian methods for nonsmooth manifold optimization , arXiv preprint arXiv:2404.05121, (2024)

  7. [15]

    Di Lorenzo and G

    P. Di Lorenzo and G. Scutari, NEXT: In-network nonconvex optimization , IEEE Transactions on Signal and Information Processing over Networks, 2 (2016), pp. 120–136

  8. [16]

    Han and J

    A. Han and J. Gao, Improved variance reduction methods for riemannian non-convex optimiza- tion, IEEE Transactions on Pattern Analysis and Machine Intelligence, 44 (2021), pp. 7610– 7623

  9. [17]

    Han and J

    A. Han and J. Gao, Riemannian stochastic recursive momentum method for non-convex optimiza- tion, in Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21, Z.-H. Zhou, ed., 8 2021, pp. 2505–2511, https://doi.org/10.24963/ijcai.2021/34...

  10. [18]

    M. Hong, D. Hajinezhad, and M.-M. Zhao, Prox-PDA: The proximal primal-dual algorithm for fast distributed nonconvex optimization and learning over networks, in International Conference on Machine Learning, PMLR, 2017, pp. 1529–1538

  11. [19]

    J. Hu, R. Ao, A. M.-C. So, M. Yang, and Z. Wen, Riemannian natural gradient methods, SIAM Journal on Scientific Computing, 46 (2024), pp. A204–A231

  12. [20]

    Hu and K

    J. Hu and K. Deng, Improving the communication in decentralized manifold optimization through single-step consensus and compression , arXiv preprint arXiv:2407.08904, (2024)

  13. [21]

    J. Hu, K. Deng, N. Li, and Q. Li, Decentralized Riemannian natural gradient methods with Kronecker-product approximations, arXiv:2303.09611, (2023)

  14. [22]

    Kasai, P

    H. Kasai, P. Jawanpuria, and B. Mishra, Riemannian adaptive stochastic gradient algorithms on matrix manifolds , in International Conference on Machine Learning, PMLR, 2019, pp. 3262– 3271

  15. [23]

    LeCun, The mnist database of handwritten digits, http://yann

    Y. LeCun, The mnist database of handwritten digits, http://yann. lecun. com/exdb/mnist/, (1998)

  16. [24]

    Mishra, H

    B. Mishra, H. Kasai, P. Jawanpuria, and A. Saroop, A Riemannian gossip approach to subspace learning on Grassmann manifold , Machine Learning, 108 (2019), pp. 1783–1803

  17. [25]

    Qu and N

    G. Qu and N. Li, Harnessing smoothness to accelerate distributed optimization, IEEE Transactions on Control of Network Systems, 5 (2017), pp. 1245–1260

  18. [26]

    Sarlette and R

    A. Sarlette and R. Sepulchre, Consensus optimization on manifolds , SIAM Journal on Control and Optimization, 48 (2009), pp. 56–76

  19. [27]

    Scutari and Y

    G. Scutari and Y. Sun, Distributed nonconvex constrained optimization over time-varying di- graphs, Mathematical Programming, 176 (2019), pp. 497–544

  20. [28]

    S. M. Shah , Distributed optimization on Riemannian manifolds for multi-agent networks , arXiv:1711.11196, (2017)

  21. [29]

    W. Shi, Q. Ling, G. Wu, and W. Yin, EXTRA: An exact first-order algorithm for decentralized consensus optimization, SIAM Journal on Optimization, 25 (2015), pp. 944–966

  22. [30]

    H. Sun, S. Lu, and M. Hong, Improving the sample and communication complexity for decentral- ized non-convex optimization: Joint gradient estimation and tracking , in International Confer- ence on Machine Learning, PMLR, 2020, pp. 9217–9228

  23. [31]

    Y. Sun, S. Chen, A. Garcia, and S. Shahrampour, Global convergence of decentralized retraction- free optimization on the stiefel manifold , arXiv preprint arXiv:2405.11590, (2024)

  24. [32]

    Tatarenko and B

    T. Tatarenko and B. Touri, Non-convex distributed optimization , IEEE Transactions on Auto- matic Control, 62 (2017), pp. 3744–3757

  25. [33]

    H.-T. W ai, J. Lafond, A. Scaglione, and E. Moulines, Decentralized frank–wolfe algorithm for convex and nonconvex problems , IEEE Transactions on Automatic Control, 62 (2017), pp. 5522–5537

  26. [34]

    W ang and X

    L. W ang and X. Liu, Decentralized optimization over the Stiefel manifold by an approximate augmented Lagrangian function, IEEE Transactions on Signal Processing, 70 (2022), pp. 3029– 3041

  27. [35]

    W ang and X

    L. W ang and X. Liu, A variance-reduced stochastic gradient tracking algorithm for decentralized optimization with orthogonality constraints , Journal of Industrial and Management Optimiza- tion, 19 (2023), pp. 7753–7776

  28. [36]

    J. Xu, S. Zhu, Y. C. Soh, and L. Xie, Augmented distributed gradient methods for multi-agent optimization under uncoordinated constant stepsizes , in IEEE Conference on Decision and Control, 2015, pp. 2055–2060

  29. [37]

    Ye and T

    H. Ye and T. Zhang, DeEPCA: Decentralized exact PCA with linear convergence rate, The Journal of Machine Learning Research, 22 (2021), pp. 10777–10803

  30. [38]

    K. Yuan, B. Ying, X. Zhao, and A. H. Sayed, Exact diffusion for distributed optimization and DECENTRALIZED PROJECTED RIEMANNIAN GRADIENT METHOD 17 learning Part II: Convergence analysis , IEEE Transactions on Signal Processing, 67 (2018), pp. 724–739

  31. [39]

    Zeng and W

    J. Zeng and W. Yin, On nonconvex decentralized gradient descent , IEEE Transactions on Signal Processing, 66 (2018), pp. 2834–2848

  32. [40]

    J. Zhao, X. W ang, and J. Lei, Distributed riemannian stochastic gradient tracking algorithm on the stiefel manifold , arXiv preprint arXiv:2405.16900, (2024)

  33. [41]

    Zhou, X.-T

    P. Zhou, X.-T. Yuan, and J. Feng, Faster first-order methods for stochastic non-convex optimiza- tion on riemannian manifolds , in The 22nd International Conference on Artificial Intelligence and Statistics, PMLR, 2019, pp. 138–147. 18 KANGKANG DENG, JIANG HU Appendix A. T ech...

Pith tools

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