On MUON optimization: From non-convergence to an error analysis with Polar Express and the Newton-Schulz polynomial from implementations

• URL: http://arxiv.org/abs/2608.04607v1
• Date: Wed, 05 Aug 2026 09:14:22 GMT
• Title: On MUON optimization: From non-convergence to an error analysis with Polar Express and the Newton-Schulz polynomial from implementations
• Abstract summary: gradient descent (SGD) optimization methods are the standard instruments for the training of deep neural networks (DNNs)<n>One of the most popular methods of such accelerated variants is the momentumized by Newton-Schulz (MUON) proposed by Jordan et al. in 2024.<n>We show for almost every mini-batch size that MUON fails to converge to the solution the optimization problem as the number of gradient steps converges to infinity.<n>We also establish an error analysis for MUON with the generalized NS steps that provides convergence rates in terms of the number of gradient steps and in terms of the mini-batch size
• Score: 1.8734449181723825
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Stochastic gradient descent (SGD) optimization methods are the standard instruments for the training of deep neural networks (DNNs). In many relevant artificial intelligence (AI) systems - such as popular large language models (LLMs)-not the standard SGD scheme is used as the optimization method but instead suitable accelerated variants of SGD are employed. One of the most popular methods of such accelerated SGD variants is the momentum orthogonalized by Newton-Schulz (MUON) optimizer proposed by Jordan et al. in 2024. The MUON optimizer exploits the special matrix structure of the weight parameters in the training of the DNNs and, in its original form, employs five Newton-Schultz (NS) matrix steps in each MUON iteration. In this work we propose and study a generalized variant of the MUON optimizer involving an arbitrary number of generalized NS steps with polynomials of possibly arbitrary high degree. The considered optimizer covers MUON with the original NS polynomial as well as MUON combined with the recently proposed Polar Express method as special cases. For a simple class of stochastic optimization problems (SOPs) we show for almost every mini-batch size that MUON fails to converge to the solution of the SOP as the number of gradient steps converges to infinity. We also establish an error analysis for MUON with the generalized NS steps that provides convergence rates in terms of the number of gradient steps and in terms of the size of the mini-batch. We illustrate our general error analysis for MUON in the case of several concrete examples including quadratic stochastic optimization problems (SOPs) as well as $\ell_2$ regularized logistic regression for binary classification.

Related papers

• Federated Compositional Muon Optimizer for Matrix-Wise Models [12.397558747278305]
  We propose an effective compositional Muon (FedCoMuon) to solve distributed matrix-wise compositional problems.<n>We show that our proposed methods are competitive with existing compositional baselines and achieve the best reported accuracy in several settings.
  arXiv  Detail & Related papers  (2026-08-13T01:46:26Z)
• Muon is Provably Faster with Momentum Variance Reduction [55.388203260208485]
  Recent empirical research has demonstrated that deep learnings based on the linear linear oracle (LMO) over specifically chosen Non-Eudean.<n>Adam-type training methods outperform the minimization of large language models.
  arXiv  Detail & Related papers  (2025-12-18T14:38:39Z)
• Preconditioned Norms: A Unified Framework for Steepest Descent, Quasi-Newton and Adaptive Methods [50.070182958880146]
  We propose a unified framework generalizing descent, quasi-Newton methods, and adaptive methods through the novel notion of preconditioned matrix norms.<n>Within this framework, we provide the first systematic treatment of affine and scale invariance in the matrix- parameterized setting.<n>We introduce two new methods, $ttMuAdam$ and $texttMuAdam-SANIA$, which combine the spectral geometry of Muon with Adam-style preconditioning.
  arXiv  Detail & Related papers  (2025-10-12T19:39:41Z)
• Leveraging Coordinate Momentum in SignSGD and Muon: Memory-Optimized Zero-Order [39.25335214877435]
  Fine-tuning Large Language Models (LLMs) is essential for adapting pre-trained models to downstream tasks.<n>Traditional first-order algorithms incur prohibitive memory and computational costs that scale poorly with model size.<n>We propose zero-order (ZO) optimization methods as a memory- and compute-efficient alternative.
  arXiv  Detail & Related papers  (2025-06-04T20:27:17Z)
• Training Deep Learning Models with Norm-Constrained LMOs [56.00317694850397]
  We propose a new family of algorithms that uses the linear minimization oracle (LMO) to adapt to the geometry of the problem.<n>We demonstrate significant speedups on nanoGPT training using our algorithm, Scion, without any reliance on Adam.
  arXiv  Detail & Related papers  (2025-02-11T13:10:34Z)
• Federated Conditional Stochastic Optimization [110.513884892319]
  Conditional optimization has found in a wide range of machine learning tasks, such as in-variant learning tasks, AUPRC, andAML. This paper proposes algorithms for distributed federated learning.
  arXiv  Detail & Related papers  (2023-10-04T01:47:37Z)
• Unified Convergence Analysis for Adaptive Optimization with Moving Average Estimator [75.05106948314956]
  We show that an increasing large momentum parameter for the first-order moment is sufficient for adaptive scaling.<n>We also give insights for increasing the momentum in a stagewise manner in accordance with stagewise decreasing step size.
  arXiv  Detail & Related papers  (2021-04-30T08:50:24Z)
• Positive-Negative Momentum: Manipulating Stochastic Gradient Noise to Improve Generalization [89.7882166459412]
  gradient noise (SGN) acts as implicit regularization for deep learning. Some works attempted to artificially simulate SGN by injecting random noise to improve deep learning. For simulating SGN at low computational costs and without changing the learning rate or batch size, we propose the Positive-Negative Momentum (PNM) approach.
  arXiv  Detail & Related papers  (2021-03-31T16:08:06Z)
• On the Promise of the Stochastic Generalized Gauss-Newton Method for Training DNNs [37.96456928567548]
  We study a generalized Gauss-Newton method (SGN) for training DNNs. SGN is a second-order optimization method, with efficient iterations, that we demonstrate to often require substantially fewer iterations than standard SGD to converge. We show that SGN does not only substantially improve over SGD in terms of the number of iterations, but also in terms of runtime. This is made possible by an efficient, easy-to-use and flexible implementation of SGN we propose in the Theano deep learning platform.
  arXiv  Detail & Related papers  (2020-06-03T17:35:54Z)
• Global Optimization of Gaussian processes [52.77024349608834]
  We propose a reduced-space formulation with trained Gaussian processes trained on few data points. The approach also leads to significantly smaller and computationally cheaper sub solver for lower bounding. In total, we reduce time convergence by orders of orders of the proposed method.
  arXiv  Detail & Related papers  (2020-05-21T20:59:11Z)
• Dual Stochastic Natural Gradient Descent and convergence of interior half-space gradient approximations [0.0]
  Multinomial logistic regression (MLR) is widely used in statistics and machine learning. gradient descent (SGD) is the most common approach for determining the parameters of a MLR model in big data scenarios.
  arXiv  Detail & Related papers  (2020-01-19T00:53:49Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.