{
  "id": "2302.10841",
  "version": "v1",
  "published": "2023-02-21T17:49:34.000Z",
  "updated": "2023-02-21T17:49:34.000Z",
  "title": "The Power of an Adversary in Glauber Dynamics",
  "authors": [
    "Byron Chin",
    "Ankur Moitra",
    "Elchanan Mossel",
    "Colin Sandon"
  ],
  "comment": "12 pages",
  "categories": [
    "math.PR"
  ],
  "abstract": "Glauber dynamics are a natural model of dynamics of dependent systems. While originally introduced in statistical physics, they have found important applications in the study of social networks, computer vision and other domains. In this work, we introduce a model of corrupted Glauber dynamics whereby instead of updating according to the prescribed conditional probabilities, some of the vertices and their updates are controlled by an adversary. We study the effect of such corruptions on global features of the system. Among the questions we study are: How many nodes need to be controlled in order to change the average statistics of the system in polynomial time? And how many nodes are needed to obstruct approximate convergence of the dynamics? Our results can be viewed as studying the robustness of classical sampling methods and are thus related to robust inference. The proofs connect to classical theory of Glauber dynamics from statistical physics.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2023-02-21T17:49:34.000Z"
    }
  ],
  "analyses": {
    "subjects": [
      "82C20",
      "60K35",
      "91D30"
    ],
    "keywords": [
      "glauber dynamics",
      "statistical physics",
      "obstruct approximate convergence",
      "important applications",
      "natural model"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 12,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}