pith. sign in

arxiv: 1907.10570 · v1 · pith:TUP6AFM3new · submitted 2019-07-24 · 🧮 math.OC · cs.IT· math.IT

Infection-Curing Games over Polya Contagion Networks

Pith reviewed 2026-05-24 16:31 UTC · model grok-4.3

classification 🧮 math.OC cs.ITmath.IT
keywords infection-curing gamesPolya contagion networksNash equilibriumexpected network exposurezero-sum gamesnetwork epidemicsgradient descent
0
0 comments X

The pith

In Polya contagion network games, an expected network exposure proxy admits a Nash equilibrium computable by gradient descent.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper studies zero-sum infection-curing games on networks whose contagion follows a Polya urn scheme that incorporates spatial dependence between neighbors. Direct analysis of the average infection cost across the network is intractable, so the authors replace it with the simpler expected network exposure measure. They prove that the resulting game possesses a Nash equilibrium. This equilibrium can be found numerically by applying gradient descent. Simulations performed on small test networks supply empirical support that equilibria also exist when the original average infection cost is used.

Core claim

The authors establish that while the zero-sum game on average network infection is too complex to handle directly, the proxy game defined by expected network exposure admits a Nash equilibrium whose strategies are numerically recoverable by gradient descent; simulations on small networks indicate that the same conclusion holds for the original infection cost.

What carries the argument

Expected network exposure, introduced as a tractable proxy for average network infection cost in the zero-sum curing game.

If this is right

  • Equilibria of the proxy game can be located by standard numerical optimization.
  • The Polya urn formulation encodes spatial contagion between neighboring nodes.
  • Simulations indicate that Nash equilibria exist for the direct average-infection game as well.
  • The zero-sum structure permits application of known existence results once the proxy is adopted.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The proxy technique could be tested on other urn-style contagion models to check whether tractability carries over.
  • Numerical computation via gradient descent may scale to moderate-sized networks for determining approximate curing strategies.
  • If the proxy equilibria consistently align with the true cost on larger instances, the method could support control policies for real epidemic networks.

Load-bearing premise

The expected network exposure measure is close enough to the true average infection cost that equilibria computed for the proxy remain meaningful for the original game.

What would settle it

On a small network, compute the proxy equilibrium via gradient descent and check whether those same strategies fail to satisfy the Nash condition when evaluated directly under the average infection cost.

Figures

Figures reproduced from arXiv: 1907.10570 by Bahman Gharesifard, Fady Alajaji, Greg Harrington.

Figure 1
Figure 1. Figure 1: Illustration of a super urn in a network. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: , which depicts the shape of E[S˜ n|Fn−1] for a single node network. However, given our fixed allocation budgets Bb and Br, we restrict ourselves to considering sets of the form X × Y = {{∆b,i(n)} N i=1 ∈ R N ≥0 | PN i=1 ∆b,i(n) ≤ Bb} × {{∆r,i(n)} N i=1 ∈ R N ≥0 | PN i=1 ∆r,i(n) ≤ Br} [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparison of simulation results for three test networks. Three cases were compared over each network. Case one used the equilibrium policies [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
read the original abstract

We investigate infection-curing games on a network epidemics model based on the classical Polya urn scheme that accounts for spatial contagion among neighbouring nodes. We first consider the zero-sum game between competing agents using the cost measure for the average infection in the network. Due to the complexity of this problem we define a game on a proxy measure given by the so-called expected network exposure, and prove the existence of a Nash equilibrium that can be determined numerically using gradient descent algorithms. Finally, a number of simulations are performed on small test networks to provide empirical evidence that a Nash equilibrium exists for games defined on the average network infection.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper investigates zero-sum infection-curing games on Polya contagion networks. It first considers the game under the average network infection cost but, citing complexity, substitutes a proxy game defined via expected network exposure; it proves existence of a Nash equilibrium for the proxy game and shows it is numerically computable by gradient descent. Simulations on small test networks are then presented as empirical evidence that a Nash equilibrium also exists for the original average-infection game.

Significance. If the proxy equilibria reliably approximate those of the original game, the work supplies a tractable analytic and computational route to strategic equilibria in a spatially contagious epidemic model. The explicit proof of Nash existence for the proxy together with the reproducible simulation protocol constitute clear strengths; however, the absence of any error bound, monotonicity relation, or fixed-point correspondence between the two payoff functions limits the significance for the stated original problem.

major comments (3)
  1. [Abstract and the section introducing the proxy game] The central substitution replaces the stated objective (average network infection) with the proxy 'expected network exposure' because the former is 'too complex,' yet no quantitative complexity measure, approximation guarantee, or monotonicity relation between the two payoff functions is supplied. This substitution is load-bearing for the claim that the proxy analysis yields insight into the original game (see the paragraph beginning 'Due to the complexity of this problem' and the final simulation paragraph).
  2. [Simulations section] Simulations on small test networks are offered as 'empirical evidence that a Nash equilibrium exists for games defined on the average network infection,' but the reported results contain no error bounds, distance metrics between proxy and original equilibria, or sensitivity analysis with respect to network size. Without such controls it remains possible that the observed equilibria are artifacts of the substitution (see the final paragraph of the abstract and the simulations section).
  3. [Section containing the Nash existence proof for the proxy] The existence and gradient-descent computability results are proved only for the explicitly defined proxy game; the manuscript does not establish any fixed-point correspondence or continuity argument that would transfer these guarantees to the original average-infection cost (see the sentence 'we define a game on a proxy measure... and prove the existence of a Nash equilibrium').
minor comments (2)
  1. [Preliminaries] Notation for the Polya urn parameters and the precise definition of 'expected network exposure' should be collected in a single preliminary section to improve readability.
  2. [Simulations section] The manuscript should state the network sizes and number of Monte-Carlo replications used in the simulations so that the empirical evidence can be assessed for statistical reliability.

Simulated Author's Rebuttal

3 responses · 2 unresolved

We thank the referee for the constructive feedback. Below we respond point by point to the major comments, clarifying the intended scope of the proxy construction and the role of the simulations.

read point-by-point responses
  1. Referee: [Abstract and the section introducing the proxy game] The central substitution replaces the stated objective (average network infection) with the proxy 'expected network exposure' because the former is 'too complex,' yet no quantitative complexity measure, approximation guarantee, or monotonicity relation between the two payoff functions is supplied. This substitution is load-bearing for the claim that the proxy analysis yields insight into the original game (see the paragraph beginning 'Due to the complexity of this problem' and the final simulation paragraph).

    Authors: The manuscript does not assert that the proxy furnishes an approximation to the original average-infection game or that its Nash equilibria yield direct insight into the original payoffs. The proxy is introduced solely to obtain a game whose payoff functions admit a rigorous existence proof and gradient-based computation; the original game is addressed separately via simulation. We will revise the abstract and the relevant paragraph to state this separation more explicitly and to avoid any implication of approximation. revision: partial

  2. Referee: [Simulations section] Simulations on small test networks are offered as 'empirical evidence that a Nash equilibrium exists for games defined on the average network infection,' but the reported results contain no error bounds, distance metrics between proxy and original equilibria, or sensitivity analysis with respect to network size. Without such controls it remains possible that the observed equilibria are artifacts of the substitution (see the final paragraph of the abstract and the simulations section).

    Authors: The simulations are executed directly on sample paths of the Polya process using the average-infection cost; they are not run on the proxy. Their purpose is to illustrate that equilibria appear to exist in the original game on the tested instances. We agree that error bounds, distance metrics to proxy equilibria, and scaling studies would strengthen the empirical section, but the manuscript contains no such additional data or analysis. revision: no

  3. Referee: [Section containing the Nash existence proof for the proxy] The existence and gradient-descent computability results are proved only for the explicitly defined proxy game; the manuscript does not establish any fixed-point correspondence or continuity argument that would transfer these guarantees to the original average-infection cost (see the sentence 'we define a game on a proxy measure... and prove the existence of a Nash equilibrium').

    Authors: The observation is accurate. The existence and computability theorems apply exclusively to the proxy game. No fixed-point or continuity argument linking the two payoff structures is provided or claimed; the original game receives only simulation-based empirical support. revision: no

standing simulated objections not resolved
  • Establishing a fixed-point correspondence or continuity argument between the proxy and original payoff functions
  • Supplying quantitative approximation guarantees, error bounds, or monotonicity relations between the two cost measures

Circularity Check

0 steps flagged

No circularity: proxy game proven independently; simulations are separate empirical checks

full rationale

The paper explicitly states that the original average-infection cost is too complex, defines a distinct proxy (expected network exposure), and proves Nash equilibrium existence plus gradient-descent computability strictly for the proxy game. Simulations are offered only as empirical evidence that equilibria may exist for the original cost on small networks; no equation, theorem, or self-citation reduces the proxy result to the original cost or vice versa, and no fitted parameter is relabeled as a prediction. The derivation chain for the proxy is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit free parameters, axioms, or invented entities; ledger left empty due to lack of detail.

pith-pipeline@v0.9.0 · 5631 in / 1223 out tokens · 31423 ms · 2026-05-24T16:31:33.531398+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages

  1. [1]

    Easley and J

    D. Easley and J. Kleinberg, Networks, Crowds and Markets: Reasoning about a Highly Connected World . Cambridge Univ. Press, 2010

  2. [2]

    Virus spread in networks,

    P. V . Mieghem, J. Omic, and R. Kooij, “Virus spread in networks,” IEEE/ACM Trans. Netw. , vol. 17, no. 1, pp. 1–14, 2009

  3. [3]

    Analysis and control of epidemics: A survey of spreading processes on complex networks,

    C. Nowzari, V . M. Preciado, and G. J. Pappas, “Analysis and control of epidemics: A survey of spreading processes on complex networks,” IEEE Control Systems Magazine , vol. 36, pp. 26–46, 2016

  4. [4]

    A communication channel modeled on conta- gion,

    F. Alajaji and T. Fuja, “A communication channel modeled on conta- gion,” IEEE Trans. Inf. Theory , vol. 40, no. 6, pp. 2035–2041, 1994

  5. [5]

    Estimating social metwork structure and propagation dynamics for an infectious disease,

    L. Kim, M. Abramson, K. Drakopoulos, S. Kolitz, and A. Ozdaglar, “Estimating social metwork structure and propagation dynamics for an infectious disease,” in Proc. Int. Conf. Social Computing, Behavioral- Cultural Modeling, and Prediction , pp. 85–93, Springer, 2014

  6. [6]

    Modeling malware spreading dynamics,

    M. Garetto, W. Gong, and D. Towsley, “Modeling malware spreading dynamics,” in Proc. Conf. Comp. Comm., vol. 3, pp. 1869–1879, 2003

  7. [7]

    E. M. Rogers, Diffusion of Innovations . 5 ed., 2003

  8. [8]

    Tracking information epidemics in blogspace,

    E. Adar and L. A. Adamic, “Tracking information epidemics in blogspace,” in Proc. IEEE/WIC/ACM Int. Conf. Web Intelligence , pp. 207–214, 2005

  9. [9]

    A Polya urn-based model for epidemics on networks,

    M. Hayhoe, F. Alajaji, and B. Gharesifard, “A Polya urn-based model for epidemics on networks,” Proc. 2017 American Cont. Conf. , 2017

  10. [10]

    A Polya contagion model for networks,

    M. Hayhoe, F. Alajaji, and B. Gharesifard, “A Polya contagion model for networks,” IEEE Trans. Cont. Netw. Sys. , vol. 5, pp. 1998–2010, 2018

  11. [11]

    ¨Uber die statistik verketteter vorg¨ange,

    F. Eggenberger and G. Polya, “ ¨Uber die statistik verketteter vorg¨ange,” Z. Angew. Math. Mech. , vol. 3, no. 4, pp. 279–289, 1923

  12. [12]

    Sur l’interpr ´etation de certaines courbes de fr ´equences,

    G. Polya and F. Eggenberger, “Sur l’interpr ´etation de certaines courbes de fr ´equences,” Comptes Rendus C. R. , vol. 187, pp. 870–872, 1928

  13. [13]

    Sur quelques points de la th ´eorie des probabilit´es,

    G. Polya, “Sur quelques points de la th ´eorie des probabilit´es,” Annales de l’institut Henri Poincar ´e, vol. 1, no. 2, pp. 117–161, 1930

  14. [14]

    Game-theoretic protection against networked sis epidemics by human decision-makers,

    A. R. Hota and S. Sundaram, “Game-theoretic protection against networked sis epidemics by human decision-makers,” IF AC- PapersOnLine, vol. 51, 03 2017

  15. [15]

    Multi-competitive viruses over static and time-varying networks,

    P. E. Par, J. Liu, C. L. Beck, A. Nedi, and T. Baar, “Multi-competitive viruses over static and time-varying networks,” in 2017 American Control Conference (ACC) , pp. 1685–1690, May 2017

  16. [16]

    Curing with the network Polya contagion model,

    M. Hayhoe, F. Alajaji, and B. Gharesifard, “Curing with the network Polya contagion model,” Proc. 2017 American Cont. Conf. , 2018

  17. [17]

    Ash and C

    R. Ash and C. Dol ´eans-Dade, Probability and Measure Theory . 2000

  18. [18]

    G. R. Grimmett and D. R. Stirzaker, Probability and Random Pro- cesses. Oxford Univ. Press, 3 ed., 2001

  19. [19]

    Equilibrium points in n-person games,

    J. Nash, “Equilibrium points in n-person games,” in Proceedings of the National Academy of Sciences USA , vol. 36, pp. 48–49, 1950

  20. [20]

    Non-cooperative games,

    J. Nash, “Non-cooperative games,” The Annals of Mathematical Statis- tics, vol. 54, pp. 286–295, 1951

  21. [21]

    D. P. Bertsekas, A. Nedi ´c, and A. E. Ozdaglar, Convex Analysis and Optimization. Belmont, MA: Athena Scientific, 1st ed., 2003

  22. [22]

    D. P. Bertsekas, Nonlinear Programming . Belmont, MA: Athena Scientific, 2nd ed., 1999