pith. sign in

arxiv: 2107.13709 · v2 · submitted 2021-07-29 · 🧬 q-bio.PE · cs.SI· math.OC

Controlling epidemics through optimal allocation of test kits and vaccine doses across networks

Mingtao Xia , Lucas B\"ottcher , Tom Chou This is my paper

Pith reviewed 2026-05-24 13:04 UTC · model grok-4.3

classification 🧬 q-bio.PE cs.SImath.OC
keywords epidemic controloptimal allocationtesting and vaccinationscale-free networksSIR dynamicscontrol theorydegree-based mean-field
0
0 comments X

The pith

Optimal testing and vaccination policies that first target high-degree nodes then shift to lower-degree nodes delay outbreaks more effectively than uniform or reinforcement-learning strategies.

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

The paper builds a degree-based model of testing and vaccination on contact networks and applies control theory to find the best way to spend limited resources over time. It shows that the resulting policies begin by focusing resources on the most connected people before moving to less connected ones. This time-dependent targeting slows the epidemic and cuts case numbers more than spreading resources evenly or using reinforcement learning, with the biggest gains appearing on certain scale-free networks. A reader would care because the result gives a concrete rule for stretching scarce test kits and vaccine doses during an outbreak.

Core claim

Within a degree-based mean-field approximation of SIR dynamics, control-theoretic optimization yields intervention policies that first target high-degree nodes before shifting to lower-degree nodes in a time-dependent manner; these policies delay outbreaks and reduce incidence rates more than uniform allocation or reinforcement-learning interventions, particularly on certain scale-free networks.

What carries the argument

degree-based testing and vaccination model derived via control-theoretic methods that produces time-dependent allocation policies indexed by node degree

If this is right

  • Optimal policies outperform uniform resource allocation in delaying outbreaks and lowering incidence.
  • The same policies outperform reinforcement-learning interventions on scale-free networks.
  • The advantage is largest when the contact network has a broad degree distribution.
  • Shifting attention from high-degree to low-degree nodes at the right moment is required for optimality.

Where Pith is reading between the lines

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

  • Contact-tracing or vaccine-distribution systems could encode the high-to-low degree shift as a simple priority rule without needing full network data.
  • The same control framework could be applied to other limited interventions such as quarantine capacity or antiviral distribution.
  • On networks that are more homogeneous than scale-free graphs the performance gap between optimal and uniform policies is expected to shrink.

Load-bearing premise

The epidemic spread can be accurately captured by averaging the dynamics over all nodes that share the same number of contacts.

What would settle it

A network simulation or field dataset in which uniform testing and vaccination delays the outbreak peak by at least as much as the degree-targeted policy would falsify the superiority claim.

Figures

Figures reproduced from arXiv: 2107.13709 by Lucas B\"ottcher, Mingtao Xia, Tom Chou.

Figure 1
Figure 1. Figure 1: Degree distribution of a Barab´asi–Albert network a [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Optimal testing and quarantining strategy for the BA network using T = 200 and discount factor δ = 0.95. (a) A heatmap of the PMP-optimal testing strategy (see Alg. 1) for the BA network. The corresponding populations of degree-k susceptibles, untested infecteds, and tested infecteds are plotted in (b-d), respectively. (e) Time-evolution of the total fraction infected 1− PK k=1 sk(t) under the PMP-optimal … view at source ↗
Figure 3
Figure 3. Figure 3: Optimized vaccination model. (a) Heatmap of the optimal vaccination strategy vk(t)/(sk(t)Nk) for the BA network given by Alg. 1. (b,c) show the corresponding susceptible and infected subpopulations sk(t) and ik(t), while (d) plots the fraction infected as a function of time, derived from solving Eqs. (20)–(22) under optimal vaccination using a discount factor δ = 0.95. The dashed red curve indicates the fr… view at source ↗
Figure 4
Figure 4. Figure 4: Total fraction infected under testing or vaccination model as a function of different intervention starting times t0. We minimize the corresponding loss function at T = 150 and use δ = 0.95. (a) The fraction infected in the BA network as a function of start times using different testing amplitudes F. At larger F, there is a decrease in infected population for later starts of testing at early times. This no… view at source ↗
Figure 5
Figure 5. Figure 5: Dependence of intervention effectiveness on the degree of the initial infected individual. (a) The PMP-optimal testing strategy computed using IC2 (ki = 20) on the BA network. Strategies for IC1 (ki = 3) and IC3 (ki = 90) are qualitatively similar (not shown) with small differences at the beginning leading to the different delays in the infection dynamics shown in (b). Specifically, for IC1 and IC3 the ini… view at source ↗
Figure 6
Figure 6. Figure 6: Illustration of the neural network used to identify effective testing and vaccination strategies. The inputs of the input layer are (sk(t);i u k(t);i ∗ k(t)) ∈ R 3K. For each hidden layer i (1 ≤ i ≤ NH ), we normalize the corresponding outputs xi,j for all samples in a minibatch such that the resulting values xˆi,j have zero mean and unit variance. These values are used as inputs to a rectified linear unit… view at source ↗
Figure 7
Figure 7. Figure 7: Reduction in fractions of infected individuals calculated as the difference between the fractions infected obtained with testing and without testing for the BA network shown in (a) and the SBM network shown in (b). The optimal control approach based on PMP reduces early infections the most. RL outperforms uniform testing in reducing the number of early-stage infections. Additionally, the effect of the opti… view at source ↗
read the original abstract

Efficient testing and vaccination protocols are critical aspects of epidemic management. To study the optimal allocation of limited testing and vaccination resources in a heterogeneous contact network of interacting susceptible, recovered, and infected individuals, we present a degree-based testing and vaccination model for which we use control-theoretic methods to derive optimal testing and vaccination policies. Within our framework, we find that optimal intervention policies first target high-degree nodes before shifting to lower-degree nodes in a time-dependent manner. Using such optimal policies, it is possible to delay outbreaks and reduce incidence rates to a greater extent than uniform and reinforcement-learning-based interventions, particularly on certain scale-free networks.

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

2 major / 2 minor

Summary. The paper introduces a degree-based mean-field SIR model on heterogeneous contact networks and applies optimal control to derive time-dependent testing and vaccination allocation policies. These policies prioritize high-degree nodes early before shifting to lower-degree nodes. The resulting interventions are claimed to delay outbreaks and reduce incidence more than uniform allocation or reinforcement-learning baselines, with the advantage most pronounced on certain scale-free networks.

Significance. If the performance gains hold under the stochastic network dynamics the authors ultimately simulate, the work supplies a control-theoretic route to resource allocation that improves on both static heuristics and black-box learning methods. The explicit time-dependent targeting schedule is a concrete, potentially actionable output.

major comments (2)
  1. [Model and control derivation] The derivation of the optimal policies rests on the deterministic degree-based mean-field closure (abstract and model section). The manuscript then applies these policies to finite stochastic networks, yet provides no error bound or consistency argument showing that the mean-field optimum remains near-optimal once stochastic extinction, local clustering around hubs, and finite-size fluctuations are restored. This gap is load-bearing for the central claim of superiority over baselines.
  2. [Numerical experiments] Table or figure reporting simulation results: the performance advantage is stated for “certain scale-free networks,” but the manuscript does not report network size N, number of independent realizations, or confidence intervals on the incidence reduction. Without these, it is impossible to judge whether the reported gains exceed sampling variability.
minor comments (2)
  1. Notation for the control inputs (testing rate per degree class, vaccination rate per degree class) should be introduced once and used consistently; several passages reuse symbols without redefinition.
  2. The abstract states that the framework “uses this degree-based testing and vaccination model,” but the precise mapping from the continuous-time control problem to the discrete-time network simulations is not summarized in one place.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive comments. We address each major comment below and indicate the revisions we plan to make.

read point-by-point responses

  1. Referee: [Model and control derivation] The derivation of the optimal policies rests on the deterministic degree-based mean-field closure (abstract and model section). The manuscript then applies these policies to finite stochastic networks, yet provides no error bound or consistency argument showing that the mean-field optimum remains near-optimal once stochastic extinction, local clustering around hubs, and finite-size fluctuations are restored. This gap is load-bearing for the central claim of superiority over baselines.

    Authors: We agree that the mean-field derivation does not come with formal error bounds for the stochastic setting. In the revised version, we will add a dedicated subsection in the discussion that explicitly acknowledges this limitation, discusses the potential impact of stochastic effects and finite-size fluctuations on the optimality of the derived policies, and suggests directions for future work on rigorous approximation guarantees. We believe this addresses the concern without requiring a full theoretical consistency proof, which would be a substantial extension. revision: partial

  2. Referee: [Numerical experiments] Table or figure reporting simulation results: the performance advantage is stated for “certain scale-free networks,” but the manuscript does not report network size N, number of independent realizations, or confidence intervals on the incidence reduction. Without these, it is impossible to judge whether the reported gains exceed sampling variability.

    Authors: We thank the referee for pointing this out. In the revised manuscript, we will update the numerical experiments section to include the network sizes used (N), the number of independent stochastic realizations performed for each experiment, and confidence intervals (or standard errors) on the reported incidence reductions. This will allow readers to assess the statistical significance of the observed performance advantages. revision: yes

Circularity Check

0 steps flagged

No circularity: standard optimal control on mean-field SIR is self-contained

full rationale

The derivation applies control-theoretic optimization to a degree-based mean-field SIR model to obtain time-dependent testing/vaccination policies, then compares them to uniform and RL baselines. No quoted step reduces a claimed result to a fitted parameter, self-definition, or load-bearing self-citation chain; the mean-field closure and optimality conditions are independent of the final performance claims. The reader's assessment of score 2.0 is consistent with this self-contained structure.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract-only; the central claim rests on the degree-based mean-field network model and standard control-theoretic optimality conditions, with no free parameters or invented entities visible at this level.

axioms (1)
  • domain assumption Degree-based mean-field approximation suffices to capture SIR dynamics on heterogeneous contact networks
    Invoked to enable the control-theoretic derivation of allocation policies.

pith-pipeline@v0.9.0 · 5635 in / 1075 out tokens · 22429 ms · 2026-05-24T13:04:20.656129+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

48 extracted references · 48 canonical work pages

  1. [1]

    Assessment of specimen pooling to conserve SARS-CoV -2 testing resources,

    B. Abdalhamid, C. R. Bilder, E. L. McCutchen, S. H. Hinric hs, S. A. Koepsell, and P . C. Iwen, “Assessment of specimen pooling to conserve SARS-CoV -2 testing resources,” American Journal of Clinical Pathol- ogy, vol. 153, no. 6, pp. 715–718, 2020

    work page 2020

  2. [2]

    Evaluation of COVID-19 RT-qPCR test in multi sample pools,

    I. Y elin, N. Aharony, E. S. Tamar, A. Argoetti, E. Messer, D. Berenbaum, E. Shafran, A. Kuzli, N. Gandali, O. Shkedi et al. , “Evaluation of COVID-19 RT-qPCR test in multi sample pools,” Clinical Infectious Diseases, vol. 71, no. 16, pp. 2073–2078, 2020

    work page 2073

  3. [3]

    Quarantine and testing strategies in contact tracing for SARS- CoV -2: a modelling study,

    B. J. Quilty, S. Clifford, J. Hellewell, T. W. Russell, A. J. Kucharski, S. Flasche, W. J. Edmunds, K. E. Atkins, A. M. Foss, N. R. Water low et al. , “Quarantine and testing strategies in contact tracing for SARS- CoV -2: a modelling study,” The Lancet Public Health , 2021

    work page 2021

  4. [4]

    Modelling optimal vaccination strategy for SARS-CoV -2 in the UK,

    S. Moore, E. M. Hill, L. Dyson, M. J. Tildesley, and M. J. Ke eling, “Modelling optimal vaccination strategy for SARS-CoV -2 in the UK,” PLoS Computational Biology , vol. 17, p. e1008849, 2021

    work page 2021

  5. [5]

    Optimal vaccination patterns in age-struc tured populations,

    J. M¨ uller, “Optimal vaccination patterns in age-struc tured populations,” SIAM Journal on Applied Mathematics , vol. 59, no. 1, pp. 222–241, 1998

    work page 1998

  6. [6]

    The Optimal V accination Strategy to Control COVID -19: A Modeling Study Based on the Transmission Scenario in Wuhan C ity, China,

    Z. Zhao, Y . Niu, L. Luo, Q. Hu, T. Y ang, M. Chu, Q. Chen, Z. Le i, J. Rui, S. Lin, Y . Wang, J. Xu, Y . Zhu, X. Liu, M. Y ang, J. Huang, W. Liu, B. Deng, C. Liu, Z. Li, P . Li, Y . Su, B. Zhao, R. Frutos, a nd T. Chen, “The Optimal V accination Strategy to Control COVID -19: A Modeling Study Based on the Transmission Scenario in Wuhan C ity, China,” 202...

    work page 2020

  7. [7]

    V accination strategies against COVID- 19 and the diffusion of anti-vaccination views,

    R. P . Curiel and H. G. Ram´ ırez, “V accination strategies against COVID- 19 and the diffusion of anti-vaccination views,” Scientific Reports , vol. 11, p. 6626, 2021

    work page 2021

  8. [8]

    COVID-19 optimal vaccination policies: A modeli ng study on efficacy, natural and vaccine-induced immunity responses,

    M. A. Acu˜ na-Zegarra, S. D´ ıaz-Infante, D. Baca-Carrasco, and D. Olmos- Liceaga, “COVID-19 optimal vaccination policies: A modeli ng study on efficacy, natural and vaccine-induced immunity responses, ” Mathemati- cal Biosciences , vol. 337, p. 108614, 2021

    work page 2021

  9. [9]

    New COVID-19 V ariants,

    CDC, “New COVID-19 V ariants,” 2021, ac- cessed: January 15, 2021. [Online]. Available: https://www.cdc.gov/coronavirus/2019-ncov/transmission/variant.html

    work page 2021

  10. [10]

    Optimal strategies for social dist ancing and testing to control COVID-19,

    W. Choi and E. Shim, “Optimal strategies for social dist ancing and testing to control COVID-19,” Journal of Theoretical Biology , vol. 512, p. 110568, 2021

    work page 2021

  11. [11]

    Optimal con- trol of epidemic size and duration with limited resources,

    L. Bolzoni, E. Bonacini, R. Della Marca, and M. Groppi, “ Optimal con- trol of epidemic size and duration with limited resources,” Mathematical Biosciences, vol. 315, p. 108232, 2019

    work page 2019

  12. [12]

    Optimal vaccination strategie s for the control of epidemics in highly mobile populations,

    P . Ogren and C. Martin, “Optimal vaccination strategie s for the control of epidemics in highly mobile populations,” in Proceedings of the 39th IEEE Conference on Decision and Control (Cat. No. 00CH37187 ), vol. 2. IEEE, 2000, pp. 1782–1787

    work page 2000

  13. [13]

    R isk- aware identification of highly suspected COVID-19 cases in S ocial 13 IoT: A joint graph theory and reinforcement learning approa ch,

    B. Wang, Y . Sun, T. Q. Duong, L. D. Nguyen, and L. Hanzo, “R isk- aware identification of highly suspected COVID-19 cases in S ocial 13 IoT: A joint graph theory and reinforcement learning approa ch,” IEEE Access, vol. 8, pp. 115 655–115 661, 2020

    work page 2020

  14. [14]

    Optimal control for heterogeneous n ode-based information epidemics over social networks,

    F. Liu and M. Buss, “Optimal control for heterogeneous n ode-based information epidemics over social networks,” IEEE Transactions on Control of Network Systems , vol. 7, no. 3, pp. 1115–1126, 2020

    work page 2020

  15. [15]

    Newman, Networks, 2nd Edition

    M. Newman, Networks, 2nd Edition . Oxford University Press, 2018

    work page 2018

  16. [16]

    The transmission dynamics of human immunodeficiency virus (HIV),

    R. M. May and R. M. Anderson, “The transmission dynamics of human immunodeficiency virus (HIV),” Philosophical Transactions of the Royal Society of London. B, Biological Sciences , vol. 321, no. 1207, pp. 565– 607, 1988

    work page 1988

  17. [17]

    V elocity and hierarchical spread of epidemic outbreaks in scale-free networks,

    M. Barth´ elemy, A. Barrat, R. Pastor-Satorras, and A. V espignani, “V elocity and hierarchical spread of epidemic outbreaks in scale-free networks,” Physical Review Letters , vol. 92, no. 17, p. 178701, 2004

    work page 2004

  18. [18]

    Epidemic dynam ics and endemic states in complex networks,

    R. Pastor-Satorras and A. V espignani, “Epidemic dynam ics and endemic states in complex networks,” Physical Review E , vol. 63, no. 6, p. 066117, 2001

    work page 2001

  19. [19]

    Epidemic spreading in scale-free networks,

    ——, “Epidemic spreading in scale-free networks,” Physical Review Letters, vol. 86, no. 14, p. 3200, 2001

    work page 2001

  20. [20]

    Ab sence of epi- demic threshold in scale-free networks with degree correla tions,

    M. Bogun´ a, R. Pastor-Satorras, and A. V espignani, “Ab sence of epi- demic threshold in scale-free networks with degree correla tions,” Phys- ical Review Letters , vol. 90, no. 2, p. 028701, 2003

    work page 2003

  21. [21]

    Dynamical patterns of epidemic outbreaks in complex heter ogeneous networks,

    M. Barth´ elemy, A. Barrat, R. Pastor-Satorras, and A. V espignani, “Dynamical patterns of epidemic outbreaks in complex heter ogeneous networks,” Journal of Theoretical Biology , vol. 235, no. 2, pp. 275–288, 2005

    work page 2005

  22. [22]

    The effect of conta ct heterogeneity and multiple routes of transmission on final e pidemic size,

    I. Z. Kiss, D. M. Green, and R. R. Kao, “The effect of conta ct heterogeneity and multiple routes of transmission on final e pidemic size,” Mathematical Biosciences , vol. 203, no. 1, pp. 124–136, 2006

    work page 2006

  23. [23]

    Effective degree network disease models,

    J. Lindquist, J. Ma, P . V an den Driessche, and F. H. Wille boordse, “Effective degree network disease models,” Journal of Mathematical Biology, vol. 62, no. 2, pp. 143–164, 2011

    work page 2011

  24. [24]

    Using exces s deaths and testing statistics to determine COVID-19 mortalities,

    L. B¨ ottcher, M. R. D’Orsogna, and T. Chou, “Using exces s deaths and testing statistics to determine COVID-19 mortalities,” European Journal of Epidemiology, vol. 36, pp. 545–558, 2021

    work page 2021

  25. [25]

    A statistic al model of COVID-19 testing in populations: effects of sampling bias a nd testing errors,

    L. B¨ ottcher, M. R. D’Orsogna, and T. Chou, “A statistic al model of COVID-19 testing in populations: effects of sampling bias a nd testing errors,” Philos. Trans. Royal Soc. A , 2021

    work page 2021

  26. [26]

    A place -focused model for social networks in cities,

    C. Brown, A. Noulas, C. Mascolo, and V . Blondel, “A place -focused model for social networks in cities,” in 2013 International Conference on Social Computing . IEEE, 2013, pp. 75–80

    work page 2013

  27. [27]

    Statistical mechanics of complex net- works,

    R. Albert and A.-L. Barab´ asi, “Statistical mechanics of complex net- works,” Reviews of Modern Physics , vol. 74, no. 1, p. 47, 2002

    work page 2002

  28. [28]

    Emergence of scaling in random net- works,

    A.-L. Barab´ asi and R. Albert, “Emergence of scaling in random net- works,” Science, vol. 286, no. 5439, pp. 509–512, 1999

    work page 1999

  29. [29]

    Stochast ic blockmodels: First steps,

    P . W. Holland, K. B. Laskey, and S. Leinhardt, “Stochast ic blockmodels: First steps,” Social networks , vol. 5, no. 2, pp. 109–137, 1983

    work page 1983

  30. [30]

    Consistency of community detection in networks under degree-corrected stochastic block models,

    Y . Zhao, E. Levina, J. Zhu et al., “Consistency of community detection in networks under degree-corrected stochastic block models, ” The Annals of Statistics , vol. 40, no. 4, pp. 2266–2292, 2012

    work page 2012

  31. [31]

    Reproduction num bers and sub-threshold endemic equilibria for compartmental model s of disease transmission,

    P . V an den Driessche and J. Watmough, “Reproduction num bers and sub-threshold endemic equilibria for compartmental model s of disease transmission,” Mathematical Biosciences , vol. 180, no. 1-2, pp. 29–48, 2002

    work page 2002

  32. [32]

    Estimating the basic reproduction number of COVID-19 in Wuhan, China,

    Y . Wang, X. Y ou, Y . Wang, L. Peng, Z. Du, S. Gilmour, D. Y on eoka, J. Gu, C. Hao, Y . Hao et al., “Estimating the basic reproduction number of COVID-19 in Wuhan, China,” Chinese Journal of Epidemiology , vol. 41, no. 4, pp. 476–479, 2020

    work page 2020

  33. [33]

    Assessment of the SARS-C oV -2 basic reproduction number, R0, based on the early phase of COVID-1 9 outbreak in Italy,

    M. D’Arienzo and A. Coniglio, “Assessment of the SARS-C oV -2 basic reproduction number, R0, based on the early phase of COVID-1 9 outbreak in Italy,” Biosafety and Health , vol. 2, no. 2, pp. 57–59, 2020

    work page 2020

  34. [34]

    Global convergence of COVID-19 basic reproduction number and esti mation from early-time SIR dynamics,

    G. G. Katul, A. Mrad, S. Bonetti, G. Manoli, and A. J. Paro lari, “Global convergence of COVID-19 basic reproduction number and esti mation from early-time SIR dynamics,” PLoS One, vol. 15, no. 9, p. e0239800, 2020

    work page 2020

  35. [35]

    COVI D-19 pandemic and its recovery time of patients in India: A pilot s tudy,

    M. P . Barman, T. Rahman, K. Bora, and C. Borgohain, “COVI D-19 pandemic and its recovery time of patients in India: A pilot s tudy,” Diabetes & Metabolic Syndrome: Clinical Research & Reviews , vol. 14, no. 5, pp. 1205–1211, 2020

    work page 2020

  36. [36]

    Reduction in effective rep roduction number of COVID-19 is higher in countries employing active c ase detection with prompt isolation,

    C. Wilasang, C. Sararat, N. C. Jitsuk, N. Y olai, P . Thamm awijaya, P . Auewarakul, and C. Modchang, “Reduction in effective rep roduction number of COVID-19 is higher in countries employing active c ase detection with prompt isolation,” Journal of Travel Medicine , vol. 27, no. 5, pp. 1–3, 2020

    work page 2020

  37. [37]

    Analysis of the impact of lockd own on the reproduction number of the SARS-Cov-2 in Spain,

    A. Hyafil and D. Mori˜ na, “Analysis of the impact of lockd own on the reproduction number of the SARS-Cov-2 in Spain,” Gaceta sanitaria , 2020

    work page 2020

  38. [38]

    Optimal stra tegy of vaccination & treatment in an SIR epidemic model,

    G. Zaman, Y . H. Kang, G. Cho, and I. H. Jung, “Optimal stra tegy of vaccination & treatment in an SIR epidemic model,” Mathematics and Computers in Simulation , vol. 136, pp. 63–77, 2017

    work page 2017

  39. [39]

    COVID-19 vaccines: a race against time,

    N. Peiffer-Smadja, S. Rozencwajg, Y . Kherabi, Y . Y azda npanah, and P . Montravers, “COVID-19 vaccines: a race against time,” Anaesthesia, Critical Care & Pain Medicine , vol. 40, no. 2, p. 100848, 2021

    work page 2021

  40. [40]

    Decisive Conditions for St rategic V accination against SARS-CoV -2,

    L. B¨ ottcher and J. Nagler, “Decisive Conditions for St rategic V accination against SARS-CoV -2,” medRxiv, 2021

    work page 2021

  41. [41]

    A global database of COV ID-19 vaccinations,

    E. Mathieu, H. Ritchie, E. Ortiz-Ospina, M. Roser, J. Ha sell, C. Appel, C. Giattino, and L. Rod´ es-Guirao, “A global database of COV ID-19 vaccinations,” Nature Human Behaviour , pp. 1–7, 2021

    work page 2021

  42. [42]

    Switchover phenomenon induced by epidemic seeding on geometric networ ks,

    G. ´Odor, D. Czifra, J. Komj´ athy, L. Lov´ asz, and M. Karsai, “Switchover phenomenon induced by epidemic seeding on geometric networ ks,” arXiv preprint arXiv:2106.16070 , 2021

    work page arXiv 2021

  43. [43]

    Percola tion central- ity: Quantifying graph-theoretic impact of nodes during pe rcolation in networks,

    M. Piraveenan, M. Prokopenko, and L. Hossain, “Percola tion central- ity: Quantifying graph-theoretic impact of nodes during pe rcolation in networks,” PloS One , vol. 8, no. 1, p. e53095, 2013

    work page 2013

  44. [44]

    Disease-induced resource constraints can tri gger explosive epidemics,

    L. B¨ ottcher, O. Woolley-Meza, N. A. Ara´ ujo, H. J. Herr mann, and D. Helbing, “Disease-induced resource constraints can tri gger explosive epidemics,” Scientific Reports , vol. 5, no. 1, pp. 1–11, 2015

    work page 2015

  45. [45]

    Neu ral ordinary differential equation control of dynamics on graphs,

    T. Asikis, L. B¨ ottcher, and N. Antulov-Fantulin, “Neu ral ordinary differential equation control of dynamics on graphs,” 2021

    work page 2021

  46. [46]

    Implicit energy regular- ization of neural ordinary-differential-equation contro l,

    L. B¨ ottcher, N. Antulov-Fantulin, and T. Asikis, “Implicit energy regular- ization of neural ordinary-differential-equation contro l,” arXiv preprint arXiv:2103.06525, 2021

    work page arXiv 2021

  47. [47]

    Generation of arbitrarily two-p oint-correlated random networks,

    S. Weber and M. Porto, “Generation of arbitrarily two-p oint-correlated random networks,” Physical Review E , vol. 76, no. 4, p. 046111, 2007

    work page 2007

  48. [48]

    Human-level control through deep reinforcement learnin g,

    V . Mnih, K. Kavukcuoglu, D. Silver, A. A. Rusu, J. V eness , M. G. Bellemare, A. Graves, M. Riedmiller, A. K. Fidjeland, G. Ost rovski et al. , “Human-level control through deep reinforcement learnin g,” Nature, vol. 518, no. 7540, pp. 529–533, 2015. Mingtao Xia is a Ph.D. candidate in the Department of Mathematics at the University of California, Los An...

    work page 2015