Multi-Dimensional Opinion Formation

Hanna Bartel; Marie-Therese Wolfram; Martin Burger

REVIEW 3 major objections 6 minor 1 cited by

Multi-Dimensional Opinion Formation

T0 review · 3 major / 6 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read When people weigh topics differently, opinion dynamics become one-way and can split or swing groups.

desk verdict New weighted multi-dimensional opinion model with interesting non-reciprocal dynamics, but the paper never states the scaling that connects its microscopic rule to the mean-field PDE it actually analyzes. read the letter →

arxiv 2601.04074 v3 pith:NBN6HZIG submitted 2026-01-07 physics.soc-ph math.APmath.DS

Hanna Bartel , Martin Burger , Marie-Therese Wolfram This is my paper · ORCID

classification physics.soc-phmath.APmath.DS MSC 35Q8335Q9182C4091D30 PACS 89.65.-s

keywords opiniondynamicsmulti-dimensionalopinionsimportanceweightsmean-fieldlimitkineticequationVlasovboundedconfidencestationarystates

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 a multi-topic opinion model in which each person has both an opinion vector and a personal importance weight per topic, and the change in any one opinion depends on the weighted similarity of the entire opinion vector between two people. It derives a mean-field (Vlasov-type) equation for the population distribution and shows that when importance weights differ across people, interactions are no longer reciprocal: one person can influence another without being influenced back. This asymmetry breaks conservation of the mean opinion and allows the variance to increase, which the authors demonstrate with explicit stationary states: consensus, separated clusters, and interacting clusters where opposing influences cancel exactly. Numerical simulations illustrate that the final opinion structure—including a left-to-right swing on all topics—is critically controlled by the importance weights and the interaction radius. A sympathetic reader cares because the model offers a mechanism, grounded in individual weighting of topics, for complex and empirically plausible opinion patterns that single-topic or symmetric multi-topic models cannot produce.

What carries the argument

The load-bearing object is the weighted interaction distance p_a(x,y,α) of equation (1), which is not a norm because p_a=0 does not imply x=y. It enters the binary interaction rule (2) through the component-wise interaction vector ϕ_xy^α = (ϕ(p_1),...,ϕ(p_d)), where ϕ is a non-increasing bounded-confidence function. The mean-field limit (grazing collision limit and time rescaling) transforms the kinetic equation (3) into the weak form (4) and the Vlasov-type equation (5), whose velocity field is ∫ ϕ_xy^α ⊙ (y−x) f(y,η,t) d(y,η). The key work of this machinery is that it makes the effective coupling between topics depend on the agent's own weights, producing the non-reciprocity that drives al

What would settle it

Run a direct numerical comparison between the particle ODE system (23) with finite N and small γ, and the mean-field PDE (5), using the same smoothed interaction function (8) and the initial data of Example 3.1. If the particle dynamics do not converge to the PDE as γ→0 and ρ→∞ (with the product γρ held fixed) for times up to T=2500, then the grazing-collision limit is invalid and the claimed stationary states, including the interacting clusters, do not describe the proposed interaction rules.

Watch

Extended reading notes

Core claim

The central claim is that coupling topics through individual importance weights changes the qualitative behavior of opinion formation. In the microscopic rule (2), the distance controlling interaction on topic a is p_a(x,y,α)=β|x_a−y_a|+(1−β)Σ_b α_b|x_b−y_b|, so the 'closeness' that gates opinion change on one topic depends on all other topics via the person's own weights. When two people have different weight vectors α and η, the interaction is in general asymmetric: person x may be moved toward y while y is not moved toward x. In the mean-field limit this yields a Vlasov equation (5) whose stationary states include Dirac measures that are separated (no interaction) or interacting clusters

Load-bearing premise

The derivation of the mean-field PDE from the binary interaction rule assumes a 'grazing collision limit' with time rescaling, but the paper never specifies the scaling (e.g., interaction strength γ→0 and interaction rate ρ→∞) that would justify this limit for the deterministic rule (2), so the PDE and all its stationary states may not describe the actual microscopic process for finite γ.

Editorial extensions

If this is right

If the weights α are identical across the population, the model reduces to a symmetric multi-dimensional bounded-confidence system: the mean opinion is conserved and the variance never increases, so only consensus or separated clusters can be stationary.
If weights differ, the mean opinion can drift and the variance can increase, so a population can become more polarized over time without any external shock or network effect.
The construction of interacting clusters (Example 4.1 and Example 4.2) shows that stationary states can exist where different opinion groups continuously influence each other but their net effect cancels exactly—such states are possible only when weights differ.
The left-to-right swing simulation (Section 5.2) indicates that a minority who highly weights one topic can be pulled to the majority position on all topics, even when their initial opinions on other topics are opposed, providing a concrete mechanism for 'ideological alignment' across topics.
The bound on the number of separated clusters in two dimensions (Section 4.2.1) gives a quantitative prediction: for fixed interaction radius R and weights, only a finite, computable number of distinct opinion groups can coexist without interacting.

Reading between the lines

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

Editorial inference: The non-reciprocity mechanism might explain persistent minority influence or one-way 'echo chamber' effects: if one group weights a topic highly and another does not, the first can be pulled toward the second without the second moving, which could be tested in controlled behavioral experiments.
Editorial inference: The interacting-cluster examples suggest that stable multi-cluster societies do not require fragmentation into non-communicating groups; a direct extension would be to test whether such clusters are observable in agent-based simulations with finite N and small but non-zero interaction strength γ, where the exact cancellations might be replaced by slow drift.
Editorial inference: The mean-field equation (5) is deterministic and ignores stochastic fluctuations; a natural extension is to add noise and check whether the non-reciprocal drift survives, which would connect the model to empirical opinion surveys where measurement noise is universal.
Editorial inference: Since the change in one topic depends on all other topics, the model predicts cross-topic 'contagion'—e.g., a shift in climate opinion could drag sustainable-energy opinion—so a testable implication is that interventions on one issue should produce correlated shifts on other issues, with the correlation strength modulated by the population's weight distribution.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The paper is worth a look if you work on multi-dimensional opinion dynamics. The weighted distance pa in (1) and the resulting non-reciprocal binary interactions are genuinely new relative to the cited multidimensional HK/DeGroot extensions, and the interacting-cluster stationary states of Section 4.3 are a real addition—I don't know of those in earlier work. The existence proof is standard but correct, and the moment computations for equal and unequal importance weights give a clear picture of when the mean is conserved and when variance can grow.

My main concern is the grazing-collision limit. The paper defines a finite-γ binary interaction rule (2), then immediately jumps to the Vlasov equation (5) by 'taking the grazing collision limit and rescaling time, as for example in [23]'. No scaling is stated. To get (4) from (2) you need γ→0, ρ→∞ with ργ held fixed (or something equivalent); otherwise the term ργ in the weak form has no finite limit. With γ fixed—which the model explicitly allows—the neglected O(ργ²) terms need not vanish, and the stationary states of (5) may not describe the finite-γ dynamics. The simulations do not resolve this: they solve (23), which is the characteristic ODE of the limit equation, not (2). So the claims about mean drift, variance increase and the stationary structures are claims about the limit model, and the link to the suggested microscopic mechanism is unquantified.

This is not a fatal flaw if the paper is read as a model of the Vlasov dynamics. But then the presentation needs to say so explicitly, and the referee should ask for the scaling to be written out.

Other issues are smaller but real. Example 4.2 contains a duplicated Dirac term—two masses at (0,ε), one of which should presumably be at (0,−ε)—so the 'arbitrarily close' interacting cluster example is not correct as written. In Section 4.2.1, the inequality R/β ≥ R/((1−β)α1) is false for many values of β≥1/2 and α1 (e.g. β=0.7, α1=0.2), so the packing bounds in that subsection need rechecking. And the conclusion defers full characterisation of stationary states and convergence for unequal weights to future work, which softens the abstract's 'critically determined' claim. None of this suggests the main idea is wrong; it suggests the paper is not yet ready in its present form.

I'd engage with it after revision, and I'd send it to a serious referee now rather than desk reject it. The referee should focus on the scaling and on getting Example 4.2 and the packing bounds right.

Referee Report

3 major / 6 minor

Summary. The paper introduces a multi-dimensional opinion-formation model in which each agent is characterized by an opinion vector x∈[-1,1]^d and an importance-weight vector α∈A. Binary interactions are defined in (2), with the interaction strength on each topic depending on a weighted L1-type distance p_a (1) that uses the agents' importance weights. The kinetic equation (3) is formally reduced, via a grazing-collision limit, to the Vlasov-type mean-field equation (5). The paper proves existence of measure-valued solutions for Lipschitz interaction functions, mass conservation, non-negativity, non-increase of the component-wise support width, and moment dynamics: the mean is conserved and variance decreases when all agents share the same weights, while unequal weights can produce mean drift and variance increase. It constructs stationary states of types (S1)-(S3), including separated and interacting clusters, and presents simulations illustrating the effect of the distance function and of the importance weights.

Significance. If the mean-field reduction is accepted, the model is a genuinely novel and interesting contribution to multi-dimensional opinion dynamics. The weighted-distance coupling is natural, and the explicit examples of non-reciprocal interactions, mean drift, variance increase, and interacting clusters with Dirac masses at arbitrarily close locations are analytically checkable and clearly demonstrate qualitative differences from equal-weight and Euclidean-distance models. The paper is also honest about what it does not prove: the stationary-state classification is not claimed to be exhaustive, and the conclusion identifies full characterization as future work. The main limitation is that all analytical and numerical results are for the limit equation (5); the connection to the original finite-γ binary rule (2) is not established, which is precisely the load-bearing issue for the paper's central claim.

major comments (3)

[Section 2, between Eq. (3) and Eq. (4)] The derivation of the Vlasov equation is not specified. Starting from (2) with fixed γ∈(0,1), the Boltzmann-type equation (3) describes jump processes; to obtain (4) one must let γ→0 and ρ→∞ with ργ→λ (typically after a time rescaling) and verify that the O(ργ^2) remainders vanish. No scaling or convergence argument is given. Since every subsequent result — Theorem 3.2, Section 3.2, the stationary-state classification, the examples, and the simulations (23) — concerns the limit equation (5), the paper's central claim that the binary rule (2) produces the described complex stationary states is only established in an unquantified asymptotic regime. Please state the scaling explicitly and either prove the limit or explicitly restrict the claims to the mean-field model.
[Section 5, Eq. (23)] The numerical experiments approximate f by a sum of Dirac measures and solve the ODE system (23), which is the characteristic system of the Vlasov equation (5). This is not a simulation of the finite-γ binary interaction rule (2) introduced in Section 2. The figures therefore confirm the behavior of the limiting model, not of the microscopic model from which the paper starts. To support the connection, the authors should either simulate (2) directly and compare with the limit, or explicitly state that the computational study is for the mean-field equation (5).
[Theorem 4.1 and Corollary 3.4] The proof of Theorem 4.1 (and the proof of Corollary 3.4) differentiates xmin_a(t) and xmax_a(t), the extremal opinions of the support J_f(t). For measure-valued solutions these extremal values are not necessarily differentiable at every time, and bounding the velocity of a characteristic at an extremal point does not by itself control the evolution of the supremum/infimum of the whole support. The argument can likely be repaired with an epsilon-neighborhood argument or by tracking the flow map of all points, but as written the consensus theorem is not fully rigorous. This is a central result, so the proof should be completed.

minor comments (6)

[Example 3.1, p. 9] The claimed value μ1(T)=73/180 does not match the distribution (13); direct computation gives μ1(T)=157/180. The inequality −5/6−μ1(T)<0 and the conclusion that the variance increases remain correct, but the numerical value should be corrected.
[Corollary 3.4, proof] The final sentence of the proof says the maximum component-wise distance is 'non-decreasing', but the statement of the corollary and the argument show it is non-increasing. Please correct the typo.
[Example 4.2, Eq. (22)] The displayed stationary state contains two copies of δ((0,ε),(1,0)) and no δ((0,−ε),(1,0)); the subsequent calculations with S((0,±ε),...) indicate that one of the copies should be δ((0,−ε),(1,0)).
[Theorem 4.1, proof] The sentence 'If we split the interval [y,z] in half' refers to undefined variables y,z; it should be the interval [u,v] or [xmin_a(t), xmax_a(t)].
[Section 4.2.1] The terms 'upper bound' and 'lower bound' are used to describe estimates on the maximum number of clusters, but the phrasing is confusing because both are bounds on the same quantity. Please clarify which quantity is being bounded and in which direction.
[Eq. (9)] The notation X((x0,α), x) for the opinion component of the characteristic map is confusing, especially where it is written as 'X(x0,α)x'. Please use a clearer notation, e.g. X_opinion or a superscript, to distinguish the pair (x,α) from its first component.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the derivation is an asymptotic mean-field limit, and the analytical and numerical results test the model's own equations without fitted predictions.

full rationale

The paper's central claim is that the binary interaction rule (2) leads, in a standard grazing-collision/time-rescaling limit, to the Vlasov-type equation (5), and that this equation exhibits complex stationary states when agents have heterogeneous importance weights. I checked for each circularity pattern. The mean-field limit is taken 'as for example in [23]' — an external, standard kinetic-theory reference, not a self-citation, and the paper gives the explicit limiting weak form (4). The grazing limit itself is not fully specified (no explicit γ→0 and ρ→∞ scaling is written), but that is a rigor/completeness concern, not a circular one: (5) is not assumed to equal (2) by definition, and the analysis honestly concerns the limit equation. The stationary-state examples in Section 4 are constructed to satisfy the stationarity condition S(x,α,f∞)=0; they are existence demonstrations, not empirical predictions, and no parameters are fitted to a target result. The moment computations in Example 3.1 evaluate the model's own evolution equation for a chosen test distribution; this illustrates a consequence of the model rather than predicting an external quantity. The simulations solve the characteristic ODE (23) of the mean-field PDE, consistent with the analytical object under study; the paper does not claim to fit or predict external data. There are no load-bearing self-citations: reference [16] is cited for the explicit smoothed interaction function, which is written out, and [5] is cited only for a discrete analogue of a consensus result proved independently here. No uniqueness theorem is imported from the authors' prior work, and no known empirical pattern is merely renamed. The acknowledged omission — full characterization of stationary states and convergence in the heterogeneous-weight case — is explicitly listed as future work, not disguised as a result. Therefore I find no significant circularity and assign score 0; the main risks are analytical (unquantified asymptotic limit, possible arithmetic slip in Example 3.1), not circularity.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The importance weights α are part of the model state, not new entities; 'interacting clusters' describe stationary-state configurations, not newly postulated physical objects. The model parameters (β, r1, r2, γ, α choices) are hand-set to illustrate dynamics, not fitted to data.

free parameters (4)

β = 0.5 (in simulations)
Weights importance of the current topic vs the weighted rest; chosen by hand in all simulations (Tables 1 and 2).
r1, r2 = e.g., r1=2/5, r2=1/2 in §5.1; r1=11/12, r2=r1+0.0001 in §5.2
Bounds of the smoothed interaction function (8); chosen per experiment to control the confidence interval.
γ = not specified; implicitly set to 1 in the ODE system (23)
Interaction strength in (2); absent from (23) and the parameter tables, so effectively fixed to 1 in simulations.
importance weights α of simulated agents = e.g., (4/5,1/5), (1/2,1/2), (7/9,1/9,1/9)
Initial weights are hand-picked to showcase different dynamics; the paper's claim that weights determine final structures depends on these choices.

assumptions (5)

domain assumption The interaction function φ is Lipschitz continuous (A1).
Required for Picard-Lindelöf existence (Section 3.1); excludes the classical discontinuous bounded-confidence function, so the analysis applies to the smoothed version (8).
domain assumption The grazing-collision limit from the kinetic equation (3) to the mean-field PDE (4) is valid with suitable time rescaling.
Stated between Eq. (3) and Eq. (4) by reference to [23]; no scaling (e.g., γ→0, rate→∞) is given for the specific interaction (2).
domain assumption People's importance weights α are fixed in time and do not change during interactions.
Stated in Section 2: the state space is Q = I_d × A and α is carried through interactions unchanged.
domain assumption No exogenous factors (media, network) and no noise are present; all-to-all interactions.
Explicit model assumptions in Section 2.
standard math Standard measure-theoretic arguments: Prokhorov compactness, push-forward measures along characteristics.
Used in Section 3.1 and Theorem 3.2 without proof.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Multi-Dimensional Opinion Formation." pith.science (2026). https://pith.science/paper/NBN6HZIG

@misc{pith2026260104074,
  author       = {Pith},
  title        = {Pith review of: Multi-Dimensional Opinion Formation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NBN6HZIG}},
  note         = {Machine review of arXiv:2601.04074}
}

read the original abstract

In this paper we propose and investigate a multi-dimensional opinion dynamics model where people are characterised by both opinions and importance weights across these opinions. Opinion changes occur through binary interactions, with a novel coupling mechanism: the change in one topic depends on the weighted similarity across the full opinion vector. We state the kinetic equation for this process and derive its mean-field partial differential equation to describe the overall dynamics. Analytical computations and numerical simulations confirm that this model exhibits a variety of qualitatively distinct stationary states, and we demonstrate that the final opinion structures are critically determined by the people's opinion weights.

Figures

Figures reproduced from arXiv: 2601.04074 by the authors.

**Figure 1.** Inferaction radius defined by the pα-distance (1) for an interaction function ϕ with compact support on [0, R] Coming back to the computations done at the beginning in Section 4.2, we see that even when people have different importance weights, the following more general but also weaker corollary holds. Corollary 4.4. Let f∞(x, α) = X M ℓ=1 cℓδ(zℓ,αℓ)(x, α) (18) with M ∈ N, zℓ ∈ Id , αℓ ∈ A and cℓ > 0 for all ℓ ∈ {1… view at source ↗

**Figure 2.** Example of an interacting cluster discussed in Example 4.1 [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗

**Figure 3.** Interacting cluster (22), in which the Dirac measures are arbitrary close (see Example [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Initial distribution given by (24) and the corresponding stationary states illustrating the [PITH_FULL_IMAGE:figures/full_fig_p020_4.png]

**Figure 5.** Figure 5: Initial and final particle density illustrating the swing from ’left’ to ’right’ discussed in [PITH_FULL_IMAGE:figures/full_fig_p022_5.png]

**Figure 6.** Figure 6: marginal plots of solution to (5) for different values of [PITH_FULL_IMAGE:figures/full_fig_p023_6.png]

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

Treasure Search Optimization
math.OC 2026-07 conditional novelty 6.0 of 10

Treasure Search Optimization — explorers that keep searching plus a hunter that only teleports to improvements — has a well-posed conditional McKean-Vlasov limit and an O(1/α) stationary guarantee near the global minimizer.

Reference graph

Works this paper leans on

22 extracted references · 3 linked inside Pith · cited by 1 Pith paper

[23]

Kinetic models of opinion formation

Giuseppe Toscani. “Kinetic models of opinion formation”. In:Communications in Mathe- matical Sciences4.3 (2006), pp. 481–496. 25

2006
[1]

Opinion dynamics over complex net- works: Kinetic modelling and numerical methods

Giacomo Albi, Lorenzo Pareschi, and Mattia Zanella. “Opinion dynamics over complex net- works: Kinetic modelling and numerical methods”. en. In:Kinetic and Related Models10.1 (Nov. 2016). Publisher: Kinetic and Related Models, pp. 1–32.issn: 1937-5093

2016
[2]

Graphon mean field systems

Erhan Bayraktar, Suman Chakraborty, and Ruoyu Wu. “Graphon mean field systems”. In: The Annals of Applied Probability33.5 (2023), pp. 3587–3619

2023
[3]

Kinetic compartmental models driven by opinion dynamics: vaccine hesitancy and social influence

Andrea Bondesan, Giuseppe Toscani, and Mattia Zanella. “Kinetic compartmental models driven by opinion dynamics: vaccine hesitancy and social influence”. In:Mathematical Models and Methods in Applied Sciences34.06 (2024), pp. 1043–1076

2024
[4]

A kinetic approach to the study of opinion forma- tion

Laurent Boudin and Francesco Salvarani. “A kinetic approach to the study of opinion forma- tion”. en. In:ESAIM: Modélisation mathématique et analyse numérique43.3 (2009), pp. 507– 522.issn: 1290-3841

2009
[5]

A modified Hegselmann-Krause model for inter- acting voters and political parties

Patrick H. Cahill and Georg A. Gottwald. “A modified Hegselmann-Krause model for inter- acting voters and political parties”. In: (Feb. 2025). arXiv:2410.13378 [physics]

arXiv 2025
[6]

Mixingbeliefsamonginteractingagents

GuillaumeDeffuantetal.“Mixingbeliefsamonginteractingagents”.In:Advances in Complex Systems3.01n04 (2000), pp. 87–98

2000
[7]

Reaching a consensus

Morris H DeGroot. “Reaching a consensus”. In:Journal of the American Statistical associa- tion69.345 (1974), pp. 118–121

1974

Show all 22 references

[8]

BoltzmannandFokker–Planckequationsmodellingopinionformation in the presence of strong leaders

BertramDüringetal.“BoltzmannandFokker–Planckequationsmodellingopinionformation in the presence of strong leaders”. en. In:Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences465.2112 (Dec. 2009), pp. 3687–3708.issn: 1364-5021, 1471-2946

2009
[9]

Breaking consensus in kinetic opinion formation models on graphons

Bertram Düring et al. “Breaking consensus in kinetic opinion formation models on graphons”. In:Journal of Nonlinear Science34.4 (2024), p. 79

2024
[10]

Vector Opinion Dynamics in a Bounded Confidence Consensus Model

Santo Fortunato et al. “Vector Opinion Dynamics in a Bounded Confidence Consensus Model”.In:International Journal of Modern Physics C16.10(Oct.2005).arXiv:physics/0504017, pp. 1535–1551.issn: 0129-1831, 1793-6586. 24

2005 arXiv
[11]

OpinionDynamicsandBoundedConfidenceModels, Analysis and Simulation

RainerHegselmannandUlrichKrause.“OpinionDynamicsandBoundedConfidenceModels, Analysis and Simulation”. In:Journal of Artificial Societies and Social Simulation5 (July 2002)

2002
[12]

Kelley and Allan C

Walter G. Kelley and Allan C. Peterson.The Theory of Differential Equations: Classical and Qualitative. en. New York, NY: Springer New York, 2010.isbn: 978-1-4419-5782-5 978- 1-4419-5783-2

2010
[13]

Multi-dimensional Hegselmann-Krause dynamics

A. Nedić and B. Touri. “Multi-dimensional Hegselmann-Krause dynamics”. In:2012 IEEE 51st IEEE Conference on Decision and Control (CDC). 2012, pp. 68–73

2012
[14]

Dynamics of interdependent multidimensional opinions

S Noipitak and M A Allen. “Dynamics of interdependent multidimensional opinions”. en. In:Journal of Physics: Conference Series1719.1 (Jan. 2021), p. 012107.issn: 1742-6588, 1742-6596

2021
[15]

From classical to modern opinion dynamics

Hossein Noorazar et al. “From classical to modern opinion dynamics”. en. In:International Journal of Modern Physics C31.07 (July 2020). arXiv:1909.12089 [physics], p. 2050101. issn: 0129-1831, 1793-6586

2020 arXiv
[16]

Convergence, Control and Continuous Ageing in Opinion Dynamics

Andrew Nugent. “Convergence, Control and Continuous Ageing in Opinion Dynamics”. PhD thesis. University of Warwick, 2025. [17]ODE Solvers - DifferentialEquations.jl.url:https : / / docs . sciml . ai / DiffEqDocs / stable/solvers/ode_solve/(visited on 05/28/2025)

2025
[18]

Social network heterogene- ity promotes depolarization of multidimensional correlated opinions

Jaume Ojer, Michele Starnini, and Romualdo Pastor-Satorras. “Social network heterogene- ity promotes depolarization of multidimensional correlated opinions”. In:Physical Review Research7.1 (Feb. 2025). Publisher: American Physical Society, p. 013207

2025
[19]

An analytical formulation for multidimensional continuous opinion models

Lucía Pedraza et al. “An analytical formulation for multidimensional continuous opinion models”. In:Chaos, Solitons & Fractals152 (Nov. 2021), p. 111368.issn: 0960-0779

2021
[20]

Collective Dynamics of Belief Evolution under Cognitive Coherence and Social Conformity

Nathaniel Rodriguez, Johan Bollen, and Yong-Yeol Ahn. “Collective Dynamics of Belief Evolution under Cognitive Coherence and Social Conformity”. en. In:PLOS ONE11.11 (Nov. 2016). Ed. by Renaud Lambiotte, e0165910.issn: 1932-6203

2016
[21]

An agent-based model of multi- dimensional opinion dynamics and opinion alignment

Simon Schweighofer, David Garcia, and Frank Schweitzer. “An agent-based model of multi- dimensional opinion dynamics and opinion alignment”. In:Chaos: An Interdisciplinary Jour- nal of Nonlinear Science30.9 (Sept. 2020), p. 093139.issn: 1054-1500

2020
[22]

A Weighted Balance Model of Opinion Hyperpolarization

Simon Schweighofer, Frank Schweitzer, and David Garcia. “A Weighted Balance Model of Opinion Hyperpolarization”. In:Journal of Artificial Societies and Social Simulation23.3 (2020), p. 5.issn: 1460-7425

2020

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [23]

Kinetic models of opinion formation

Giuseppe Toscani. “Kinetic models of opinion formation”. In:Communications in Mathe- matical Sciences4.3 (2006), pp. 481–496. 25

2006

[2] [1]

Opinion dynamics over complex net- works: Kinetic modelling and numerical methods

Giacomo Albi, Lorenzo Pareschi, and Mattia Zanella. “Opinion dynamics over complex net- works: Kinetic modelling and numerical methods”. en. In:Kinetic and Related Models10.1 (Nov. 2016). Publisher: Kinetic and Related Models, pp. 1–32.issn: 1937-5093

2016

[3] [2]

Graphon mean field systems

Erhan Bayraktar, Suman Chakraborty, and Ruoyu Wu. “Graphon mean field systems”. In: The Annals of Applied Probability33.5 (2023), pp. 3587–3619

2023

[4] [3]

Kinetic compartmental models driven by opinion dynamics: vaccine hesitancy and social influence

Andrea Bondesan, Giuseppe Toscani, and Mattia Zanella. “Kinetic compartmental models driven by opinion dynamics: vaccine hesitancy and social influence”. In:Mathematical Models and Methods in Applied Sciences34.06 (2024), pp. 1043–1076

2024

[5] [4]

A kinetic approach to the study of opinion forma- tion

Laurent Boudin and Francesco Salvarani. “A kinetic approach to the study of opinion forma- tion”. en. In:ESAIM: Modélisation mathématique et analyse numérique43.3 (2009), pp. 507– 522.issn: 1290-3841

2009

[6] [5]

A modified Hegselmann-Krause model for inter- acting voters and political parties

Patrick H. Cahill and Georg A. Gottwald. “A modified Hegselmann-Krause model for inter- acting voters and political parties”. In: (Feb. 2025). arXiv:2410.13378 [physics]

arXiv 2025

[7] [6]

Mixingbeliefsamonginteractingagents

GuillaumeDeffuantetal.“Mixingbeliefsamonginteractingagents”.In:Advances in Complex Systems3.01n04 (2000), pp. 87–98

2000

[8] [7]

Reaching a consensus

Morris H DeGroot. “Reaching a consensus”. In:Journal of the American Statistical associa- tion69.345 (1974), pp. 118–121

1974