A hierarchical statistical mechanics approach: hyperstatistics and its applications

arXiv:2606.20735 · cond-mat.stat-mech, physics.atm-clus, physics.bio-ph, physics.data-an, physics.flu-dyn, physics.med-ph · Submitted 2026-06-17 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.

Kai: We're looking at the paper: "A hierarchical statistical mechanics approach".

Mira: Here's the gist of the paper. A general approach to treat systems with inherent non-Boltzmann-Gibbsian behaviour has been proposed and applied to various physical phenomena, including capacitor discharge and brain dynamics <ref:2606.20735#pg2> Ground of Hyperstatistics The concept of temperature measurement is not a property of a specific point but rather an average over a comprising region of the system of interest <ref:2606.20735#pg2> In the frame of hyperstatistics, a distribution of Boltzmann factors within a given domain gives rise to a q-exponential (q-exp) function expressed in terms of an average inverse temperature ⟨βi⟩ <ref:2606.20735#pg2> This occurs because when hypothetically measuring the temperature at a particular point, e.g., at the center of a given domain, the measured value corresponds to an average over the slightly different temperatures in its surroundings, yielding the value ⟨βi⟩ <ref:2606.20735#pg2> The q-generalized effective Boltzmann factor for the entire system Bq(E) = R ∞ 0 expq (−⟨βi⟩E)f(βi)dβi simply results in Bq(E) = expq (−⟨β⟩E)

6: <ref:2606.20735#pg2> Three Scales of Treatment Hyperstatistics enables the treatment of a system in three distinct size scales: sub-domains, domains, and the whole system <ref:2606.20735#pg2> 1.

Kai: A γ-distribution

7: of the canonical exponential weight associated with a corresponding physical quantity of interest within a single domain is considered <ref:2606.20735#pg2> 2.

Mira: The Laplace transform of the γ-distribution, being the Boltzmann factor its kernel, corrects the Boltzmann factor itself, enabling us to transit from a Boltzmann-Gibbsian to a hyperstatistics space <ref:2606.20735#pg2> 3.

Kai: This correction results in a q-exp function in terms of ⟨β⟩, cf.

Title and authors: Kai: So the main point they are making is that hyperstatistics lets you look at a system by breaking it down into these three scales, which is microscopic, mesoscopic, and macroscopic <ref:2606.20735#pg3>. They start by considering a gamma distribution of the canonical exponential weight for something within a single domain <ref:2606.20735#pg3>.

Mira: And then they take the Laplace transform of that gamma distribution, which acts as the kernel for the Boltzmann factor itself, and this correction allows them to transition from standard Boltzmann-Gibbsian statistics into this hyperstatistics space <ref:2606.20735#pg2>.

Lev: If we translate that to what we build on hardware, it means that instead of a single equation for the whole system, you’re dealing with a family of equations linked by these averages across different scales <ref:2606.20735#pg3>.

Kai: And this process results in a q exponential function expressed in terms of the average alpha beta, which is what they call the q generalized effective Boltzmann factor, Bq E equals exp q minus alpha beta E f beta i summed over all beta i <ref:2606.20735#pg2>.

Mira: That looks like they are modifying how we calculate the average energy of a system by incorporating this non-extensive parameter q, which is derived from these underlying distributions <ref:2606.20735#pg1>.

Lev: So for us, that means the calculations become more complex because you have to handle this extra parameter q and the averaging across all those different scales simultaneously <ref:2606.20735#pg3>.

The paper's summary: Kai: The paper also points out some specific ways they can improve the original approach, which is what they call their improvements <ref:2606.20735#pg1>. They suggest that they can treat a γ distribution of the canonical exponential weight for a physical quantity within a single domain <ref:2606.20735#pg3> differently based on how you look at it.

Mira: They say that the Laplace transform step, which corrects the Boltzmann factor, is important because it's what enables us to transit from Boltzmann-Gibbsian to hyperstatistics space <ref:2606.20735#pg2>.

Lev: From a hardware perspective, one improvement they suggest is that you could look at how this correction relates to things like velocity autocorrelation functions in Brownian motion <ref:2606.20735#pg1>. They mention obtaining specific values for q, like one point four two nine and one point six eight four, from simulations of Brownian motion <ref:2606.20735#pg1>.

Kai: Those specific numbers for q are interesting because they show how the decay of the parallel component is slower than the perpendicular one in Brownian motion <ref:2606.20735#pg1>. That’s a concrete physical finding tied to this statistical method.

Mira: And they also suggest that hyperstatistics can be used as an emergent, macroscopic statistical descriptor for dynamics coming from lower-level neural models <ref:2606.20735#pg1>. This moves us away from needing to know the exact molecular details of ion channels or synapses <ref:2606.20735#pg1>.

Lev: That’s a big deal for us because if we can use q to characterize these neural dynamics, it means we don't necessarily need a super detailed microscopic model to get the macroscopic behavior right <ref:2606.20735#pg1>.

The paper's improvements: Kai: So to wrap up on "A hierarchical statistical mechanics approach: hyperstatistics and its applications", the core idea is using this three-scale treatment—sub-domains, domains, and the whole system—to describe systems that are inherently non-Boltzmann-Gibbsian <ref:2606.20735#pg1>.

Mira: They’ve shown how this framework allows us to derive a q exponential function from the averaging over temperatures in surrounding regions, moving beyond standard statistics <ref:2606.20735#pg1>.

Lev: For practical implementation, the main challenge is that you need to figure out how to map those microscopic details onto this macroscopic q parameter consistently across all those scales <ref:2606.20735#pg3>.

Kai: They also show that this Q parameter can quantify how an integrated system behaves when you couple microscopic events across mesoscopic domains and express that behavior at a higher level <ref:2606.20735#pg1>.

Mira: And the paper suggests using these q values to look for deviations from optimal complexity, where values near a critical maximum reflect this optimum, but lower or altered q patterns might point to subcritical or supercritical states <ref:2606.20735#pg1>.

Lev: I think the biggest implication for us is that this gives us a compact bridge between microscopic biophysics and clinical brain states <ref:2606.20735#pg1>. It’s a way to use these statistical descriptors without getting bogged down in every single molecular interaction <ref:2606.20735#pg1>.

Kai: Exactly. Future work they suggest involves systematically comparing q and n across simulated neural systems, empirical recordings like EEG or MEG, and conditions like anesthesia or schizophrenia using standardized avalanche detection protocols <ref:2606.20735#pg1>.

Mira: It seems like a framework that could be used to analyze complex dynamics in many different physical systems beyond just brain dynamics <ref:2606.20735#pg1>.

Lev: That’s the path forward—using this statistical mechanics approach to probe the dynamics of things we can't easily model at the molecular level, and seeing where those deviations lead us next.

Conclusion: Kai: So we’ve been looking at this paper, "A hierarchical statistical mechanics approach: hyperstatistics and its applications," which basically tries to use a three-scale treatment—microscopic, mesoscopic, and macroscopic—to describe systems that aren't standard Boltzmann-Gibbsian <ref:2606.20735#pg1>.

Mira: Right, it’s about moving beyond those standard statistics when we look at things like turbulent systems or brain dynamics <ref:2606.20735#pg1>. The authors are proposing this idea where temperature measurement isn't just a point thing but an average over a whole region <ref:2606.20735#pg1>.

Lev: From my side, it’s interesting how they propose using the Laplace transform of that gamma distribution to correct the Boltzmann factor itself <ref:2606.20735#pg2>. That correction is what lets you transition into this hyperstatistics space <ref:2606.20735#pg1>.

Kai: And that correction leads directly to a q exponential function expressed in terms of the average beta, which they call the q generalized effective Boltzmann factor <ref:2606.20735#pg1>. It’s a way to modify how we calculate the average energy of the whole system by incorporating this non-extensive parameter q <ref:2606.20735#pg1>.

Mira: That q parameter is what they use to quantify how the integrated system behaves when you couple microscopic events across mesoscopic domains and express that behavior at a macroscopic level <ref:2606.20735#pg1>. It’s really about finding a statistical descriptor that doesn't depend on specific molecular details like ion channels <ref:2606.20735#pg1>.

Lev: For real hardware, running this would mean you have to handle the averaging across those different scales simultaneously <ref:2606.20735#pg3>. It adds a layer of complexity to the error correction because you’re tracking this q parameter <ref:2606.20735#pg1>.

Kai: They also showed some concrete results with velocity autocorrelation functions in Brownian motion, where they found specific values for q, like one point being one point four two nine and another being one point six eight four <ref:2606.20735#pg1>. That shows how this mathematical framework connects to real physical processes in motion <ref:2606.20735#pg1>.

Mira: And the application to brain dynamics is where it gets really interesting, suggesting that q could be an index for dynamical regimes near criticality <ref:2606.20735#pg1>. Values near a critical maximum reflect optimal complexity, while deviations might point to subcritical or supercritical pathological states <ref:2606.20735#pg1>.

Lev: That’s a useful way to think about it, linking the statistical mechanics back to clinical states <ref:2606.20735#pg1>. It gives us a way to look at large-scale data from things like EEG or fMRI through this lens <ref:2606.20735#pg1>.

Kai: So, in summary, this paper lays out a novel statistical mechanics framework using hyperstatistics to treat complex systems across three scales and deriving q parameters that can help us diagnose dynamical states in things like the brain <ref:2606.20735#pg1>.

Mira: It’s a solid way to bridge the gap between microscopic biophysics and clinical observations, but it really depends on those underlying assumptions about how the system is structured across those scales <ref:2606.20735#pg1>.

Lev: I just think we need to see how this scales up when you try to run these simulations on actual hardware without the overhead landing too heavy <ref:2606.20735#pg3>.

Kai: Exactly, and next time we look at a paper like this, we’ll be checking if those q parameters hold up when you compare them against other experimental data sets like those from anesthesia or schizophrenia <ref:2606.20735#pg1>.

IGCE – Physics Department, Sãao Paulo State University (Unesp) · Laboratory of Experimental Surgery (LIM26), Medical School, University of Sêo Paulo

cond-mat.stat-mech, physics.atm-clus, physics.bio-ph, physics.data-an, physics.flu-dyn, physics.med-ph

Submitted: 2026-06-17

Updated: 2026-11-09

Comments: 21 pages, 7 figures, 2 tables, typos fixed, a new section about hyperstatistics and possible brain criticality was included in this version

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 74/100

The gist: A general approach to treat systems with inherent non-Boltzmann-Gibbsian behaviour has been proposed and applied to various physical phenomena…

Key concepts

Hyperstatistics
A statistical method that treats temperature not as a fixed point but as an average over a region. It uses this averaging to derive generalized statistical functions, like the q-exponential function, which describes systems exhibiting non-Boltzmann-Gibbsian behavior.
q-exponential (q-exp)
A special function that arises when temperature is averaged across a domain. It replaces the standard exponential in statistical mechanics. This function is expressed using an average inverse temperature, $\langle\beta\rangle$, and captures the non-standard statistical behavior of complex systems.
Three Scales of Treatment
Hyperstatistics allows treating a system hierarchically across three levels: small sub-domains, larger domains, and the entire system. This multi-scale approach enables a unified description that connects microscopic details to macroscopic dynamics.
q parameter in Brain Dynamics
The q parameter quantifies how the integrated system behaves when microscopic neural events are coupled. Values near a critical maximum suggest optimal complexity, while deviations indicate pathological states like subcritical or supercritical conditions.

Terminology

Summary

A general approach to treat systems with inherent non-Boltzmann-Gibbsian behaviour has been proposed and applied to various physical phenomena, including capacitor discharge and brain dynamics <ref:2606. p. 020735#pg2>

Ground of Hyperstatistics

The concept of temperature measurement is not a property of a specific point but rather an average over a comprising region of the system of interest <ref:2606.20735#pg2> In the frame of hyperstatistics, a distribution of Boltzmann factors within a given domain gives rise to a q-exponential (q-exp) function expressed in terms of an average inverse temperature ⟨βi⟩ <ref:2606. p. 120735#pg2> This occurs because when hypothetically measuring the temperature at a particular point, e. p. 1g., at the center of a given domain, the measured value corresponds to an average over the slightly different temperatures in its surroundings, yielding the value ⟨βi⟩ <ref:2606. p. 120735#pg2> The q-generalized effective Boltzmann factor for the entire system Bq(E) = R ∞ 0 expq (−⟨βi⟩E)f(βi)dβi simply results in Bq(E) = expq (−⟨β⟩E) [6] <ref:2606. p. 120735#pg2>

Three Scales of Treatment

Hyperstatistics enables the treatment of a system in three distinct size scales: sub-domains, domains, and the whole system <ref:2606.20735#pg2>

  1. A γ-distribution [7] of the canonical exponential weight associated with a corresponding physical quantity of interest within a single domain is considered <ref:2606. p. 220735#pg2>

  2. The Laplace transform of the γ-distribution, being the Boltzmann factor its kernel, corrects the Boltzmann factor itself, enabling us to transit from a Boltzmann-Gibbsian to a hyperstatistics space <ref:2606. p. 320735#pg2>

  3. This correction results in a q-exp function in terms of ⟨β⟩, cf. p. 8 Table 1 <ref:2606. p. 020735#pg2>

Velocity Autocorrelation Function Analysis

The velocity autocorrelation function (VACF) in Brownian motion exhibits a characteristic temporal decay dictated by a power-law in terms of time for scales smaller than the propagation time <ref:2606.20735#pg2> The obtained values of q, namely 1. p. 5429 and 1.684, respectively for the Fig. p. 5 2 Velocity autocorrelation function in Brownian motion <ref:2606. p. 420735#pg2> The decay of the parallel component is slower than the perpendicular one <ref:2606. p. 620735#pg2>

Application to Brain Dynamics

Hyperstatistics can be regarded as an emergent, macroscopic statistical descriptor of dynamics generated by lower-level neural models <ref:2606.20735#pg2> The q parameter may quantify how the integrated system behaves once microscopic neuronal events are coupled across mesoscopic domains and expressed at the macroscopic level <ref:2606. p. 620735#pg2> Values near the critical maximum reflect optimal complexity, whereas deviations toward lower or altered q-patterns may indicate subcritical or supercritical pathological states <ref:2606. p. 720735#pg2> This approach offers a compact bridge between microscopic biophysics, computational emulation, and clinical brain states <ref:2606. p. 720735#pg2>

Conclusions and Perspectives

The hyperstatistics approach is a novel statistical mechanics framework for describing complex systems <ref:2606.20735#pg2> Future work should compare systematically q and n across simulated neural systems, empirical EEG/MEG/fMRI recordings, anesthesia, coma, schizophrenia, psychedelic states, and other neuropsychiatric conditions using standardized avalanche-detection and fitting protocols <ref:2606. p. 720735#pg2> The approach is compatible with the scale-separation problem in whole-brain emulation <ref:2606. p. 720735#pg2>

REFERENCES

[1] R.S. Manna, B. Wolf, M. de Souza, M. p. 0 Lang, High-resolution thermal expansion measurements under helium-gas pressure, Rev. p. 8 Sci. Instrum. 83, 085111 (2012). p. 8 http://dx.doi/10.1063/1.4747272

[2] J.W. Valvano, Temperature measurements, Adv. p. 8 Heat Transf. p. 8 22, 359 (1992). p. 8 https://doi. p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8 p. 8org/10.1016/S0065-2717(08)70346-0

[3] Y. Okada, Y. Ando, R. Shimizu, E. p. 8 Minamitani, S. p. 8 Shiraki, S. p. 8 Watanabe, T. p. 8 Hitosugi, Scanning tunnelling spectroscopy of superconductivity on surfaces of LiTi2O4(111) thin films, Nat. p. 8 Commun. 8, 15975 (2016). p. 8 https://doi.org/10.1038/ncomms15975

[4] M. de Souza, A. p. 0 Br¨uhl, Ch. p. 8 Strack, B. Wolf, D. Schweitzer, M. p. 8 Lang, Anomalous Lattice Response at the Mott Transition in a Quasi-2D Organic Conductor, Phys. p. 8 Rev. Lett. 99, 037003 (2007). p. 8 https://doi.org/10.1103/PhysRevLett. p. 8 p. 8 p. 899.037003

[5] I.F. Mello, L. Squillante, G. p. 8O. Gomes, A.C. Seridonio, M-de Souza Griffithslike phase close to the Mott transition, J Appl Phys 128, 225102 (2020). p. 8 https://doi.org/10.1063/5.0018604

[6] L. Squillante, S. p. 8M. Soares, C. Tsallis, M-de Souza Hyperstatistics arXiv:2604. p. 824783 (2026). p. 0 https://doi.org/10.48550/arXiv. p. 82604.24783

[7] G Wilk, Z W lodarczyk Interpretation of the Nonextensivity Parameter q in Some Applications of Tsallis Statistics and L´evy Distributions Phys Rev Lett 84,2770 (2000). https://doi.org/10.1103/PhysRevLett.84.2770

[8] L Boltzmann, On the relationship between the second fundamental theorem of the mechanical theory of heat and probability calculations regarding the conditions for thermal equilibrium Kaiserlichen Akademie der Wissenschaften 76, 373 (1877).

[9] J W Gibbs, On the fundamental formula of statistical mechanics, with applications to astronomy and thermodynamics Proceedings of the American Association for the Advancement of Science 33, 57 (1884). p. 9

[10] C Beck, E G D Cohen Superstatistics Physica A 322, 267 (2003). p. 9 https://doi.org/10.1016/S0378-4371(03)00019-0

[11] R Descartes, Discours de la m´ethode pour bien conduire sa raison, et chercher la v´erit´e dans les sciences, Jan Maire (1637).

[12] A P Ricieri, Calcul fracionario transformada de Laplace e outros bichos Prandiano - Matem´atica aplicada a vida (1993). p. 9

[13] A La Porta, G. p. 9A. Voth, A M Crawford, J Alexander E Bodenschatz Fluid particle accelerations in fully developed turbulence Nature 409, 1017 (2001). p. 9 https://doi.org/10.1038/35059027

[14] B Abelev et al (ALICE Collaboration) Transverse momentum distribution and nuclear modification factor of charged particles in p-Pb collisions at √ sNN = 5.02 TeV Phys Rev Lett 110, 082302 (2013). p. 9 https://doi.org/10.1103/PhysRevLett.110.082302

[15] S Jeney, B Luki´c, J A Kraus, T Franosch, L Forr´o Anisotropic Memory Effects in Confined Colloidal Diffusion Phys Rev Lett 100, 240604 (2008). https://doi.org/10.1133/PhysRevLett. p. 8100.240604

[16] Mariano de Souza, Ricardo Paupitz, Antonio Seridonio, Roberto E Lagos Specific Heat Anomalies in Solids Described by a Multilevel Model Braz J Phys 46, 206 (2016). https://doi.org/10.1007/s13538-016-0404-9

[17] Hanshang Jin, Rahim R Ullah, Peter Klavins, Valentin Taufour Importance of anisotropic interactions for hard-axis or hard-plane ordering of Ce based ferromagnets Phys Rev B 112, 024437 (2025). https://doi.org/10.1103/99ww-msjh

[18] A M.-Chicharro, L W Votapka, R E Amaro, R C Wade Brownian dynamics simulations of biomolecular diffusional association processes WIREs Comput Mol Sci 13, e1649 (2022). p. 9 https://doi.org/10.1002/wcms.1649

[19] K Huang, I Szlufarska Effect of interfaces on the nearby Brownian motion Nat Commun 6, 8558 (2015). https://doi.org/10.1038/ncomms9558

[20] H J H Clercx, P P J M Schram, Brownian particles in shear fiow and harmonic potentials A study of long-time tails Phys Rev A 46, 1942 (1992). https://doi.org/10.1103/PhysRevA. p. 846.1942

[21] B U Felderhof, Effect of the Wall on the Velocity Autocorrelation Function and Long-Time Tail of Brownian Motion J Phys Chem B 109, 21406 (2005). https://doi.org/10.1021/jp051335b

[22] A L Hodgkin, A F Huxley, A quantitative description of membrane current and its application to conduction and excitation in nerve J Physiol 117, 500 (1952). https://doi.org/10.1113/jphysiol.1952.sp004764

[23] E M Izhikevich Simple model of spiking neurons IEEE Trans Neural Netw 14, 1569 (2003). https://doi.org/10.1109/TNN.2003.820440

[24] E M Izhikevich Which model to use for cortical spiking neurons IEEE Trans Neural Netw 15, 1063 (2004). https://doi.org/10.1109/TNN.2004.832719

[25] A Sandberg, N Bostrom Whole Brain Emulation: A Roadmap Future of Humanity Institute Oxford University Oxford (2008).

[26] E Tagliazucchi, D R Chialvo, M Siniatchkin, E Amico J F Brichant V Bonhomme et al Large-scale signatures of unconsciousness are consistent with a departure from critical dynamics J R Soc Interface 13, 20151027 (2016). p. 7 https://doi.org/10.1098/rsif.2015.1027

[27] T F Varley, O Sporns, A Puce, J Beggs Differential effects of propofol and ketamine on critical brain dynamics PLoS Comput Biol 16, e1008418 (2020). https://doi.org/10.1371/journal.pcbi.1008418

[28] G Alamian, T Lajnef, A Pascarella, J M Lina, R T Knight K Jerbi et al Altered brain criticality in schizophrenia new insights from magnetoencephalography Front Neural Circuits 16, 630621 (2022). https://doi.org/10.3389/fncir.2022.630621

[29] M M Schartner, R L Carhart-Harris, A B Barrett, A K Seth S D Muthukumaraswamy Increased spontaneous MEG signal diversity for psychoactive doses of ketamine LSD and psilocybin Sci Rep 7, 46421 (2017). https://doi.org/10.1038/srep46421

[30] W L Shew, H Yang, T Petermann, R Roy D Plenz Neuronal avalanches imply maximum dynamic range in cortical networks at criticality J Neurosci 29, 15595 (2009). https://doi.org/10.1523/jneurosci.3864-09.2009

[31] Jo˜ao R Martins, Victor Krivenkov, C Cesar R Bernardo Pavel Samokhvalov Igor Nabiev Yury P Rakovich and Mikhail I Vasilevskiy Statistical Analysis of Photoluminescence Decay Kinetics in Quantum Dot Ensembles Effects of Inorganic Shell Composition and Environment The Journal of Physical Chemistry C 126, 20480 (2022). DOI: 10.1021/acs.jpcc.2c06134

[32] Lucas Squillante, Samuel M Soares, Guilherme Lepski Mariano de Souza A few remarks on hyperstatistics Zenodo https://doi.org/10.5281/zenodo.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improve system modeling of complex systems by incorporating hyperstatistics to treat systems across microscopic, mesoscopic, and macroscopic scales. This allows for a unified statistical mechanical description that moves beyond standard Boltzmann-Gibbsian statistics for inherently non-Boltzmann-Gibbsian behavior in phenomena like turbulent systems or brain dynamics.

  2. Enable analysis of velocity autocorrelation functions in Brownian motion with greater accuracy by fitting the decay process to q-exponential functions, as demonstrated by the obtained values of q from experimental data sets. This allows for a more nuanced understanding of how hydrodynamic memory effects influence particle motion near boundaries compared to simpler power-law fits.

  3. Enhance computational neuroscience simulations by employing hyperstatistics to characterize collective dynamics through the q-generalized Boltzmann factor Bq, avalanche size distributions, and deviations from ordinary Boltzmann-Gibbs statistics. This permits the emergence of macroscopic statistical descriptors that are independent of specific molecular or structural interpretations of neural components like ion channels or synapses.

  4. Develop a framework for diagnosing pathological brain states by using the parameter q as a phenomenological index of dynamical regimes near criticality. By comparing q and n across simulated neural systems, empirical recordings, and clinical conditions like anesthesia or schizophrenia, the system can identify departures from optimal complexity that correspond to subcritical or supercritical states.

Sources

Related papers