WorksPhysics and Astronomy
Nonlinear Physics and Self-Organized Criticality in Astrophysics
Abstract
Since standard photon statistics is not applicable to nonlinear physics processes, it is of utmost importance to properly include the nonlinearity in astrophysical instabilities and emission processes. Bak et al.~(1987) postulated sandpile avalanches and called the underlying principle ``self-organized criticality'' (SOC), but its definition remained ambiguous and controversial. Here we discuss the nonlinear physics of six basic SOC definitions, such as: (i) phase transitions, (ii) fractal structures, (iii) power laws, (iv) SOC contingency, (v) exponential growth, and (vi) diffusive decay. Statistics of SOC events or avalanche processes can be inferred from their size distributions, or from their spatio-temporal evolution, described with the logistic equation. Here we quantify each nonlinear process in terms of quantitative scaling laws.
Introduction
The concept of self-organized criticality (SOC) processes has been introduced in statistical mechanics, such as phase transitions in terms of the Ising model ([18]), and other cellular automaton concepts, known as the Bak-Tang-Wiesenfeld (BTW) model ([12, 13, 25]). A schematic representation of the range of perception of SOC is shown in Fig. 1 of Watkins et al. [27], which illustrates four levels of SOC definitions: The most restrictive definition of a SOC process is the self-tuned phase transition, the second level is fractality, the third level is power law-like size distributions, and the fourth level is contingency of nature caused by SOC. Some of the most common nonlinear physics processes involve instabilities with an exponential growth phase, followed by a diffusive decay phase, which may be called “avalanche events”. The schematic shown in Table 1 indicates that the definition of avalanches or SOC events corresponds to at least six different nonlinear physics categories.
| PARAMETERS | NONLINEAR PHYSICS | EVENT STATISTICS |
| size distributions | phase transition<br>fractality<br>power laws<br>contingency | Self-organized criticality<br>(SOC) events |
| spatio-temporal evolution | exponential growth<br>diffusive decay | avalanche events |
Table 1. Nomenclature of nonlinear physics processes (adapted from Watkins et al. [27]).
The parameterization of SOC events is generally given in terms of a fitted power law slope () of their size distribution function, , while the parameterization of an avalanche event is generally available from the time profile of a single event. For instance, the SOC parameter of energy characterizes the power law slope , while the time evolution of each event reveals information on the exponential growth phase of an instability.
In this study we describe six processes that are relevant for the definition of nonlinear physics processes, which provide statistics of SOC events or avalanche events. In principle, each of the six nonlinear physics processes can be used as an individual SOC definition or model, while the combination of all six models attempts a more comprehensive approach.
Discussion
In the following we discuss the six nonlinear physics categories enumerated in Table 1. For each of the six SOC processes we provide a brief description of the underlying physics and a quantitative parameterization that can be used as a SOC definition for observed or predicted astrophysical phenomena.
Self-tuned phase transitions
Criticality refers to the behavior of extended systems as a phase transition where no characteristics scale exists. Thermodynamically, a phase transition occurs when there is a singularity in the free energy. The liquid-gas, conductor-superconductor, fluid-superfluid, or paramagnetic-ferromagnetic phase transition are common examples. The Ising model of a ferromagnet is one of the simplest models displaying the paramagnetic-ferromagnetic phase transition, that is, the spontaneous emergence of magnetization in zero external field as the temperature is lowered below a certain critical temperature. At the critical point of critical temperature and zero external field there is no characteristic scale (Christensen and Moloney [18]).
A definition of the Ising model can be given by a two-dimensional grid (in two dimensions), , where every site is occupied by a so-called spin . For a magnetic material, one may think of the spins as the magnetic dipoles positioned on the crystal structure lattice. In uni-axial magnetic materials, the magnetic dipole interactions constrain the spins to point in parallel () or anti-parallel direction (). In the Ising model, the total internal interaction energy is quantified by the coupling constants between the spins and , and the sum runs over all distant pairs of spins. In addition, the total energy is characterized by an external field, where the sum runs only over the nearest neighbors, and , so that the combined internal and external contributions add up to (Ising 1925),
where is the uniform external field and is a shorthand notation for the full microscopic description of a particular microstate.
The behavior of the Ising model as a function of the temperature, defined by the energy scale (with the Boltzmann’s constant ), the spin-spin interaction is relatively strong, so that the spin tends to align with one another, reducing the system’s internal energy. For temperatures , the spin-spin interaction is relatively weak, so that the spins are effectively non-interacting and point up and down randomly. The unpredictable aggregation of small and large magnetic (spin) domains in the Ising model became a paradigm of self-organized criticality. The Ising model was also applied to biology, neuroscience, economics, and sociology (see references in Christensen and Moloney 2005, chapter 2).
A theoretical prediction that could be tested is the fractal dimension in ferro-magnetic domains. If the fractal dimension is measured in 2-dimensional areas, the fractal (or Euclidean) dimension would be , while curvi-linear phenomena should show a fractal (or Euclidean) dimension of , or a mean value of the fractal dimension . As an example, a critical behavior of the Ising model on fractal structures has been found in dimensions between one and two [24].
Fractality
The seminal SOC paper [13] states the following role of fractal structures in SOC processes: We show that certain extended dissipative dynamical systems naturally evolve into a critical state, with no characteristic time or length scales. The temporal “fingerprint” of the self-organized critical state is the presence of flicker noise or noise; its spatial signature is the emergence of scale-invariant (fractal) structure [12, 13].
While the Euclidean relationship implies non-fractal areas, the fractal area can be expressed in terms of a power law index of the length scale , also called the Hausdorff dimension [20, 21, 22],
Likewise, while the Euclidean relationship implies non-fractal volumes, the fractal volume can be expressed in terms of a power law index of the length scale , also called the Hausdorff dimension,
There is no universal theory that predicts the values of the fractal dimensions, but some educated guess can be made by assuming the Euclidean dimension as upper limit, and one dimension less, , as lower limit. This yields the predictions,
and
Since the values and are observables, their statistical means and standard deviations can be compared with the theoretical predictions (Eq. 4, 5). Low values of the fractal dimension are characteristic for curvi-linear geometries (d=1), while high values of the fractal dimension indicate Euclidean geometries with solid filling (d=3). This dimensional diagnostic yields information on the spatial topology of astrophysical structures. Magnetic field lines tend to display 1-dimensional geometries, while 3-dimensional plasma structures reveal solidly-filled 3-dimensional volumes.
2-dimensional structures have been observed and theoretically modeled as quasi-linear separatrix surface layers in magnetic reconnection processes [e.g. 11]. Multi-fractal solar EUV intensity fluctuations were found to be able to discriminate between coronal heating models (e.g., [16]. An example of fractal structures is shown in Figure 1 that yields a mean value and standard deviation of , observed in a large solar flare. This is close to the theoretical prediction of [3]. Similar values were obtained from sunspots () and from active regions (), while photospheric and transition region phenomena reveal distinctly lower values ( and ) [10].

Figure 1. Measurement of the fractal area of a solar flare, observed by the spacecraft TRACE 171 Å on 2000 July 14, 10:59.32 UT. The Hausdorff dimension is evaluated with a box counting algorithm for pixels above a threshold of 20% of the peak flux value, yielding a mean of for the 8 different spatial scales (Aschwandenk and Aschwanden 2008a).
Power laws
Many avalanche processes have been interpreted in terms of “self-organized criticality (SOC) processes, which often display power law size distributions [17, 5, 6, 7, 8, 23, 26] (Aschwanden et al. 2016). Self-organized criticality phenomena ubiquitously display power laws in their size distribution functions, which can be analytically derived for each SOC parameter. We need only two fundamental assumptions: (i) the universal scale-free probability distribution function of length scales , which simply expresses a reciprocal relationship between the number density distribution and the fractal volume ,
and (ii) a scaling law relationship between the length scale and a SOC parameter, for instance the volume (),
which can be obtained from the fractal dimension , where are the Euclidean dimensions. A dimension is often used for three-dimensional data, but other dimensions can also be useful for fractal areas. In order to derive the power law of (fractal) volumes we invert the SOC parameter, , and calculate its derivative, , which yields,
and the desired power law index is,
which yields for and a value of . Similar derivations of power law indices can be obtained for other SOC parameters, such as for areas , fluxes , energies , or fluences [see 9]. An exhaustive list of power law slope measurements in energies and fluxes are compiled in a recent textbook, covering astrophysical phenomena such as solar and stellar flares, magnetospheric events, planetary systems, galactic and black-hole systems [9].
SOC Contingency
Watkins et al. [27] considered the contingency of nature caused by SOC as the most visionary perception of SOC phenomena. Generally, the aspect of contingency is used in the context of an event (i) that may occur but that is not likely or intended, (ii) a possibility that must be prepared for, or (iii) the condition of being dependent on chance (Wikipedia).
In the paper ”Complexity, Contingency, and Criticality” by Bak and Paczuski [14] it is stated that: Complexity originates from the tendency of large dynamical systems to organize themselves into a critical state, with avalanches or ”punctuations” of all sizes. In the critical state, events which would otherwise be uncoupled become correlated. The apparent, historical contingency in many sciences, including geology, biology, and economics, finds a natural interpretation as a self-organized critical phenomenon. These ideas are discussed in the context of simple mathematical models of sandpiles and biological evolution. Insights are gained not only from numerical simulations but also from rigorous mathematical analysis.
In practice, contingency may be used as a strategy to distinguish nonlinear physics from linear physics. Linear physics is traditionally applied to random processes, while nonlinear physics invokes a “fat tail” in the size distribution of SOC events. The statistics of event sizes in linear physics applications can be approximated with a Poissonian, an exponential, or a Gaussian size distribution, while nonlinear physics requires generally a power law-like distribution. The practical goal of any contingency measure is the estimation of the largest possible event. If the size distribution is characterized with a SOC power law slope , i.e., , the largest event (or contingency measure), is
where is the maximum volume of a SOC event, while corresponds to the maximum length scale. The maximum length scale is also called the finite system size, which is detectable by a steep cutoff in the size distribution. Of course, the maximum event of a SOC size distribution also depends on the number of events in the size distribution.
In order to give some examples, the maximum size corresponds to the size of continental tectonic plates in the case of earthquakes, i.e., km. In the case of solar (or stellar) flares, the size of (magnetic) active regions size amounts to m, [3, 4]. These limits can be considered as reliable contingency measures.
Exponenial Growth - Logistic Avalanches
An instability in astrophysics can be formulated with an exponential growth phase and a subsequent diffusion phase. The initial growth phase can be expressed with a first-order differential equation,
where is the time parameter, and is the exponential growth time scale, also known as e-folding time scale, and is the time evolution of an energy parameter.
In order to avoid exponential growth to infinity, which is an unphysical property, it is appropriate to introduce a factor of , which leads to the so-called logistic equation,
discovered by Piere Francois Verhulst in 1845 (Beltrami 1987, pp.61-64; Aschwanden et al. 1998). The logistic equation represents a generalization of the exponential growth curve (Eq. (11)), by including the growth limitation at the asymptotic saturation level . The solution of this differential equation is initially exponential (because the term is close to unity for ), and approaches asymptotically the value for large times. The physical basis of the logistic term is the constraint of an uppermost limit , which is also called carrying capacity in ecology sciences [e.g. 15, 2].
The exact solution of the first-order differential equation (Eq.12) is,
where is the reference time. We see that the time evolution of the energy exhibits a monotonic function in the range of (Figure 2a).
The most directly observed parameter of an astrophysical source is given in flux units , (in terms of physical cgs units [erg s]), while the energy is defined by the time integral of the flux, , or vice versa by the time derivative . The flux can be expressed in explicit form ,
This explicit form of the energy as a function of time is shown in Figure 2b, which ressembles to a Gaussian function. Herewith we have explicit functions of the enegy (Eq. (13)) and the flux (Eq. (14)), parameterized by three variables, i.e., the finite energy maximum and the time scale , as well as the reference time . The exponential growth dynamics predicts spatio-temporal functions for the energy and the flux , which can be tested for each avalanche event separately.
Four examples of logistic curve fits are shown in Figure 3, after subtraction of a linear background. The solar flare of 1991-May-31 featured in Figure 3 displays a hierarchy of “elementary” time profiles of the hard X-ray flux, which fit the logistic equation well, and thus underscore the nonlinear avalanche behavior during the rise time, and subsequent diffusion during the decay phase.
Figure 3. Examples of and keV hard X-ray time profiles, observed with BATSE/CGRO with 64-ms time resolution. Four sub-bursts at the beginning of the flare are enlarged (marked with the labels A, B, C, and D.) Note that the solar flare (1991, May 31) consists of significant sub-bursts with sub-second time scales, while the sub-bursts themselves show no significant sub-structures, and thus may represent time-resolved elementary structures of particla acceleration processes. Four of the sub-bursts are fitted with the logistic equation, after subtraction of a linear background each [2].
The property of exponential growth is also called coherent growth, in contrast to the case of incoherent growth, which does not exceed the random noise level. The predicted exponential growth phase represents the most obvious manifestation of self-organized criticality, especially because each SOC event amplifies nonlinear growth by several order of magnitudes, as expected for models of sandpile avalanches.
Diffusive Decay
As the previous description of avalanches governed by exponential growth implies, accumulation of flux or energy leads to a monotonic increase of energy during the rise phase and subsequent diffusion during the subsequent decay phase. It was found that the coefficient of turbulent diffusion changes in direct proportion to both the temporal and spatial scales [e.g. 1]. Statistics of SOC events can be separately obtained during the rise time and the decay time , since the two evolutions are likely to be dominated by two different nonlinear physics processes, i.e., exponential growth during the rise time, and diffusion during the decay time. Consequently, the time-integrated fluxes in the two time domains are expected to be correlated,
and
which yields the prediction,
If a proportionality between these two energy parameters is observed, the (nonlinear) physical processes are predicted to be incoherent, such as X-ray bremsstrahlung and gyrosynchrotron emission in some astrophysical SOC processes.
Conclusions
There is no unique definition of self-organized criticality (SOC), but we identify at least six nonlinear physics processes that may serve as a SOC definition and provide theoretical predictions. The various nonlinear physics categories complement (and overlap) with each other (see also Figure 3 in Watkins et al. [27]). We conclude that the various nonlinear physics categories can be used to test the agreement between the observed SOC phenomena and theoretical SOC predictions, following the schematic described in Table 1 and in the § 2.1 to § 2.6:
Self-tuned phase transitions: The Ising model predicts an initially randomly distributed grid of positive and negative spins to produce ordered magnetic domains with equal spin directions in a ferromagnet, after the temperature has self-tuned to a particular critical value. A theoretical prediction that could be tested is the fractal dimension in ferromagnetic domains, as defined in Eq. (4) for an Euclidean dimension .
Fractality: The Hausdorff fractal dimension in 2-dimensional space, , and in 3-dimensional space, , respectively, are predicted to have mean values of and , which can be tested from imaging data. For area parameters , each image pixel has a value or 0, while 3-D fluxes are obtained by including the line-of-sight component, . Deviations from the theoretically predicted fractal dimensions may reveal more complex topologies than the mean values of and .
Power laws: SOC processes predict not only the mathematical form of size distribution functions for each SOC parameter, , but also their specific power law slope value . The physical basis of size distribution functions is the universal scale-free probability distribution (Eq. (6)) and the choice of the fractal dimension , which depends on the dimensionality of the Euclidean geometry. For a dimensionality of , the mean fractal dimension is , and for a dimensionality of , the mean fractal dimension is . If we adopt the mean fractal dimension as a representative value for , the power law slope depends on the dimensionality as
SOC contingency: A contingency measure of SOC processes is the maximum value of length scales that are observed in a size distribution , which is also called the finite system size limit. The corresponding SOC parameter for the maximum volume follows from (Eq. (7)), i.e., .
Exponential growth - Logistic equation: The exponential growth phase of the spatio-temporal evolution of a SOC process predicts the time evolution of the energy (in normalized units , and ),
and the flux ,
which can be compared with the observed flux and energy values in Fig. (3). The histogrammed time profiles shown in Fig. 3 (in subpanels A, B, C, and D) represent the observed data , while the smoothed background-subtracted curves render the best fits of the logistic equation (Eq. 20). The exponential growth behavior of the logistic equation fit confirms the role of nonlinear instabilities most directly.
Diffusive decay: The energy during the growth phase and the energy during the diffusive decay phase may belong to two different physical processes. A linear regression analysis may reveal whether the two physical processes are correlated or not. In solar (or stellar) flares, the rise time is dominated by nonthermal hard X-ray emission, while the diffusive decay time is mostly produced by the conductive loss rate and the radiative loss rate. Fitting the exponential growth phase (with Eq. 20) yields the finite energy of a SOC avalanche and its time duration , but it is still an open question whether the diffusive decay phase is an dfferent SOC process or not. Nevertheless, the logistic equation satisfactorily fits the observed data during both the rise time and the decay time, which confirms the role of a nonlinear diffusion process for both the exponential growth and the diffusive decay phases.
These six conclusions demonstrate that information on at least six nonlinear physics processes is available that has not much been used in the context of astrophysical phenomena. At the same time, each of the six nonlinear physics categories can be considered as a possible definition of a SOC process. However, we should not restrict to a single definition, unless each of the six models can be explained with a single universal model. A complete SOC model should also be able to explain the differences in power law slopes of various astrophysical phenomena. Relevant efforts are in the works.
!Energy function plot labelled (a), with axes Time and Energy [arbitrary units] bbox=0.29,0.12,0.71,0.38
(a)
!Peak flux function plot labelled (b), with axes Time and Peak flux [arbitrary units] bbox=0.29,0.43,0.71,0.65
(b)
Figure 2. Exact solution of the energy function (a), and the (peak) flux function of an avalanche event (b).
References
- [1]Abramenko, V.I., Carbone, V. and Yurchyshyn, V. 2011, ApJ 743/2, article id. 133, 9pp, Turbulent diffusion in the photosphere as derived in the photospheric bright point motion
- [2]Aschwanden, M.J., Dennis, B.R., and Benz, A.O. 1998, ApJ 497, 972-993, Logistic avalanche processes, elementary time structures, and frequency distributions in solar flaresDOI
- [3]Aschwanden, M.J. and Aschwanden, P.D. 2008a, Solar flare geometries: 1. The area fractal dimension, ApJ 674, 530-543
- [4]Aschwanden, M.J. and Aschwanden, P.D. 2008b, Solar flare geometries: 2. The volume fractal dimension, ApJ 674, 544-553DOI
- [5]Aschwanden, M.J. 2011, Self-organized criticality in astrophysics. The statistics of nonlinear processes in the universe, Springer/PRAXIS: Berlin
- [6]Aschwanden, M.J. 2012, A&A 539, A2, A statistical fractal-diffusive avalanche model of a slowly-driven self-organized criticality system
- [7]Aschwanden, M.J. 2014, ApJ 782, 54, A macroscopic description of self-organized systems and astrophysical applications
- [8]Aschwanden, M.J. 2022, ApJ 934:33 (27pp), The fractality and size distributions of astrophysical self-organized criticality systemsDOI
- [9]Aschwanden, M.J. 2025, Power Laws in Astrophysics. Self-Organized Criticality Systems, Cambridge University Press: Cambridge.DOI
- [10]Aschwanden, M.J. and Nhalil, N.V. 2022, Interface region imaging spectrograph (IRIS) observations of the fractal dimension in the solar atmosphere, Frontiers in Astronomy and Space Sciences 9, id. 999319arxiv.org/abs/2207.12894
- [11]Aulanier,G., Pariat,E., and Demoulin,P. 2005, Current sheet formation in quasi-separatrix layers and hyperbolic flux tubes, AA 444, 961-976DOI
- [12]Bak, P., Tang, C., and Wiesenfeld, K. 1987, PhRvL 59/4, 381-384, Self-organized criticality: An explanation of 1/f noisepubmed.ncbi.nlm.nih.gov/10035754
- [13]Bak, P., Tang, C., and Wiesenfeld, K. 1988, PhRvL 38/1, pp.364-374. Self-organized criticality
- [14]Bak, P. and Paczuski, M. 1995, Complexity, contingency, and criticality, Proc. National Academy of Sciences of the United States of America 92, 6689-6696.pubmed.ncbi.nlm.nih.gov/11607561
- [15]Beltrami, E. 1987, Mathematics for dynamic modeling, Academic Press, Inc, Boston.
- [16]Cadavid, A.C., Rivera, Y.J., Lawrence, J.K., et al. 2016, ApJ 831/2, article id. 186, 18pp, Multi-fractal solar EUV intensity fluctuations and their implications for coronal heating modelsDOI
- [17]Charbonneau,P., McIntosh,S.W., Liu,H.L., Bogdan,T.J. 2001, Solar Phys. 203, 321-353, Avalanche models for solar flares
- [18]Christensen, K. and Moloney, N.R. 2005, Complexity and Criticality, Imperial College Press, London, UK.
- [20]Mandelbrot, B.B. 1977, Fractals: form, chance, and dimension, Translation of Les objects fractals, W.H. Freeman, San Francisco.
- [21]Mandelbrot, B.B. 1983, The fractal geometry of nature, W.H. Freeman, San Francisco.
- [22]Mandelbrot, B.B. 1985, Self-affine fractals and fractal dimension, Physica Scripta 32, 257-260DOI
- [23]McAteer,R.T.J., Aschwanden, M.J., Dimitropoulou, M., Georgoulis, M.K., Pruessner, G., Morales, L., Ireland, J., and Abramenko, V. 2016, SSRv 198 217-266, 25 Years of Self-Organized Criticality: Numerical Detection MethodsDOI
- [24]Monceau, P. and Perreau, M. 2001, Phys.Rev.B63, 184420, Critical behavior of the Ising model on fractal structures in dimensions between one and two: Finite-size scaling effectsDOI
- [25]Pruessner, G. 2012, Self-organised criticality. Theory, models and characterisation
- [26]Sharma,A.S., Aschwanden, M.J., Crosby, N.B., Klimas, A.J., Milovanov, A. V., Morales, L., Sanchez, R., and Uritsky, V. 2016, SSRv 198, 167-216, 25 Years of Self-Organized Criticality: Space and Laboratory PlasmasDOI
- [27]Watkins, N.W., Pruessner, G., Chapman, S.C., Crosby, N.B., and Jensen, H.J. 2016, SSRv 198, 3-44, 25 Years of Self-Organized Criticality: Concepts and ControversiesDOI
