Skimmed: about 30 of 262 pages, chosen for the record’s question. The review’s technical core (amplitude and phase equations, defects, pattern selection, chaos, and the system-by-system chapters) was not read, so there is no claims table.
Contribution
The review puts convection, Taylor–Couette flow, parametric waves, reaction–diffusion chemistry, biological pattern models, solidification and optics into one framework. Patterns are classified by the wave-number q₀ and frequency ω₀ of the linear instability of a uniform state, and described near onset by universal amplitude equations and further above it by phase equations. Throughout, the authors weigh theory against quantitative experiment.
Key insight
What unifies non-equilibrium patterns is not a thermodynamic principle but the shared mathematics of instabilities: near threshold, very different systems obey the same amplitude equation and differ only in its coefficients. Away from threshold there is no free energy to minimize, so which pattern appears is settled by boundaries, defects and the protocol, not by an extremum principle.
Assumptions
What the read sections take as their setting:
- Constant external conditions. The systems are held under constant non-equilibrium driving, with a control parameter R raised through a threshold R_c (§I.B).
- Dissipation. The authors consider “almost exclusively systems where dissipation is important” (fn. 1.2). Dissipative dynamical systems are those whose phase-space volumes contract onto attractors (§III.A.1).
- Determinism. Deterministic “microscopic” equations (Navier–Stokes, reaction laws) are the starting point. Noise is usually negligible in macroscopic experiments (§XIII.B(iii)).
- Large systems. The systems are large compared with the instability length 1/q₀, so boundaries are a perturbation (§I.B).
Key results
From the sections read:
- Classification (§I.B).
- Type I_s (q₀ ≠ 0, ω₀ = 0): stationary periodic patterns, such as convection rolls and Turing patterns.
- Type III_o (q₀ = 0, ω₀ ≠ 0): uniform oscillation, such as the stirred BZ reaction.
- Type I_o (q₀ ≠ 0, ω₀ ≠ 0): travelling or standing waves. Footnote 1.1 credits Turing (1952) with first linking macroscopic pattern formation to linear instabilities.
- Rayleigh–Bénard convection (§II.A). The control parameter is the Rayleigh number R = gαΔTd³/(κν). The conducting state goes unstable at R_c ≈ 1708, whatever the fluid, with q₀ of order 1/d. The finite wave-length comes from a conservation law: the fluid “cannot rise as a whole”.
- Bénard–Marangoni (§II.B.3). “The original experiment of Bénard (1900)” in an open dish gave hexagons. Rayleigh (1916) explained them by bulk buoyancy, but “it was later discovered (Pearson, 1958) that the hexagon pattern resulted from a surface instability caused by a temperature dependent surface tension”. The control parameter is the Marangoni number, which does not involve g.
- Reaction–diffusion (§II.E). In the linear two-species model (2.7) a stationary finite-wave-number instability needs the activator’s diffusion length (D₁/a₁)^½ to be shorter than the inhibitor’s (“local activation with lateral inhibition”). Large cross-coupling instead gives the uniform oscillation seen in BZ. “A closed chemical system, just as a closed fluid system, ultimately must come to equilibrium. Nonequilibrium phenomena of interest to us either occur as a transient … or in response to some external chemical pumping” (p. 865).
- Lyapunov functions (§I.B, §III.A.3). Near equilibrium a coarse-grained free energy selects the pattern, but “in general no such free energy (or other so-called Lyapunov potential) can be defined for nonequilibrium systems, though there are notable exceptions” (p. 857). Graham’s “nonequilibrium potential” exists for general systems but is singular and “does not determine the dynamics” (p. 868).
- BZ (§X.A.1). Belousov found the reaction in 1951; Zaikin and Zhabotinsky (1970) “already revealed the existence of chemical waves”. The Oregonator is “a simplified model”. Closed pattern experiments “run down in finite time (typically less than 100 periods of oscillation)”, which open gel reactors fed at their rims overcame (§X.B.5, p. 1046).
- Turing patterns (§X.B.5). Stationary Turing structures were long sought and found only recently: Ouyang et al. (1989, 1991) with an imposed gradient; Castets et al. (1990) in a gel, from differing diffusivities; and Ouyang and Swinney (1991), without macroscopic gradients, giving stripes and hexagons in the chlorite–iodide–malonic-acid reaction (Fig. 87).
- Morphogenesis (§XI.A). “The chemical identification of the morphogens has proved elusive, so that the theories are essentially phenomenological” (p. 1050). The activator–inhibitor condition is necessary within (2.7) but “by no means a general necessary condition” for a type I_s instability.
- What has not been accomplished (§XIII.B, p. 1077).
- No “useful” extremum principle for non-equilibrium steady states, “useful” meaning one that replaces and simplifies the dynamics (fn. 13.1).
- They quote Glansdorff and Prigogine (1971, p. 108): “the search for a universal kinetic potential has proved to be unsuccessful”. Those authors proposed instead “more specialized variational principles” and a “universal evolution criterion” in the form of an inequality.
- Landauer: relative occupation of competing states depends on the trajectories joining them, not only on local properties.
- The maximum-entropy formalism has produced no new physical results on pattern formation known to the authors.
- Self-organized criticality is “not supported by our detailed studies”.
- On life (§XIII.C, p. 1080). “There is no evidence for the existence of any global minimization principles controlling the structure, except as a perturbative statement near threshold.” Such a principle “would make it easier to generalize” from laboratory patterns to biological complexity. That reaction–diffusion mechanisms carry molecular information to the cellular level is “plausible to us (although by no means demonstrated)”. Against Anderson (1981), they hold that spontaneously broken continuous symmetry, with its phase dynamics, generalized rigidity and topological defects, is “established” in non-equilibrium laboratory systems.
Concepts
- dissipative structures: one of three names (with “synergetics” and “self-organization”) for macroscopic spatial structure in steady state under constant non-equilibrium conditions, cited to Nicolis and Prigogine (LIT-524) (§I.B). The review does not use it as a technical term.
- type I_s / III_o / I_o instability: the classification by (q₀, ω₀) at threshold (§I.B).
- amplitude equation: the universal equation for the slow envelope of the unstable mode near threshold (§I.C).
- Lyapunov function: V with V(U₀) = 0 and dV/dt < 0 away from the attractor U₀, which exists for gradient systems (§III.A.3).
Connections
- Prigogine’s Nobel lecture (LIT-516), read. On the point the record cares about, they agree. The lecture restricts minimum entropy production to the strictly linear regime and offers only a sufficient stability criterion far from equilibrium. The review finds no global principle at all beyond threshold. The difference is emphasis. The lecture presents dissipative structures as “a new type of dynamic states of matter” with their own thermodynamics. The review treats them as instabilities of dynamical systems, uses the term once, and the phrase “entropy production” does not occur anywhere in the OCR text. On convection the review corrects the lecture: the lecture calls the buoyancy-driven Rayleigh problem the “Bénard instability”, but Bénard’s own cells were Marangoni convection.
- Glansdorff and Prigogine (LIT-537), Deferred. The review is the record’s only read source for that book’s p. 108 admission.
- Turing (LIT-535), read. Credited as the origin of the instability-based view (fn. 1.1) and of the morphogen mechanism (§XI.A). The first chemical realizations came 37–39 years later.
- BZ (LIT-531), read. The review’s BZ section and Zhabotinsky’s agree on mechanism and phenomenology. The review supplies the point that closed pattern experiments run down, which makes the open reactor the setting in which BZ patterns are dissipative structures in the steady-state sense.
- Emergence survey (LIT-021). NOTE-007 records that its history of emergence runs from Anderson’s “More is different” to Prigogine. The review’s closing section takes up a later Anderson paper (1981) that doubted broken symmetry means much away from equilibrium, and argues that for laboratory patterns it is “established”.
Bearing on the record
- LIT-192 (demarcation). The review gives the physicist’s reason the demarcation cannot come from the physics of self-organization. The same instabilities produce rolls, hexagons, spirals and Turing spots in fluids, gels and models of embryos, and there is no global principle that would single out one class of dissipative system as directed toward an end. Its closing doubt about using such structures as a first step toward life is the strongest read statement in the record that dissipative pattern formation is necessary at most, not distinctive.
- LIT-211 (“Dissipative Self-Organizing Systems”). The review’s remark that closed systems’ patterns are transients, and open ones’ depend on “external chemical pumping”, sharpens what a dissipative definition of life must add. An organism is not held away from equilibrium by an experimenter’s feed; whatever a definition says instead is the part that does the work.
- For the parallel filings on entropy-production principles: §XIII.B is a read, citable statement that, as of 1993, no general extremum principle for non-equilibrium steady states had produced physical results on pattern formation, maximum-entropy formalisms included.
- No ML instruction.
Limitations
- The review’s limits. By its own account it treats only states “related to modes appearing at linear instabilities” (§XIII.C). Dendrites, fractal growth and strong turbulence are outside it.
- This skim’s limits. The technical core was not read, so nothing here certifies the amplitude-equation results, the stability balloons or the chaos chapters. Quotations come from OCR.
Open questions
- Is there any non-perturbative principle selecting patterns far from threshold? The review says there is none known, and that finding one would most likely come from “continu[ing] investigating specific systems”.
- Does any of the post-1993 work on entropy-production principles meet the review’s standard of “useful”, that is, replace the dynamics rather than restate it? That is the question for the parallel filings.
Corrections
- none to a seeded skim (there was no seed)
- Request’s choice. The request offered Bénard (1900) or a standard review, and preferred this one if readable. It was readable, as a scanned copy in the Caltech Authors repository, and is filed. Citation details are right: Rev. Mod. Phys. 65(3):851–1112, 1 July 1993, DOI 10.1103/RevModPhys.65.851.
- Section numbering. The OCR shows the conclusion’s heading as “XII. CONCLUSION”, but the contents list it as XIII (“XII. Other Systems” precedes it). The text’s own cross-reference to “Sec. XII.C” for spin waves agrees with the contents, so the conclusion is §XIII.