Tendency towards maximum complexity in a nonequilibrium isolated system Xavier Calbet Instituto de Astrofı ´sica de Canarias Vı ´a La ´ctea, s/n, E-38200 La Laguna, Tenerife, Spain Ricardo Lo ´ pez-Ruiz* Departamento de Fı ´sica Teo ´rica Facultad de Ciencias, Edificio A, Universidad de Zaragoza, E-50009 Zaragoza, Spain Received 30 January 2001; published 22 May 2001 The time evolution equations of a simplified isolated ideal gas, the ‘‘tetrahedral’’ gas, are derived. The dynamical behavior of the Lo ´pez-Ruiz–Mancini–Calbet complexity R. Lo ´pez-Ruiz, H. L. Mancini, and X. Calbet, Phys. Lett. A 209, 321 1995 is studied in this system. In general, it is shown that the complexity remains within the bounds of minimum and maximum complexity. We find that there are certain restrictions when the isolated ‘‘tetrahedral’’ gas evolves towards equilibrium. In addition to the well-known increase in entropy, the quantity called disequilibrium decreases monotonically with time. Furthermore, the trajectories of the system in phase space approach the maximum complexity path as it evolves toward equilibrium. DOI: 10.1103/PhysRevE.63.066116 PACS numbers: 89.75.Fb, 05.45.-a, 02.50.-r, 05.70.-a I. INTRODUCTION Several definitions of complexity, in the general sense of the term, have been presented in the literature. These can be classified according to their calculation procedure into two broad and loosely defined groups. One of these groups is based on computational science and consists of all definitions based on algorithms or au- tomata to derive the complexity. Examples are the logical depth 1, the -machine complexity 2, and algorithmic complexity 3. These definitions have been shown to be very useful in describing symbolic dynamics of chaotic maps, but they have the disadvantage of being very difficult to calculate. Another broad group consists of those complexities based on the measure of entropy or entropy rate. Among these, we may cite the metric or K -S entropy rate 4,5, the thermody- namic depth 6, the effective measure complexity 7, and the simple measure for complexity 8. These definitions have also been very useful in describing symbolic dynamical maps, the latter having been applied to a nonequilibrium Fermi gas 9. They suffer the disadvantage of either being very difficult to calculate or having a simple relation to the regular entropy. New definition types of complexity have recently been introduced. These are based on quantities that can be calcu- lated directly from the distribution function describing the system. One of these is based on ‘‘metastatistics’’ 10and the other on the notion of ‘‘disequilibrium’’ 11. This latter definition will be referred to hereafter as the Lo ´ pez-Ruiz– Mancini–Calbet LMCcomplexity. These definitions, to- gether with the simple measure for complexity 8described above, have the great advantage of allowing easy calcula- tions within the context of kinetic theory and of permitting their evaluation in a natural way in terms of statistical me- chanics. The disequilibrium-based complexity is easy to calculate and shows some interesting properties 11, but suffers from the main drawback of not being very well behaved as the system size increases, or equivalently, as the distribution function becomes continuous 12. Feldman and Crutchfield tried to solve this problem by defining another equivalent term for disequilibrium, but ended up with a complexity that was a trivial function of the entropy. Whether these definitions of complexity are useful in non- equilibrium thermodynamics will depend on how they be- have as a function of time. There is a general belief that, although the second law of thermodynamics requires average entropy or disorderto increase, this does not in any way forbid local order from arising 13. The clearest example is seen with life, which can continue to exist and grow in an isolated system for as long as internal resources last. In other words, in an isolated system the entropy must increase, but it should be possible, under certain circumstances, for the com- plexity to increase. In this paper we will examine how LMC complexity evolves with time in an isolated system and we will show that it indeed has some interesting properties. The disequilibrium-based complexity defined in Ref. 11actu- ally tends to be maximal as the entropy increases in a Bolt- zmann integrodifferential equation for a simplified gas. In Sec. II LMC complexity definition is reviewed. We proceed to calculate the distributions which maximize and minimize the complexity and its asymptotic behavior, and also introduce the basic concepts underlying the time evolu- tion of LMC complexity in Sec. III. Later, in Sec. IV, by means of numerical computations following a restricted ver- sion of the Boltzmann equation, we apply this to a special system, which we shall term ‘‘tetrahedral gas.’’ Finally, in Sec. V, the results and possible future lines of investigation are discussed, together with their possible applications. Ana- lytical and numerical demonstrations of the results of the numerical calculations for the tetrahedral gas are shown in the appendixes. *Present address: Area de Ciencias de la Computacion, Facultad de Ciencias, Edificio B, Universidad de Zaragoza, E-50009 Zara- goza, Spain. PHYSICAL REVIEW E, VOLUME 63, 066116 1063-651X/2001/636/0661169/$20.00 ©2001 The American Physical Society 63 066116-1