PHYSICAL REVIEW E 86, 067101 (2012) Conservation-dissipation structure of chemical reaction systems Wen-An Yong * Zhou Pei-Yuan Center for Applied Mathematics, Tsinghua University, Beijing 100084, China (Received 31 August 2012; published 5 December 2012) In this Brief Report, we show that balanced chemical reaction systems governed by the law of mass action have an elegant conservation-dissipation structure. From this structure a number of important conclusions can be easily deduced. In particular, with the help of this structure we can rigorously justify the classical partial equilibrium approximation in chemical kinetics. DOI: 10.1103/PhysRevE.86.067101 PACS number(s): 82.20.w, 05.70.Ln, 82.40.Qt Chemical reactions are fundamental for many natural phenomena, ranging from the rusting of iron to cell cycling and from photosynthesis to apoptosis. Since J. Wei [1] proposed a set of axioms to characterize general chemical systems in 1962, there have been huge efforts in seeking a unified mathematical formulation of chemical reactions. Oster and Perelson [2], for example, presented a geometric structure by casting much of the classical theory of kinetics into the language of differential geometry in 1974. Very recently in Ref. [3] van der Schaft et al. derived a general graph-theoretic formulation, which is basically already contained in the innovative paper by Sontag [4]. Other general studies of the mathematical structure of chemical systems include those of Aris [57], Bowen [8,9], Coleman and Gurtin [10], Feinberg [1113], Horn [14], Horn and Jackson [15], Krambeck [16], Sellers [17], Shapiro and Shapley [18], Wei [19], Wei and Prater [20], and so on. See Refs. [21,22] for more references. In this report, we present a conservation-dissipation struc- ture of the chemical reaction equations. It will be seen that this new structure is different from all those mentioned above. Consider a reaction system with n s species participating in n r reversible reactions ν i 1 S 1 + ν i 2 S 2 +···+ ν in s S n s k if k ir ν ′′ i 1 S 1 + ν ′′ i 2 S 2 +···+ ν ′′ in s S n s (1) for i = 1,2,...,n r . Here S k is the chemical symbol for the kth species, the nonnegative integers ν ik and ν ′′ ik are the stoichiometric coefficients of the kth species in the i th reaction, and k if and k ir are the respective direct and reverse constants of the i th reaction. The reversibility means that both k if and k ir are positive. Denote by u k the concentration of the kth species S k . According to the law of mass action, the evolution of u k = u k (t ) obeys the ordinary differential equations (see, e.g., Ref. [22]): du k dt = n r i =1 (ν ′′ ik ν ik ) k if n s j =1 u ν ij j k ir n s j =1 u ν ′′ ij j . (2) * wayong@tsinghua.edu.cn Set u = (u 1 ,u 2 ,...,u n s ) T with the superscript T for the transpose. The free energy of the reaction system is defined as F (u) = n s k=1 (u k ln u k u k ln u k u k ) with u = (u 1 ,u 2 ,...,u n s ) T a constant state to be specified below. Clearly, the gradient of F = F (u) is ∂F ∂u = ( ln u 1 ln u 1 , ln u 2 ln u 2 ,..., ln u n s ln u n s ) T (3) and F (u) is convex with respect to u. The aforesaid conservation-dissipation structure is given in the following theorem, which will be proved at the end of this report. Theorem. Assume that the system described by Eq. (2) satisfies the principle of detailed balance: there are n s positive numbers, u i > 0, such that k if n s j =1 (u j ) ν ij = k ir n s j =1 (u j ) ν ′′ ij (4) for i = 1,2,...,n r . Then there is a symmetric and nonpositive definite matrix, S = S (u), defined for u with u i > 0(i = 1,2,...,n s ), such that the kinetic Eq. (2) can be rewritten as du dt = S (u) ∂F ∂u (5) and the null space of S (u) is span ( ν ′′ i 1 ν i 1 ′′ i 2 ν i 2 ,...,ν ′′ in s ν in s ) T ,i = 1,2,...,n r , which is independent of u with u i > 0(i = 1,2,...,n s ). Remark that the constant state u = (u 1 ,u 2 ,...,u n s ) T needs not be unique and is often not unique. The symmetry of S (u) is reminiscent of the celebrated Onsager reciprocal relation [23], but it is different from that due to the dependence of S (u) on u. The null space of S (u) is just the right null space of the n r × n s -stoichiometric matrix ν ′′ 11 ν 11 ν ′′ 12 ν 12 ··· ν ′′ 1n s ν 1n s ν ′′ 21 ν 21 ν ′′ 22 ν 22 ··· ν ′′ 2n s ν 2n s . . . . . . . . . . . . ν ′′ n r 1 ν n r 1 ν ′′ n r 2 ν n r 2 ··· ν ′′ n r n s ν n r n s and is just the orthogonal complement of the stoichiometric subspace [13]. Its independence on u has a clear physical 067101-1 1539-3755/2012/86(6)/067101(3) ©2012 American Physical Society