388 Energy Optimization in Process Systems p s (n + 1) -p s (n) = L s1 ~ ~ S n+1 (p 0 , ·p s ) S n+1 (p 0 , ·p s ) + L s2 ··· p 1 p 0 ~ S n+1 (p 0 , ·p s ) +L ss p s (10.37) as in classical irreversible thermodynamics. Therefore, the evolution processes can be imbedded into a relatively large family of thermodynamic processes. 10.8. CONCLUDING REMARKS By applying the tensor calculus, one can develop a discrete, non-linear represen- tation of evolution dynamics in metric spaces that may be curvilinear. Dynamic programming algorithms (Bellman's equations) can be derived and computer- aided simulations of their solutions can be performed. Systems governed by non-classical q-entropies may exhibit a quantitatively more distinct picture of instabilities than classical ones. Evolution of living organisms can be described in terms of variational principles for maximum generalized entropy along with suit- able transversality conditions. General gradient dynamics (in curvilinear spaces), which govern the evolution problems, are of Onsager's structure and consistent