A note on Burg's modified entropy in statistical mechanics (Q272095)

This paper offers a comparative study of the different entropy concepts, that of a) S. Carnot and R. Clausius, b) L. Boltzmann, c) C. E. Shannon, d) E. T. Jaynes and e) Burg and J. N. Kapur. It starts evidently with the phenomenological thermodynamic (classical) notion of S. Carnot and R. Clausius (a), according to which the system's thermal energy per unit temperature is showing that the energy of an isolated system is unavailable for doing useful work. The second law of thermodynamics states that in an isolated system the irreversible process always proceeds in the direction of the increasing entropy. In the history, the next step (b) was due to Boltzmann. This happened in the early steps of statistical mechanics (thermodynamics), stating that \(S=-\ln W\), where \(W\) is the thermodynamic probability. This concept was used only in a narrower sense, making use of a simpler concept of identical particles, think of the notion of microstates and macrostates of a system of identical particles, composing the medium (the gas). The third step (c) was that of C. E. Shannon in information theory, introduced to measure the information of a message, the measure of the uncertainty in different distribution of signs (in a message). The fourth (d) event in the history was the step of E. T. Jaynes, his maximum entropy principle, the remarkable difference here compared to the former efforts is the (sometimes infinite) number of constraints which relates to the respected components. The entropy expressions quoted hitherto are similar in the fundamental form: the entropy has a negative sign. The fifth approach is due to Burg, saying that \(S=^\Sigma_i\ln p_i\) (with \(p_i\) the probability of the \(i\)-th group of events). Remark the change of sign! The result of this change of sign was studied by \textit{J. N. Kapur} [Int. J. Math. Educ. Sci. Technol. 13, 693--714 (1982; Zbl 0503.90027)]. The present paper makes a comparative study to show the interesting features of this new concept and even its slight modifications. The illustrative part of this work shows that the main problems occur when one has to solve the simultaneous (transcendental) equations for the Lagrange multiplier of the eventual additional constaints. The result of the (approximative) calculations (by means of Lingo1) are shown in figures.