Different motivations and goals in the historical development of the theory of systems of linear inequalities (Q1862868): Difference between revisions

The theory of systems of linear inequalities is characterized as developed for several different reasons, in different mathematical, scientific, and social contexts, therefore having very fragmented history. At the end of the 19th century it was developed within the context of analytical mechanics by Julius Farkas, who proved the important theorem, today referred to as Farkas's lemma. He gave a solution method for a finite system of linear inequalities. (Before only Fourier seems to have had the idea of constructing such a theory, but did not get very far.) It is demonstrated how Farkas's main theorem arose from analytical mechanics, in the framework of Fourier's Inequality Principle. Approximately at the same time Hermann Minkowski published his book ``Geometrie der Zahlen'', where the study of convex bodies was initiated. In the present paper his theory of linear inequalities from the appendix of the first chapter of this book is discussed. About twenty years later Alfred Haar was the first to prove the results of Farkas and Minkowski using the theory of convexity. In the 1920's a theory of linear inequalities emerged in USA, at first independently of the earlier developments. The starting point was a small paper by William V. Lovitt in 1916 titled ``Preferential voting'' [Am. Math. Mon. 23, 363-366 (1916; JFM 46.0220.01)]. A year later Llyod L. Dines started his series of papers dealing directly with the theory of systems of linear inequalities, first for their own sake. He worked out the matrix treatment, introduced the concepts of \(I\)-rank, \(I\)-minor etc. The development of this subject was continued by Walter B. Carver, and then by Dines himself, from 1928 also outside the USA by Fujiwara in Japan. The first presentation, after Haar's papers, of a theory of linear inequalities, completely imbedded in the theory of convexity, appeared in 1935 in a paper by H. Weyl. A year later Theodore Motzkin published his thesis, which can be seen as a culmination of all the work done so far and eventually became a main source of inspiration. He gave an important generalization for Gordan's ``disguised'' transposition theorem from 1873, discussed profoundly in the present paper. The paper is concluded with a historical overview of the theory of linear inequalities in the context of game theory and linear programming in the postwar period in the USA (J. von Neumann, G. B. Dantzig, ONR project).