Sulle connessioni affini non-posizionali (Q2650251)

transformation equations of the general geometry of paths it is stipulated that the \(L^i_{hk}\) are homogeneous of degree zero in each of the directional arguments \(\underset{\alpha}{dx}\) (or \(\underset{\alpha}{\dot x})\). For the case \(m=0\) we obtain a non-Riemannian space (in the sense of Eisenhart); when \(m=1\) we have the type of space first considered by \textit{J. Douglas} [Ann. Math. (2) 29, 154--166 (1928)] and \textit{Knebelman} [Am. J. Math. 51, 527--564 (1929)]. Furthermore, when a metric is introduced, \(X_n\) becomes a Finsler space for the case \(m=1\). Denoting the symmetrical part of \(L^i_{hk}\) by \(\Gamma^i_{hk}\), the author obtains the curvature tensor \[ \Gamma^{\cdot\ \cdot\ i}_{h\,k\, l} = \partial_k \Gamma^i_{hl} - \partial_h\Gamma^i_{kl} + \Gamma^p_{hl}\Gamma^i_{kp} - \Gamma^p_{kl}\Gamma^i_{hp} + \overset{\alpha}{\partial_j} \Gamma^i_{lh}\partial_k \underset{\alpha}{\dot x^j} - \overset{\alpha} {\partial_j} \Gamma^i_{lk}\partial_h \underset{\alpha}{\dot x^j} \] when limiting himself to the symmetrical case. Here \(\partial_l \equiv \partial/\partial x^l\) and \(\partial_l^\alpha \equiv \partial/\underset{\alpha}{\partial \dot x^l}\). Even for the case of a Finsler space this tensor is not related to the curvature tensor introduced by \textit{E. Cartan} [Les espaces de Finsler. Paris: Hermann. 42 p. (1934; Zbl 0008.41805), Actual. Sci. industr. Nr. 79, , p. 35 seq. of the book] and, in the opinion of the reviewer, it seems to be better adapted to the study of such spaces rather than the latter. \( \Gamma^{\cdot\, \cdot\, i}_{h\,k\, l}\) can only be calculated at a point when not only the \(\underset{\alpha}{\dot x^i}\) but also their first derivatives are known at this point. Thus \(m\) fields of directions must be defined in the neighbourhood of this point. By means of integrability conditions the author obtains a specialised form \( \overset{*}{\Gamma}^{\cdot\, \cdot\, i}_{h\,k\, l}\) of \( \Gamma^{\cdot\, \cdot\, i}_{h\,k\, l}\), which is a generalisation of the curvature tensor introduced by Knebelman (loc. cit. p. 530) for the case \(m=1\). By restricting himself to this case in \S 4, the author introduces a metric function \(F(x, \dot x)\); and putting \(\Gamma^i(x, \dot x) = \tfrac14 \{\partial_s \dot\partial_j F^2\dot x^s - \partial_j F^2\} g^{ij}\), \((\dot\partial_j\equiv\partial/\partial i^j)\) (taking into account a printing error in the corresponding equation of the paper), he defines \(\Gamma^i_{hk} = \dot\partial_h \dot\partial_k \Gamma^i\). This leads to the covariant derivative of \textit{Berwald} [Math. Z. 25, 40--73 (1926)] and becomes in this case the curvature tensor as defined by Berwald. Finally the author discusses the connection between his covariant derivative and the one defined by the reviewer [ Math. Z. 54, 115--128 (1951; Zbl 0042.40403)].