No label defined (Q4754963)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 3010468
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 3010468 |
Statements
1933
0 references
differential geometry
0 references
Seminar
0 references
Vektor- und Tensoranalysis
0 references
This collection of papers, published by the University of Moscow, deals essentially with three subjects, with the so-called subprojective spaces, with the integrability conditions of covariant differential equations and with tensors of which the components belong to a linear associative algebra. The text is written either in German or in French, with short abstracts in Russian. The papers take their origin from the seminar of Professor B. Kagan. They have been composed with care, go often into the details of the computation and are very readable. Contents: B. Kagan, Über eine Erweiterung des Begriffes vom projektiven Raume und dem zugehörigen Absolut. Tr. Sem. Vektorn. Tenzorn. Anal. 1, 12--101 (1933); H. Schapiro, Über die Metrik der subprojektiven Räume. ibid. 1, 102--125 (1933); P. Rachewsky (P. K. Rashevskiĭ), Caractères tensoriels de l'espace sousprojectif. ibid. 1, 126--142 (1933); G. B. Gurewitsch (Gurevich), Über einige Integralaufgaben der Tensoranalysis. ibid. 1, 143--195 (1933); J. Dubnow (Ya. S. Dubnov), Über Tensoren mit nichtskalaren Komponenten. ibid. 1, 196--222 (1933); J. Dubnow (Ya. S. Dubnov), Die Differentialgeometrie der Strahlenkongruenzen in tensorieller Darstellung. ibid. 1, 223--303 (1933). I. Subprojective spaces. This is the subject of the first three papers, the conception is due to B. Kagan. This author first defines a \(k\) times projective space \(P_n^k\) by the condition that it is an \(n\)-dimensional manifold of symmetric linear connection, which admits a set of \(n\) ``projective coordinates'' such that the paths (``geodesic lines'') lie in linear manifolds of \(k\) dimensions. A \(P_n^{n-1}\) is a projective space of \(n\) dimensions. The subprojective spaces are those \(P_n^{n-2}\) of which all \(\infty^{2(n- 2)}\) ``geodesic planes'' pass through a fixed point, they form a class of preferred character. Kagan's paper deals first with the \(P_n^{n-1}\) and then extends the results to the subprojective \(P_n^{n-2}\). The connection of the \(P_n^{n-1}\) is of the form \[ \Gamma_{kl}^i = E_{kl}^i + \delta_l^i p_k + \delta_k^i p_l, \] where the \(E_{kl}^i\) are the Christoffel symbols of a Euclidean connection, the \(p_i\) and \(\delta_l^i\) are an arbitrary vector and the unit tensor. In projective coordinates the \(E_{kl}^i\) vanish. The cases in which this manifold is Riemannian (or ``metrical projective'') depends on certain differential equations, of which the integrability conditions determine whether we find the Euclidean or the non-Euclidean geometries. A similar reasoning is now applied to the subprojective spaces. In projective coordinates \(x^i\) we get as necessary and sufficient condition \[ \Gamma_{kl}^i = f_{kl}x^i + \delta_l^i p_k + \delta_k^i p_l, \] the \(f_{ki}\) form a symmetric tensor. Now the question arises when the spaces are metrical. The answer depends on the nature of the integrability conditions of the equations \[ \frac{\partial g_{ik}}{\partial x^l} = f_{kl}g_i + f_{il}g_k + 2g_{ik}p_l + g_{il}p_k + g_{kl}p_i, \] where \(g_l = g_{li}x^i\). It is found that in this case both \(p_i\) and \(f_{kl}x^l\) are gradient vectors, they define two scalars \(\vartheta\) and \(\varphi\). The equations \(\vartheta = 0\), \(\varphi = 0\) give the two ``absolute surfaces''. For a \(P_n^{n-1}\) and projective coordinates \(\varphi\) is constant and the other absolute surface goes to infinity. When the integrability conditions are identically satisfied we call the subprojective spaces ``Euclidean''. There are three types, which are studied in detail. In the case that the integrability conditions are only satisfied by certain values of \(g_{ik}\), we get the ``non-Euclidean'' forms. To this case also belongs a conform-Euclidean type, studied by A. Buchholz. According to a theorem of Rachewsky, it is always possible to find in a subprojective space a special set of coordinates, the ``canonical coordinates'', for which \(p_l = 0\). Shapiro's paper takes up the study of subprojective spaces in these coordinates. This allows him to simplify many of Kagan's results. He finds the condition that such spaces are projective ``in the narrow sense'', that is Riemannian of constant curvature. Other results deal with Riemannian manifolds of the Einstein type and of the type ``with a center'', investigated by Weyl, and the theorem is established that a subprojective space of \(n\) dimensions can always be imbedded in a Euclidean space of \(n+1\) dimensions. Rachewsky's paper discusses the conditions that a metric given in arbitrary coordinates be subprojective. The conclusion is that the manifold must be conformal Euclidean, \[ R_{kl,ji} = g_{ik} T_{lj} + g_{lj} T_{ik} - g_{il}T_{jk} - g_{jk}T_{il} \] \[ \nabla_{[m}T_{i]j} = 0, \] with the additional condition that two non-constant scalar fields \(P\), \(g\) exist for which \[ T_{ij} = Pg_{ij},\quad P_i= \partial P/\partial x^i,\quad g = g(P). \] The differential equations for canonical systems of coordinates in terms of arbitrary coordinates are also found. II. Integrability conditions of covariant differential equations. This paper, by Gurewitch, deals with the equation \[ \nabla_{[l}w_{i_1i_2\cdots i_p]} = w_{[i_1i_2\cdots i_p} v_{k]}, \quad w_{i_1i_2\cdots i_p} = w_{[i_1i_2\cdots i_p]}. \] Methods are given to get the general solution for \(p = 1\), \(p = 2\), and for an arbitrary \(p\) the necessary and sufficient conditions are given for the existence of not identically vanishing solutions, at the same time solutions are given ``that probably are the general solutions''. The paper also gives the method of integration of \[ \nabla_kw_i = w_{[i}v_{k]} + u_{ik}, \quad w_i\text{ unknown}, \quad v_i\text{ and }u_{ik} = -u_{ki}\text{ given}. \] It would be of interest to compare these results with those of \textit{J. A. Schouten}, ``On the conditions of integrability of covariant differential equations'', Trans. Am. Math. Soc. 27, 441--473 (1925; JFM 51.0596.05), who also treats more general cases. III. Tensors belonging to a linear associative algebra. To this subject are devoted two papers by Dubnow, the first paper explaining the theory and giving applications to the differential geometry of surfaces in Euclidean three-space; the second paper dealing with the differential geometry of line congruences, also in Euclidean three-space. We shall confine ourselves to the explanation of the principle. Let \(\mathbf r\) be the radius vector of a point \(M(u^1, u^2)\) of a surface \(V_2\). Let \(\mathbf r_1\) and \(\mathbf r_2\) represent the derivatives of \(\mathbf r\) with respect to \(u^1\) and \(u^2\). These vectors are now considered as components of a field of covariant ``tensors'' \(\mathbf r_i\), given at the points \(M\) of the \(V_2\). This means that under the transition to new coordinates \('u^1\), \('u^2\) on \(V_2\) the vectors \(\mathbf r_1\), \(\mathbf r_2\) are replaced by the new pair according to the formulas \[ '\mathbf r_i = \mathbf r_\alpha \frac {\partial u^\alpha}{\partial 'u^i}\qquad i,\alpha = 1,2.\] If \(\mathbf r^1\), \(\mathbf r^2\) are the reciprocal vectors to \(\mathbf r_1\), \(\mathbf r_2\) in the tangential plane of \(V_2\) at \(M\) \[ \mathbf r^i \mathbf r_j = \delta_j^i, \] then we can take \(\mathbf r^1\), \(\mathbf r^2\) as components of a contravariant ``tensor'' \(\mathbf r^i\). A covariant differentiation can also be defined. Many formulas of ordinary surface theory become very simple in this new tensor notation. In the differential geometry of line congruences, to which the author devotes an extensive study of 77 pages, he can considerably improve, not only on Kummer, but also on Sannia. The theory now becomes that of two tensors in Dubnow's sense, a tensor \(b_{ij}\), and a tensor \(\beta_i\) related by a differential relation \[ \nabla_{(j}\beta_{i)} = b_{ij}.\] Many properties of line congruences are now discussed, also of special congruences (parabolic, Guichard, Bianchi, isotropic).
0 references