The primary purpose of this paper is to expose, in as simple and clear a form as is possible, the fundamentals of the geometric structure of a Riemannian space. It is a general truth that the methods which pierce most deeply into the heart of a geometric theory are invariant methods, that is, methods which are independent of the choice of the coordinates in terms of which the theory is expressed analytically. In the case of Riemannian geometry, these are the methods of tensor analysis. As important, perhaps, as the use of invariant methods is the expression of the analytic theory, so far as possible, in terms of invariant quantities alone. For it is in this form that the theory becomes most illuminating and suggestive. But, in ordinary tensor analysis, the components of a tensor are not invariants. A first step toward our goal will be, then, to introduce for Riemannian geometry an intrinsic tensor analysis, that is, a form of tensor analysis in which the components of all tensors are invariants. Any theory of the geometry of a Riemannian space presupposes that the space is referred to a certain ennuple of congruences of curves. In the ordinary theory, this ennuple consists of the parametric curves. In the intrinsic theory, it is an ennuple E, whose choice, as will presently be evident, is entirely arbitrary. The ordinary components of a tensor, that is, the components in the ordinary theory, are referred to the differentials of the coordinates xi pertaining to the ennuple of the parametric curves. The intrinsic components are referred to the differentials of arc, dsi, of the curves of the ennuple E. There is no need in the intrinsic theory of actual coordinates pertaining to the ennuple E; the differentials of arc dsi suffice. Accordingly, E can be chosen arbitrarily; it does not have to be a parametric ennuple, that is, an ennuple with which coordinates can be associated. It may be, and we shall ordinarily take it to be, an ennuple of general type, and hence, of course, not necessarily orthogonal. Intrinsic components, referred to E, of the covariant derivative of a tensor are made possible by the introduction of invariant Christoffel symbols in
Read full abstract