In physics, a sign convention is a choice of the physical significance of signs (plus or minus) for a set of quantities, in a case where the choice of sign is arbitrary. "Arbitrary" here means that the same physical system can be correctly described using different choices for the signs, as long as one set of definitions is used consistently. The choices made may differ between authors. Disagreement about sign conventions is a frequent source of confusion, frustration, misunderstandings, and even outright errors in scientific work. In general, a sign convention is a special case of a choice of coordinate system for the case of one dimension.
In relativity, the metric signature can be either (+,−,−,−) or (−,+,+,+). (Note that throughout this article we are displaying the signs of the eigenvalues of the metric in the order that presents the timelike component first, followed by the spacelike components). A similar convention is used in higher-dimensional relativistic theories; that is, (+,−,−,−,...) or (−,+,+,+,...). A choice of signature is associated with a variety of names:
+ − − −:
− + + +:
We catalog the choices of various authors of some graduate textbooks:
The signature + − − − corresponds to the metric tensor:
whereas the signature − + + + corresponds to:
The Ricci tensor is defined as the contraction of the Riemann tensor. Some authors use the contraction , whereas others use the alternative . Due to the symmetries of the Riemann tensor, these two definitions differ by a minus sign.
In fact, the second definition of the Ricci tensor is . The sign of the Ricci tensor does not change, because the two sign conventions concern the sign of the Riemann tensor. The second definition just compensates the sign, and it works together with the second definition of the Riemann tensor (see e.g. Barrett O'Neill's Semi-riemannian geometry).
It is often considered good form to state explicitly which sign convention is to be used at the beginning of each book or article.