In mathematics the concept of a vector can be made quite general and abstract with the idea of vector space. We define what is a vector space $V$ and we say that a vector is an element of a vector space $V$.

This concept is algebraic, and the important thing is the behavior of the operations performed on elements of $V$ inasmuch as it happens with groups, rings and fields. In that case, a vector space might include concepts quite far from the geometric idea of an arrow, including spaces of functions and much more.

On the other hand there is a quite particular case of this concept that is indeed related to geometry. That would be the tangent spaces to a smooth manifold $M$.

Given a smooth manifold $M$, for each $x\in M$, the tangent space $T_x M$ is a vector space comprising all directions tangent to $M$ at $x$. This implies that $T_x M$ is actually defined precisely so that we can picture its elements as arrows at $x$.

Physicists usually define vectors in a different way, with transformation laws. Quoting Arfken:

The set of $N$ quantities $V_j$ is said to be the components of an $N$-dimensional vector $\mathbf{V}$ if and only if their values relative to the rotated coordinate axes are given by

$$V_i'= \sum_{i=1}^N a_{ij}V_j, \quad i=1,2,\dots,N$$

As before, $a_{ij}$ is the cosine of the angle between $x_i'$ and $x_j$.

Now my question here is: when this standard definition is made, nothing is mentioned about what are the assumptions made over the space of all vectors.

Certainly it should be a vector space, however I ask: when physicists perform this standard definition of a vector, what is being defined is an element of a general vector space, that could be simply anything, or one has in mind exactly the tangent spaces to a smooth manifold?

Is this definition, defining a vector in a general vector space, or it is defining a vector that belongs to the tangent space to a manifold?


There are in fact several additional interpretations of tangent vectors. An equivalent definition of a tangent vector involves equivalence classes of curves.

The appropriate interpretation largely depends on the context of a problem. As QMechanic stated, we may take a linear algebra viewpoint, or view them in terms of differential geometry.

In the latter case, one has the bundle $TM \to^\pi M$ and sections are maps $s : M \to TM$ such that for the projection, $\pi \circ s = \mathrm{id}_M$. Thus a tangent vector is a section of $TM$. It can be viewed as,

$$TM = \bigcup_{p \in M}T_pM$$

that is, the union of all tangent spaces on the manifold. A tangent vector at a point $p$ lies in $T_pM$ which is a fibre of the bundle $TM \to^\pi M$.

This interpretation is useful in general relativity, wherein physics is cast in the language of differential geometry. On the other hand, if say, I were solving a problem in classical mechanics involving firing a cannonball, I would think of its velocity vector as simply some $v\in \mathbb R^3$.

  • $\begingroup$ Thanks @JamalS. I do know these definitions. Also here I'm not considering fields, so when I talk about smooth manifolds, I'm thinking about a vector at a point (an element of $TM$ rather than a section). What I want to know is: when the definition via transformation law that I stated is performed, Physicists are assuming that there is a background manifold? In other words, they are assuming that the vector being defined is an element of a tangent bundle of a manifold? Or they are not assuming anything like that, and thinking of a general vector space? Thanks again for the aid! $\endgroup$ – user1620696 Feb 25 '17 at 14:54

OP is essentially asking if tensors$^1$ in physics should be understood

  1. within the category of vector spaces and multilinear maps, i.e. linear algebra;

  2. or within the category of vector bundles and bundle maps, i.e. differential geometry?

The answer is: Both, depending on context. In the latter case, tensors are more properly called tensor fields.


$^1$ A vector is a (1,0) tensor.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.