This is my understanding of a Cartesian tensor: Any vector in three-dimensional space can be written in a Cartesian system as $$\vec{A}=A^1\hat{e}_1+A^2\hat{e}_2+A^3\hat{e}_3$$ where $A^1, A^2, A^3$ are the components of $\vec{A}$ in a Cartesian basis $\hat{e}_1, \hat{e}_2, \hat{e}_3$. Under a rotation of the coordinate system, the components transform according to $$A^{\prime i}=\frac{\partial x^{\prime i}}{\partial x^j}A^j=R^i_{~ j} A^j$$ where $R$ is the orthogonal rotation matrix. This is a Cartesian vector.

What is not a Cartesian vector (or tensor, in general)?




I will expand the question after some time and will also give textbook references. I am also thoroughly confused by the terminology but it does exist.

  • 1
    $\begingroup$ 1. What reference did you get this notion of "Cartesian tensor" from? 2. I don't really understand what kind of answer this question is looking for: Things that aren't vectors in 3d space aren't "Cartesian" vectors, either. There are lots of things that aren't vectors. Where is this question coming from? $\endgroup$
    – ACuriousMind
    Nov 13, 2022 at 20:07
  • 1
    $\begingroup$ For an example of how things can break, see the Christoffel symbols. But in general, it's hard to answer this, as it's not clear why you think things might not be tensors. (To be clear, not all things are tensors, but it would help if you indicate what you're reading that made you come up with the question.) $\endgroup$ Nov 13, 2022 at 20:23
  • 2
    $\begingroup$ @ACuriousMind for example in Sakurai, page 234, revised edition: "The simplest example of a Cartesian tensor of rank 2 is a dyadic formed out of two vectors U and V." $\endgroup$
    – hyportnex
    Nov 13, 2022 at 20:23
  • $\begingroup$ It is strange terminology. Calling something a "Cartesian tensor" is almost paradoxical. The whole idea behind tensors is to have objects that preserve physical laws regardless of the coordinate system. So a "cartesian tensor" implies a preferred coordinate system - the antithesis of a tensor. $\endgroup$
    – joseph h
    Nov 13, 2022 at 22:33

2 Answers 2


It's meaningless, or wrong, to define a "Cartesian tensor" or a "Cartesian vector". You can define the components of a tensor or a vector in a Cartesian reference fame, or coordinate systems if you're dealing with tensor fields.

A tensor, or a vector, is an implicit mathematical object, i.e. it doesn't depend on the reference frames you can use to write it. It's the most important property of this mathematical objects to be used in Physics, translating the independence of the physical process, i.e. the nature, from the observer, i.e. the reference frames or the coordinates used to describe it.


According to Synge & Schild: TENSOR CALCULUS, pp:127-128

[...] define as Cartesian tensors quantities which transform according to the same laws when the coordinates undergo an orthogonal transformation, i.e. when we pass from one set of homogeneous coordinates to another.

and then

We use the word "Cartesian" because homogeneous coordinates in a flat space are analogous to rectangular Cartesian coordinates in a Euclidean plane or 3-space, and indeed reduce to rectangular Cartesians when the space reduces to the Euclidean plane or 3-space. In order to qualify as a tensor, a set of quantities has to satisfy certain laws of transformation when the coordinates undergo a general transformation. This is a much more stringent condition than that which we impose on a set of quantities in order that it may qualify as a Cartesian tensor; in the latter case only orthogonal transformations are involved. Consequently, every tensor is a Cartesian tensor, but the converse is not true. (This is an abuse of language comparable to the following: "All horses are black horses, but the converse is not true." To avoid it, we might call the tensors of Chapter I "complete tensors." A simpler plan is to regard the expression "Cartesian tensor" as a single noun, not divisible into "Cartesian" and "tensor.")

where a homogeneous coordinate transformation is defined as $$z'_m = A_{mn}z_n + A_m \\ A_{mp}A_{nq} = \delta_{pq}$$

By this definition a Cartesian tensor is more general than a "normal" tensor.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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