Suppose $A^{\mu_1 \cdots \mu_n}_{\nu_{n+1}\cdots \nu_m}$ is a tensor. That means it transforms a tensor. How do I show that it transforms as a tensor? How do I see that $\cos (A^{\mu_1 \cdots \mu_n}_{\nu_{n+1}\cdots \nu_m})$ or some other non-tensor does not transform like a tensor?

  • $\begingroup$ Can you clarify exactly what definition of tensor you're using? Otherwise, your question just seems to be "is a tensor a tensor, and is a non-tensor not a tensor". $\endgroup$ – knzhou Mar 9 '17 at 19:03
  • $\begingroup$ Related: physics.stackexchange.com/q/32011/2451. Related question by OP: physics.stackexchange.com/q/317502/2451 $\endgroup$ – Qmechanic Mar 9 '17 at 19:20
  • $\begingroup$ @knzjou I come from a maths background, and i think the way stuff is defined seems to change as tutors sees necessary. They say a tensor is a tensor if it transforms as a tensor. This all tutors seems to agree on. $\endgroup$ – Mikkel Rev Mar 9 '17 at 20:29

A tensor (or more appropriately tensor field) $T^{i_1 \dots i_m}_{j_1 \dots j_n}$ of rank $(m,n)$ transforms as,

$$T'^{i'_1 \dots i'_m}_{j'_1 \dots j'_m} = \left( \frac{\partial\tilde{x}^{i'_1}}{\partial x^{i_1}} \dots \frac{\partial\tilde{x}^{i'_m}}{\partial x^{i_m}} \right) \left(\frac{\partial x^{j_1}}{\partial \tilde{x}^{j'_1}}\dots \frac{\partial x^{j_m}}{\partial \tilde{x}^{j'_m}} \right)T^{i_1 \dots i_m}_{j_1 \dots j_n} $$

under a coordinate transformation $x \to \tilde{x}$. Now, if we have some multi-dimensional array, we need to know specifically what it is to be able to check if it is a tensor. Without further details, the best we can say is it transforms as a tensor if it satisfies the above definition.

As an example, consider the Christoffel symbols,

$$\Gamma^i_{kl} = \frac12 g^{im} \left( \partial_l g_{mk} + \partial_k g_{ml} - \partial_m g_{kl} \right).$$

We know that the metric $g$ is a tensor and thus we know exactly how it transforms under a coordinate transformation. Thus, to figure out if the Christoffel symbols are tensors, we need only plug this in and compute. But we find that,

$$\Gamma'^{k}_{ij} = \frac{\partial x'^p}{\partial x^i}\frac{\partial x'^q}{\partial x^j} \Gamma^r_{pq} \frac{\partial x^k}{\partial x'^r} + \frac{\partial x^k}{\partial x'^m} \frac{\partial^2 x'^m}{\partial x^i \partial x^j}$$

and thus clearly does not transform in the manner defined for a tensor, which is why we call them the Christoffel symbols. As you can see, we need to know what the quantity in question is to be able to check if it transforms the right way.


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.