# Deriving the norm of four-velocity from the definition of a time-like interval

I am going through Sean Carroll's Spacetime and Geometry, trying to learn GR. On page 31, he introduces the concept of the four-velocity $$U^\mu$$ as $$U^\mu = \frac{dx^\mu}{d\tau}$$ where $$\tau$$ is the proper time along a timelike path, with a timelike interval given by $$d{\tau}^2 = -\eta_{\mu\upsilon}dx^{\mu}dx^{\upsilon}$$

He further mentions that, because of the formula above for the timelike interval, the four-velocity is automatically normalized: $$\eta_{\mu\upsilon}U^{\mu}U^{\upsilon} = -1$$

In trying to derive this fact for myself, I tried to take the derivative of the timelike interval w.r.t the proper time twice. For the first:$$\frac{d}{d\tau}(d\tau^2) = \frac{d}{d\tau}(-\eta_{\mu\upsilon}dx^{\mu}dx^{\upsilon})$$ $$2d\tau = -\eta_{\mu\upsilon} . \frac{d}{d\tau}(dx^{\mu}dx^{\upsilon})$$ $$2d\tau = -\eta_{\mu\upsilon}\left(dx^\mu\frac{dx^\upsilon}{d\tau} + dx^\upsilon\frac{dx^\mu}{d\tau}\right)$$ And for the second derivative: $$\frac{d}{d\tau}(2d\tau) = -\eta_{\mu\upsilon}\frac{d}{d\tau}\left(dx^\mu\frac{dx^\upsilon}{d\tau} + dx^\upsilon\frac{dx^\mu}{d\tau}\right)$$ which results in $$2 = -\eta_{\mu\upsilon}\left(2\frac{dx^\mu}{d\tau}\frac{dx^\upsilon}{d\tau} + dx^\mu\frac{dx^\upsilon}{d\tau^2} + dx^\upsilon\frac{dx^\mu}{d\tau^2}\right)$$ I realize that the normalization value for the four-velocity would hold if only the first term of the last equation were present, and I've seen a lot of second-order terms dropped from equations before (mainly in engineering, though). However, I don't think that I can make that assumption in this case, since these derivatives are more symbolic in nature.

My question is this: Have I made a mistake in either the calculations, or made any assumptions that don't really hold? If not, why doesn't this approach work?

In $$-\eta_{\mu\nu}\mathrm{d}x^\mu\mathrm dx^\nu$$ there is an implicit tensor product, namely $$-\eta_{\mu\nu}\mathrm{d}x^\mu\otimes\mathrm dx^\nu$$. This means to contract this one-form with a derivative one must use \begin{align} \frac{\mathrm d}{\mathrm d\tau}\otimes\frac{\mathrm d}{\mathrm d\tau}(-\eta_{\mu\nu}\mathrm{d}x^\mu\otimes\mathrm dx^\nu)=-\eta_{\mu\nu}\frac{\mathrm d x^\mu}{\mathrm d\tau}\frac{\mathrm d x^\nu}{\mathrm d\tau} \end{align} Similarly, $$\frac{\mathrm d}{\mathrm d\tau}\otimes\frac{\mathrm d}{\mathrm d\tau}(\mathrm{d}\tau\otimes\mathrm d\tau)=\frac{\mathrm d\tau}{\mathrm d\tau}\frac{\mathrm d\tau}{\mathrm d\tau}=1$$

• Could you show me where I could read about using the derivative in this way? In the book I mentioned, Carroll has mentioned tensor products but not how derivatives interact with them. Yet Commented Jun 10, 2023 at 23:30
• As the manifold is the spacetime, how do you define $d/d\tau\otimes d/d\tau$? Commented Jun 11, 2023 at 19:19

It may be not the most rigorous explanation, but essentially you are already working with differentials, so instead of deriving it you can just divide as if they were fractions, so just divide your second equation by $$d\tau$$ twice and you get the result.

Actually the correct derivative is like: $$$$\frac{d}{d\tau}\tau=1$$$$ and not $$$$\frac{d}{d\tau}d\tau=1$$$$ (with $$d\tau$$ instead of $$\tau$$)

• I see. I assumed that $d\tau$ was more like "a function of $\tau$", so the derivative of a function w.r.t itself is 1. But I see my error. Also, was never quite sure about treating denominators of derivatives like actual denominators and moving them around; I don't quite know when that is valid and when it isn't. Commented Jun 10, 2023 at 23:28
• For a mathematician you can never do that, but physically speaking you usually deal with "very small but finite" quantities. Of course you must be careful, for example you cannot treat partial derivatives as fractions. Commented Jun 10, 2023 at 23:40

We have: $$\mathrm{d}\tau^2 = -\eta_{\mu\nu} \mathrm{d} x^\mu \otimes\mathrm{d}x^\nu = -\eta_{\mu\nu} \left(\frac{\mathrm{d}x^\mu}{\mathrm{d\tau}} \mathrm{d}\tau\right) \otimes \left(\frac{\mathrm{d}x^\nu}{\mathrm{d\tau}} \mathrm{d}\tau\right) = -\eta_{\mu\nu} \frac{\mathrm{d}x^\mu}{\mathrm{d\tau}} \otimes \frac{\mathrm{d}x^\nu}{\mathrm{d\tau}} \mathrm{d}\tau^2$$ for every $$\mathrm{d}\tau^2$$. For the chain rule, $$x^\mu = x^\mu(\tau)$$ is implied.