Talk:Contorsion tensor

Source/Reference missing
There are no references is this article, if you know one, please add it. — Preceding unsigned comment added by 134.96.96.228 (talk) 10:05, 5 June 2014 (UTC)


 * I believe that it is the same thing as the below, occuring in the proof of theorem 6.2.5 in David Bleecker, "Gauge Theory and Variational Principles" (1982) D. Reidel Publishing


 * There it is not actually given a name, but is written as
 * $$\Sigma(X,Y,Z)=\langle\phi(Z), \Theta(X,Y)\rangle +

\langle\phi(Y), \Theta(Z,X)\rangle - \langle\phi(X), \Theta(Y,Z)\rangle $$


 * where $$\Theta$$ is the torsion form, and $$\phi$$ is the solder form (tautological one-form) and $$\langle,\rangle$$ is the scalar product. The non-cyclic nature of the x,y,z in the above always bugged me although its obviously needed in the proof. What is remarkable is this: one gets
 * $$\Sigma(X,Y,Z)=2\langle\phi(Z), \sigma(X)\phi(Y)\rangle$$


 * and the $$\sigma(X)$$ is exactly what you need to add to an arbitrary connection to get the torsion-free Levi-Civita connection. I copied above from the book; someone should really double check that this really is the contorosion tensor (I am really pretty sure it is) and add it to the article.  Its mentioned explicitly half way down, here: vertical bundle. Sorry, its the lower-case sigma that is the contorsion form, not the upper-case sigma. Note also the factor of two agrees with what is in this article. The only point of having the phi's in there is to do the index raising/lowering fiddle-faddle in an index-free-way. That is, the lower-case sigma is the contorsion form, defined in terms of the upper-case expression.67.198.37.16 (talk) 22:16, 30 April 2016 (UTC)

Sign convention
Added a note about the sign of the contorsion tensor, which is reversed due to using the opposite sign for the torsion tensor as is usual when using the convention $$\nabla_{i}v^{j} = \partial_{i}v^{j} + {\Gamma^{j}}_{ki} v^{k}$$ for the lower index ordering of the connection coefficients. Adam Marsh (talk) 00:57, 3 September 2020 (UTC)


 * Thanks! Double-thanks, it wasn't till I was lying in bed that I realized which sign you were referring to! Oh, but wait, huh, looking at the edit history, I see that recent edits, just a few months ago, changed around the signs and the order of the indexes. ... and introduced an extra minus sign into the works. Which was not propagated into all sections of the article. So there are clearly several conventions floating about, for signs and for indexes. Harrumph. Whoever invented signs should be shot. 67.198.37.16 (talk) 06:02, 3 November 2020 (UTC)