While reviewing some basic field theory, I once again encountered the Bianchi identity (in the context of electromagnetism). It can be written as $$\partial_{[\lambda}\partial_{[\mu}A_{\nu]]}=0.$$ Here, $A_\nu$ is of course the electromagnetic potential. This formula is immediately reminiscent of the Jacobi identity: $$[A,[B,C]]+[B,[C,A]]+[C,[A,B]]=0.$$

This is even clearer in general relativity, where we have $$\nabla_{[\lambda}R_{\rho\sigma]\mu\nu},$$ which we can rewrite, remembering the definition of the Riemann tensor in terms of the commutator of covariant derivatives, as $$[[\nabla_\lambda,\nabla_\rho],\nabla_\sigma]+[[\nabla_\rho,\nabla_\sigma],\nabla_\lambda]+[[\nabla_\sigma,\nabla_\lambda],\nabla_\rho]=0. $$ This all looks like there should be some profound connection here, but I'm incapable of pinpointing it. Maybe one of the experts here can make this more precise? I'd love to get to know more about this. Any comments are much appreciated. I'd also be grateful if someone could suggest (more) appropriate tags to use.

  • 1
    $\begingroup$ I had the same question, but finally didn't ask it. I think the reason is related to the fact that the Bianchi identity can be derived from the Jacobi identity: diffgeom.subwiki.org/wiki/… $\endgroup$
    – jinawee
    Nov 3, 2013 at 15:19
  • $\begingroup$ Fair enough, but honestly, I was hoping for a 'deeper' answer. Maybe I'm looking for something that simply isn't there, but it feels like there should be a profound connection between two seemingly very different things. Maybe something connected to geometrical properties of the fields? $\endgroup$
    – Danu
    Nov 3, 2013 at 15:36
  • 2
    $\begingroup$ @jinawee : OP is speaking about the second (differential) Bianchi identity $\nabla_{[\lambda}R_{\rho\sigma]\mu\nu}$ $\endgroup$
    – Trimok
    Nov 3, 2013 at 16:31

2 Answers 2


I) The proofs of both the first (algebraic) Bianchi identity and the second (differential) Bianchi identity crucially use that the connection $\nabla$ is torsionfree, so they are not entirely consequences of the Jacobi identity. Proofs of the Bianchi identities are e.g. given in Ref. 1.

II) The second Bianchi identity may be formulated not only for a tangent bundle connection but also for vector bundle connections.

III) The Lie bracket in the pertinent Jacobi identities is the commutator bracket $[A,B]:=A\circ B -B\circ A$. The Jacobi identity follows because operator composition "$\circ$" is associative.

IV) In the context of Yang-Mills theory and EM, the second Bianchi identity follows because the gauge potential $A_{\mu}$ and the field strength $F_{\mu\nu}$ may be viewed as (part of) a covariant derivative and corresponding curvature tensor, respectively.


  1. M. Nakahara, Geometry, Topology and Physics, Section 7.4.

Here is a proof of the Bianchi identity using the Jacobi identity: if we denote by $F$ the curvature of the connection $A$, then locally $$F = dA + \frac{1}{2}[A,A]$$ and therefore $$d_A F = dF + [A,F] = d(\frac{1}{2}[A,A] ) + [A,dA] = 0.$$ Where $[A,[A,A]$ is zero because of the Jacobi identity.


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.