- #1
- 2,970
- 1,513
Hi,
I have a question concerning gravity obtained by gauging symmetry algebras, a method used often in the context of supergravity.
One can gauge the Poincaré algebra with generators {P,M}. P generates translations, M generators Lorentz transformations, schematically given by
[tex]
[P,P]=0, \ \ \ \ [M,P]=P, \ \ \ \ \ [M,M]=M
[/tex]
This gives two gauge fields, one for P and one for M:
[tex]
e_{\mu}{}^a, \ \ \ \omega_{\mu}{}^{ab} \ \ \ (1)
[/tex]
In the first order formalism one then imposes a curvature constraint. The curvature R of translations, R(P), is put to zero: R(P)=0. This amounts to putting the torsion to zero. The effect is that the [itex]\omega_{\mu}{}^{ab}[/itex] can be solved for if one introduces inverse fields [itex]e^{\mu}{}_a[/itex], and the local P-translations on the remaining field [itex]e_{\mu}{}^a[/itex] can be regarded as a combination of a general coordinate transformation and a Lorentz transformation (this is not trivial, but nevertheless true). This makes one to identify [itex]e_{\mu}{}^a[/itex] as vielbein.
One can then impose the Vielbein postulate to express the Levi-Civita connection [itex]\Gamma^{\rho}_{\mu\nu}[/itex] in terms of the gauge fields (1). Then the Riemann tensor can be expressed in terms of the curvature of Lorentz transformations R(M),
[tex]
R^{\mu}{}_{\nu\rho\sigma} = -e^{\mu}{}_a R_{\rho\sigma}{}^{ab}(M) e_{\nu \ b}
[/tex]
In this way one can "obtain general relativity by a gauging procedure on the Poincare algebra".
My question is: how does this first-order analysis extend to the (A)dS algebra? The problem is that the (A)dS algebra is semi-simple (in constrast with the Poincare algebra), so effectively one has 1 generator. Splitting things up becomes messy. If one applies the above analysis naively to the (A)dS algebra written as
[tex]
[P,P]=M, \ \ \ \ [M,P]=P, \ \ \ \ \ [M,M]=M
[/tex]
then one doesn't get the deformation [P,P]=M into the Riemann tensor via the Vielbein postulate; because the curvature R(P) doesn't change, the spin connection doesn't change, and if the Vielbein postulate doesn't change, the Riemann tensor doesn't change, and so effectively one is gauging the Poincaré algebra again.
I looked up some literature about this (Mansouri, Ortin's "gravity and strings", Wilczek) but I would like to understand the precise reason why the 1-order formalism a la Poincaré goes wrong. And how to improve it, of course.
So if anyone has an idea: I would be happy with some comments :)
I have a question concerning gravity obtained by gauging symmetry algebras, a method used often in the context of supergravity.
One can gauge the Poincaré algebra with generators {P,M}. P generates translations, M generators Lorentz transformations, schematically given by
[tex]
[P,P]=0, \ \ \ \ [M,P]=P, \ \ \ \ \ [M,M]=M
[/tex]
This gives two gauge fields, one for P and one for M:
[tex]
e_{\mu}{}^a, \ \ \ \omega_{\mu}{}^{ab} \ \ \ (1)
[/tex]
In the first order formalism one then imposes a curvature constraint. The curvature R of translations, R(P), is put to zero: R(P)=0. This amounts to putting the torsion to zero. The effect is that the [itex]\omega_{\mu}{}^{ab}[/itex] can be solved for if one introduces inverse fields [itex]e^{\mu}{}_a[/itex], and the local P-translations on the remaining field [itex]e_{\mu}{}^a[/itex] can be regarded as a combination of a general coordinate transformation and a Lorentz transformation (this is not trivial, but nevertheless true). This makes one to identify [itex]e_{\mu}{}^a[/itex] as vielbein.
One can then impose the Vielbein postulate to express the Levi-Civita connection [itex]\Gamma^{\rho}_{\mu\nu}[/itex] in terms of the gauge fields (1). Then the Riemann tensor can be expressed in terms of the curvature of Lorentz transformations R(M),
[tex]
R^{\mu}{}_{\nu\rho\sigma} = -e^{\mu}{}_a R_{\rho\sigma}{}^{ab}(M) e_{\nu \ b}
[/tex]
In this way one can "obtain general relativity by a gauging procedure on the Poincare algebra".
My question is: how does this first-order analysis extend to the (A)dS algebra? The problem is that the (A)dS algebra is semi-simple (in constrast with the Poincare algebra), so effectively one has 1 generator. Splitting things up becomes messy. If one applies the above analysis naively to the (A)dS algebra written as
[tex]
[P,P]=M, \ \ \ \ [M,P]=P, \ \ \ \ \ [M,M]=M
[/tex]
then one doesn't get the deformation [P,P]=M into the Riemann tensor via the Vielbein postulate; because the curvature R(P) doesn't change, the spin connection doesn't change, and if the Vielbein postulate doesn't change, the Riemann tensor doesn't change, and so effectively one is gauging the Poincaré algebra again.
I looked up some literature about this (Mansouri, Ortin's "gravity and strings", Wilczek) but I would like to understand the precise reason why the 1-order formalism a la Poincaré goes wrong. And how to improve it, of course.
So if anyone has an idea: I would be happy with some comments :)
Last edited: