Revision history [back]

For a scalar field, it is much faster to compute the Laplace-Beltrami operator as $\nabla_\mu \nabla^\mu\psi$, instead of $\star\mathrm{d}\star\mathrm{d}\psi$. For your example:

sage: nabla = g.connection()
sage: LBpsi = nabla(nabla(psi).up(g)).trace()

takes only 12 s on my computer (and of course yields the same result), see the online notebook.

The slowness you've experienced is due to the lack of optimization in the Hodge dual computation (especially for the Hodge dual of a 6-form, as indicated by the timing in cell 21 of the notebook), which is all the more severe that the dimension is high. This certainly will be improved in some future version. Thanks for pointing it out!