I wonder if the current PETSc implementation supports using AMG monolithically for a multi-field problem and using fieldsplit in the smoother.

Thank you very much,

Mike