A weak formulation of a specifies admissible unknowns, , and identities obtained by transferring derivatives to tests through . A weak solution is an admissible unknown satisfying every prescribed identity.

Example and scope

For Δu=f-\Delta u=f on an open set, the distributional formulation for locally integrable u,fu,f is u(Δϕ)=fϕ\int u(-\Delta\phi)=\int f\phi for every compactly supported smooth ϕ\phi. A variational formulation may instead require a Sobolev space and use uϕ=fϕ\int\nabla u\cdot\nabla\phi=\int f\phi. Initial and boundary conditions need their own trace or test conventions. Nonlinear products must exist in the stated function spaces; an integral identity alone does not supply their integrability.