Bernoulli's formula extends integration by parts to integrate ∫u(x)v(x)dx in one pass, whenever u(x) is a polynomial (so its successive derivatives eventually vanish) and v(x) can be integrated repeatedly with ease (e.g. sinnx, cosnx, eax).
Notation. Write u(1)=dxdu, u(2)=dxdu(1), u(3)=dxdu(2),… for the successive derivatives of u, and v(1)=∫vdx, v(2)=∫v(1)dx, v(3)=∫v(2)dx,… for the successive anti-derivatives of v.
Bernoulli's formula.
∫uvdx=uv(1)−u(1)v(2)+u(2)v(3)−u(3)v(4)+⋯
— an alternating sum that terminates because u is a polynomial: once a derivative u(m) hits 0, every later term vanishes.
Derivation sketch. Apply integration by parts once with dv(1)=vdx to get ∫uvdx=uv(1)−∫u(1)v(1)dx; then apply by-parts again to ∫u(1)v(1)dx with dv(2)=v(1)dx, and so on — each step trades one derivative of u for one further anti-derivative of v.
Working rule. List u,u(1),u(2),… in one column down to 0, and v(1),v(2),v(3),… (one integration behind) in a parallel column; multiply pairs on the diagonal with alternating signs +,−,+,−,… and sum.
For a definite integral, evaluate the entire right-hand side (every u(k)v(k+1) term) between the limits at the end, rather than finding the indefinite antiderivative first — this avoids re-deriving the formula and keeps sign bookkeeping simple, exactly as in Examples 9.31-9.34.