Core idea

Suppose U(t,x)=A(t)xU(t,x)=A(t)x, with trA=0\operatorname{tr}A=0, is a smooth unforced solution with pressure pUp_U. Let k(t)Rd{0}k(t)\in\mathbb R^d\setminus\{0\} and a(t)Cda(t)\in\mathbb C^d solve

k=ATk,a=Aa+2k(kAa)k2νk2a,k'=-A^T k,\qquad a'=-Aa+2\frac{k(k\cdot Aa)}{|k|^2}-\nu|k|^2a,

with k(0)a(0)=0k(0)\cdot a(0)=0. Then the shearing wave

u=U+Re(a(t)eik(t)x)u=U+\operatorname{Re}(a(t)e^{ik(t)\cdot x})

is an exact solution with pressure

p=pU+Re(2i(kAa)k2eikx).p=p_U+\operatorname{Re}\left(\frac{2i(k\cdot Aa)}{|k|^2}e^{ik\cdot x}\right).
Why the nonlinear wave term vanishes

The wavevector equation transports the phase by UU. Differentiating kak\cdot a with the displayed ODEs gives zero, preserving transversality. Since the amplitude is spatially constant and perpendicular to kk, every term in (w)w(w\cdot\nabla)w, for w=Re(aeikx)w=\operatorname{Re}(ae^{ik\cdot x}), vanishes. The remaining linear and viscous terms give exactly the amplitude and pressure formulas.

A simple shear and scope

For U=Sx1e2U=Sx_1e_2, one has k1(t)=k1(0)Stk2(0)k_1(t)=k_1(0)-Stk_2(0), with the other components constant. General affine backgrounds use the same displayed evolution. These waves on all of Euclidean space are generally not finite-energy fields. Spatial localization creates additional terms that need correction; the exact plane-wave identity alone does not supply a localized solution.

References