Static compliance and directional instability in indefinite conformation states

Abstract

A conformation tensor is positive definite for every physically realizable polymer microstructural state. Numerical discretization can move the conformation tensor outside the positive-definite domain. This raises a question: can the least eigenvalue alone identify the first unstable direction? We answer it by linearizing Oldroyd-B, equilibrium-normalized FENE-P and Giesekus models about uniform frozen states. All include solvent viscosity and stress diffusion. We examine every non-zero planar Fourier mode, assuming each model's uncoupled constitutive tangent is strictly stable. The margin $1+r\chi$ measures the balance between solvent damping and the zero-frequency polymer response. The complete velocity--conformation system is stable if and only if this margin is positive. At zero margin, a simple stationary root appears; finite inertia changes growth rates but not the neutral boundary. At fixed wavenumber and other parameters, decreasing $\lambda_1$ identifies the first neutral direction. Oldroyd-B and equilibrium-normalized FENE-P first become neutral along principal directions; Giesekus mobility can instead make an oblique direction neutral first. For the reference case, onset is $\lambda_{1,c}=-1.933$ at $\theta_c=23.94^\circ$, before the principal-axis prediction. Along this family, the all-direction threshold approaches $-2.319$ as the other principal stretch grows, whereas the formal principal-axis extrapolation tends to negative infinity. A rational-parameter counterexample, matrix spectra and uniform forced-base calculations test the neutral boundary and both sides. These results concern one linear, uniform, planar Fourier mode, not nonlinear or inhomogeneous-flow stability. Within this scope, onset depends not on indefiniteness alone but also on constitutive-tangent geometry and wavevector direction.