Spontaneous periodic orbits in the Navier-Stokes flow (Q2022571)

In this paper, the authors propose a computer-assisted approach to the proof of the existence of time-periodic solutions of the Navier-Stokes equations corresponding to a time-independent force. Consider the Navier-Stokes equations on the three-torus \({\mathbb T}^3\) with size length \(2\pi\): \[ \begin{cases} \partial_t u+(u\cdot\nabla)u-\nu\Delta u+\nabla p=f \\ \nabla\cdot u=0 \end{cases}\quad\text{in }\mathbb{T}^3\times\mathbb{R},\tag{1} \] where \(u=u(x,t)\) and \(p=p(x,t)\) denote the fluid velocity and pressure field (scaled by the fluid -- constant -- density), respectively, \(\nu\) is the coefficient of kinematic viscosity, and \(f=f(x)\) is the given external force which is assumed to be independent of time and having zero spatial average. The problem is whether there exist a period \(T\) and a corresponding analytic periodic solution \((u,p)\) of (1) with period \(T\). This solution corresponds to what the authors call ``spontaneous periodic motion'' as it represents a (non-stationary) time-periodic flow of a fluid driven by a time-independent force. The existence of spontaneous periodic motions of a fluid governed by the Navier-Stokes equations has been investigated in the following works: [\textit{V. I. Yudovich}, J. Appl. Math. Mech. 35, 587--603 (1971; Zbl 0247.76044); translation from Prikl. Mat. Mekh. 35, 638--655 (1971)], [\textit{G. Iooss}, Arch. Ration. Mech. Anal. 47, 301--329 (1972; Zbl 0258.35057)], [\textit{D. D. Joseph} and \textit{D. H. Sattinger}, Arch. Ration. Mech. Anal. 45, 79--109 (1972; Zbl 0239.76057)], and [\textit{G. P. Galdi}, Arch. Ration. Mech. Anal. 222, No. 1, 285--315 (2016; Zbl 1352.35096)]. These papers are concerned with periodic solutions branching off from a steady state undergoing bifurcation. In contrast, the authors of the present paper provide a proof of the existence of spontaneous periodic motions which are not necessarily ``close'' to a bifurcation point of a steady-state solution. The strategy of the proof consists in three main steps. The first step is to identify a zero finding problem \(\mathcal F(W)=0\) on the Banach space of geometrically decaying Fourier coefficients. The solution \(W\) corresponds to an angular frequency \(\Omega\) and a time-periodic solution \(\omega\) to the vorticity equation with period \(2\pi/\Omega\). In Lemma 2.5, it is then proved that solutions to \(\mathcal F(W)=0\) correspond to time-periodic solutions \((u,p)\) of (1). As a second step, consider a numerical approximation \(\bar W\) of \(W\), i.e., \(\mathcal F(\bar W)\approx 0\), then the exact zero \(W\) of \(\mathcal F\) will be found as a fixed point of the operator \[ T:\; W\mapsto W-D\mathcal F(\bar W)^{-1}\mathcal F(W), \] in a neighborhood of \(\bar W\). By Banach fixed point theorem, it is enough to show that \(T\) is a contraction in a ball centered at \(\bar W\). To prove the latter, computable estimates of \(\|D\mathcal F(\bar W)^{-1}\|\) are needed. Instead of working with \(D\mathcal F(\bar W)^{-1}\), the authors construct approximations \(\hat A\) and \(A\) of \(D\mathcal F(\bar W)\) and \(D\mathcal F(\bar W)^{-1}\), respectively, and use these operators to find sufficient conditions to ensure that \(T\) is contraction in a ball centered at \(\bar W\) (Theorem 2.15). The final step is to derive and implement explicit bounds that meet the hypothesis of Theorem 2.15. Given the high computational cost to evaluate such bounds, the authors use the symmetries of the model to reduce the size of the zero finding problem (Theorem 4.23). The implementation of the bounds in the symmetric setting can be found in [\textit{J. B. van den Berg} et al., MATLAB code for ``Spontaneous periodic orbits in the Navier-Stokes flow'' (2019), \url{https://www.math.vu.nl/~janbouwe/code/navierstokes/}]. The results for time-periodic solutions which are homogeneous in one space variable (more precisely, they are independent of the third space variable and their third component is zero) corresponding to the Taylor-Green forcing \[ f(x)=\left(\begin{matrix} 2\sin x_1\cos x_2 \\ -2\cos x_1 \sin x_2 \\ 0 \end{matrix}\right) \] are presented.