Abstract
In this note we investigate the initial-boundary value problem for a Stokes system arising in a free surface viscous flow of a horizontally periodic fluid with fractional boundary operators. We derive an integral representation of solutions by making use of the multiple Fourier series. Moreover, we demonstrate a unique solvability in the framework of the Sobolev space of \(L^2\)-type.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In this note we investigate a Stokes system on the horizontally periodic domain of the mean depth \(\frac{\pi }{2}\):
where \(x_1\) is the vertical coordinate and \(x'=(x_2,x_3)\) is the horizontal coordinates in the 2-dimensional torus \(\mathbb {T}^2=(-\pi ,\pi ]^2 \cong \mathbb {R}^2/(2\pi \mathbb {Z})^2\). The upper and lower boundaries are denoted by
Our target system is the governing equations:
subject to the boundary conditions:
with \(\Delta ':=\frac{\partial ^2}{\partial x_2^2} + \frac{\partial ^2}{\partial x_3^2}\) and the initial conditions: for \(x=(x_1, x') \in \Omega \),
satisfying certain compatibility conditions. Here \(\eta =\eta (x',t)\) is an unknown graph of the free surface on \(S_F\) and \(u=(u_i(x,t))_{1 \le i \le 3}\) is an unknown velocity field of the fluid with a pressure function \(p= p(x,t)\). Furthermore, \(\nu >0\) is a constant coefficient of the viscosity of the fluid, \(\sigma >0\) is the surface tension coefficient, and \(g \ge 0\) is a constant describing the effect of gravitational acceleration on the fluid. For simplicity, we assume that only gravity acts on the fluid as an external force.
The motion of a viscous incompressible fluid bounded above by an atmospheric pressure on an upper free surface and below by a rigid bottom is modelled by the Navier–Stokes equations with appropriate boundary conditions (cf. [10]), where the effect of the surface tension is taken into account by using the mean curvature of the free interface \(x_1= \eta (x',t)\):
In [2], J. T. Beale studied the initial-boundary value problem of an infinite layer of a viscous fluid having a non-compact upper free surface \(\{x_1=\eta (x',t)\}\) and a rigid bottom surface \(\{x_1=-b(x)\}\) under the effect of surface tension at free surface, where he showed by using the contraction mapping principle that there exists a unique global solution to the free surface Navier-Stokes problem for a sufficiently small initial data with certain compatibility conditions. For this task, he transformed the Navier-Stokes system to a linearized system on the infinite layer \((-b(x), 0) \times \mathbb {R}^2\) by stretching/compressing on vertical line segments instead of using the Lagrangian formulation (cf. [2, (2.1)–(2.6) in Sect. 2]).
In [6], Nishida–Teramoto–Yoshihara studied the motion of horizontally periodic Navier–Stokes fluid with surface tension. Then they showed such a global-in-time existence for small initial data and the exponential-in-time decay of energy via the linearized problem with \(\alpha =1\) at the equilibrium domain \((-b,0) \times \mathbb {T}^2\) (cf. [6, (2.4)–(2.9) in Sect. 2]).
In [9], Tice and Zbarsky introduced the generalized boundary condition of fractional Laplacian type (1.4) in the Stokes system and then studied the decay rate of energy when \(0 < \alpha \le 1\) in both cases of the infinite layer \((0,b) \times \mathbb {R}^{d-1}\) and the horizontally periodic domain \((0,b) \times \mathbb {T}^{d-1}\) for arbitrary dimension \(d \ge 3\).
In this note, we are concerned with construction of solution operator for the initial-boundary value problem (1.1)–(1.7) constituted of the horizontally periodic Stokes system with fractional boundary operators. Indeed, we show that the problem possesses an integral representation of solutions by making use of the multiple Fourier series (cf. Theorem 3.1). From the alternative formulation, we demonstrate the unique solvability, provided that \(0< \alpha < 3/2\) (cf. Theorem 4.1).
This note is organized as follows. In Sect. 2, as a preliminary, we review the almost-everywhere convergence results of the multiple Fourier series in the Sobolev spaces and also introduce the Sobolev–Slobodeckij spaces of \(L^2\)-type. In Sect. 3, we derive a Fourier representation of solutions to the IBVP (1.1)–(1.7). In Sect. 4, we establish the unique solvablity of the IBVP (1.1)–(1.7).
2 Preliminaries
Let \(\partial _i = \partial /\partial x_i\) for \(i=1,\dots ,d\). We use the standard multi-index notation for the spatial differential operator: \(\partial ^\beta = \partial _1^{\beta _1} \dots \partial _d^{\beta _d}\) with the order \(|\beta |=\beta _1 + \dots + \beta _d\) for \(\beta =(\beta _1,\dots ,\beta _d) \in \mathbb {N}_0^d\), and \(\partial ^0 = Id\).
We denote by \(\mathbb {T}^d\) the d-dimensional torus \((-\pi , \pi ]^d \cong \mathbb {R}^d/(2\pi \mathbb {Z})^d\), and specially, \(\mathbb {T}:= \mathbb {T}^1\).
Let \(\{\mathcal {F} [f](n)\}_{n \in \mathbb {Z}^d}\) denote the Fourier coefficients of an absolutely integral function \(f(x)=f(x_1,\dots ,x_d)\) on \(\mathbb {T}^d\):
where \(n \cdot x = n_1 x_1+ \dots + n_d x_d\) and \(dx = dx_1 \dots dx_d\). The multiple Fourier series is formally defined by the series
Here we recall the celebrated almost-everywhere convergence result of the multiple Fourier series on \(\mathbb {T}^d\) (cf. [3, 4, 7, 8]). If \(f=f(x)\) lies in the Lebesgue space \(L^p(\mathbb {T}^d)\) with \(p >1\), the rectangular partial sum
converges to f(x) for almost every \(x \in \mathbb {T}^d\) as \(N \rightarrow \infty \). Recently, for the Sobolev (or Bessel potential) space \(H^{r,p}(\mathbb {T}^d)\) with the norm
Ashurov [1] proved that the spherical partial sum
converges to f(x) in \(H^{r,p}(\mathbb {T}^d)\) for almost every \(x \in \mathbb {T}^d\) as \(N \rightarrow \infty \), provided that \(1 < p \le 2\) and \(r > (d-1)(\frac{1}{p}-\frac{1}{2})\). Therefore, when \(p=2\), if \(f \in H^{r,2}(\mathbb {T}^d)\) for \(r \ge 0\), then we obtain the inversion formula \(f(x)= \mathcal {F}^*[\mathcal {F}[f]](x)\) for a.e. \(x \in \mathbb {T}^d\), where the Fourier series \(\mathcal {F}^*[\,\cdot \,](x)\) corresponds to the pointwise limit of the spherical (resp. rectangular) partial sum when \(r>0\) (resp. \(r=0\)). Hence, from the Parseval formula
we get the characterization of \(H^{r,2}(\mathbb {T}^d) = \{ f \in L^2(\mathbb {T}^d): \Vert f\Vert _{H^{r,2}(\mathbb {T}^d)} < \infty \}\). From now on, we shall use the shorthand notation \(H^r(\mathbb {T}^d):= H^{r,2}(\mathbb {T}^d)\).
For \(m \in \mathbb {N}_0\), we set the Slobodeckij space of \(L^2\)-type on \(D= \mathbb {T}^d\) or \(\Omega \):
In particular, \(W^m(\mathbb {T}^d)\) (resp. \(W^0(D)\)) coincides with \(H^m(\mathbb {T}^d)\) (resp. \(L^2(D)\)). We also define the space of the \(\frac{1}{2}\)-fractional type on \(\mathbb {T}^d\):
with the norm
We should notice that there exists a trace operator \(\gamma : W^m (\Omega ) \rightarrow W^{m-\frac{1}{2}}(S_B \cup S_F)\) such that \(\gamma f = f|_{S_B \cup S_F}\) for \(m \in \mathbb {N}\) (cf. [5]).
Let us introduce the subspace of \(W^m(\mathbb {T}^3)\) (resp. \(H^r(\mathbb {T}^3)\)) consisting of functions with the sine-like symmetry in \(x_1 \in \mathbb {T}\):
The space restricted to \(\Omega \) is denoted by
with the norm
In particular, when \(m = 2\) (resp. \(m = 3\)), \(f \in W_{s_1}^m(\mathbb {T}^3)\) satisfies \(f=0\) on \(S_B\) (resp. \(f=0\) on \(S_B\) and \(\partial _1 f=0\) on \(S_F\)), since \(W^2(\mathbb {T}^3) \subset C(\mathbb {T}^3)\) (resp. \(W^3(\mathbb {T}^3) \subset C^1(\mathbb {T}^3)\)) by the embedding theorem.
For a function \(f(x)=f(x_1,x_2,x_3)=f(x_1, x'): \mathbb {T}^3 \rightarrow \mathbb {R}\), we denote the tangential Fourier coefficients by
for \((x_1, n') \in \mathbb {T}\times \mathbb {Z}^2\), where \(n'=(n_2, n_3)\), \(n' \cdot x' = n_2x_2+ n_3 x_3\) and \(dx' = dx_2dx_3\). The Fourier series of a given function \(\hat{f}(x_1, n'): \mathbb {T}\times \mathbb {Z}^2 \rightarrow \mathbb {C}\) is denoted by
We define the \(x_1\)-tangential fractional Laplacian acting on \(f(x): \Omega \rightarrow \mathbb {R}\) by
with \(|n'|^2 = n_2^2+n_3^2\) and \(\hat{f}(x_1,n')=\mathcal {F}'[f](x_1,n')\).
For a function \(g(x_1): (0, \frac{\pi }{2}) \rightarrow \mathbb {R}\), we set the sine-type (resp. cosine-type) Fourier coefficients
for \(n_1 \in \mathbb {N}\) and the associated Fourier series
for \(x_1 \in \mathbb {T}\).
For a function \(f(x): \Omega \rightarrow \mathbb {R}\), we define the coefficients of the hybrid \(x_1\)-sine-type Fourier series by
for \(n=(n_1,n') \in \mathbb {N}\times \mathbb {Z}^2\) and the associated series of \(\mathcal {F}_{s_1}[f]: \mathbb {N}\times \mathbb {Z}^2 \rightarrow \mathbb {C}\) by
Here we note that for \(f(x):\mathbb {T}^3 \rightarrow \mathbb {R}\) and \(l_1 \in \mathbb {Z}\),
for \(n_1 \in \mathbb {N}_0\), and thus, \(\mathcal {F}^*[\mathcal {F} [f]] = \mathcal {F}^*_{s_1}[\mathcal {F}_{s_1}[f]]\). On the other hand, thanks to the convergence results as mentioned above, the inversion formula \(\mathcal {F}_{s_1}^*[\mathcal {F}_{s_1}[f]] =f\) holds on \(W_{s_1}^m(\Omega )= H_{s_1}^m(\mathbb {T}^3) \) for \(m \in \mathbb {N}_0\) under the suitable limit operation. Therefore, we may redefine
For \(\rho ,r >0\), we define the fractional differential operator \(|\partial _1|^\rho |\nabla '|^r\) acting on \(f \in W_{s_1}^0(\Omega ) \cong L^2(\Omega )\) by
for \(n=(n_1,n') \in \mathbb {N}\times \mathbb {Z}^2\).
We finish this section to provide basic properties regarding the heat semigroups.
Proposition 2.1
For \(f: \mathbb {T}^d \rightarrow \mathbb {R}\), let
Then \(\{ e^{\nu t \Delta } \}_{t \ge 0}\) is a strongly continuous semigroup on \(H^r(\mathbb {T}^d)\).
Moreover, for every \(m \in \mathbb {N}\), there exists a constant \(C=C(m) >0\) such that
and \(v:= e^{\nu t \Delta }f\) is a smooth solution to the heat equation \(\partial _t v= \nu \Delta v\) on \(\mathbb {T}^d \times (0,\infty )\), where \(\Delta = \partial _1^2 + \dots +\partial _d^2\).
In addition, \(\Vert e^{\nu t \Delta } f\Vert _{H^r(\mathbb {T}^2)} \le \Vert f\Vert _{H^r(\mathbb {T}^2)}\) for \(t \ge 0\).
Proof
We can verify the main statement by using the inversion formula and the Parseval formula. We omit the detail of the proof. \(\square \)
Proposition 2.2
Let \(m \in \mathbb {N}_0\) be arbitrary. For \(f: \Omega \rightarrow \mathbb {R}\), let
Then \(\{ e^{\nu t \Delta _{s_1}} \}_{t \ge 0}\) is a strongly continuous semigroup on \(W_{s_1}^m(\Omega )\) such that \(v:= e^{\nu t \Delta _{s_1}} f\) is a smooth solution to the heat equation \(\partial _t v = \nu \Delta v\) in \(\Omega \times (0,\infty )\). Moreover, the solution \(v=v(t)\) satisfies \(\partial _1^{2k} v=0\) on \(S_B\) and \(\partial _1^{2k+1} v=0\) on \(S_F\) for all \(t>0\) and every \(k \in \mathbb {N}_0\).
Proof
For \(f \in W_{s_1}^m (\Omega )\), let \(f^\star \in W_{s_1}^m (\mathbb {T}^d)\) with \(f^\star |_{\{0< x_1 < \frac{\pi }{2}\}} =f\). Let \(v:=e^{\nu t \Delta } f^\star \). Since \(v= e^{\nu t \Delta _{s_1}} f\), the main assertion follows from Proposition 2.1. Furthermore, it is easily verified that \(\partial _1^{2k} v|_{x_1=0}=\partial _1^{2k+1} v|_{x_1=\frac{\pi }{2}}=0\). \(\square \)
3 Fourier Representation of Solutions
In this section, we present a Fourier representation of solutions to the IBVP (1.1)–(1.7).
For a function \(z=z(x_2,x_3)=z(x'): \mathbb {T}^2 \rightarrow \mathbb {R}\), we set the heat semigroups:
and
which is consistent with the semigroup (2.15) when \(f= x_1z(x_2,x_3)\).
For a set of functions \(\{ \eta (x',t),\eta _0(x'), u(x,t), a(x), p(x,t)\}\) in (1.1)–(1.7), we set
Theorem 3.1
Consider the initial-boundary value problem (1.1)–(1.7) with the additional boundary condition:
Let
and let
We set the auxiliary functions:
and
for \(i=2,3\). Then the solution \((\eta , u)\) satisfies the equations
and
for \(i=2,3\).
Proof
We have that the initial datum \(a=u|_{t=0}\) satisfies the conditions:
Set
The boundary condition (1.4) can be rewritten as
which gives
Applying the divergence operator \(\nabla \cdot \) to the second equation (1.2), we obtain that \(\Delta p =0\), i.e., \((\partial _1^2 -|n'|^2) \hat{p}=0\). Therefore, \(\hat{p}=C_1(n',t) e^{x_1|n'|}+ C_2(n',t) e^{-x_1|n'|}\). Here we impose the boundary condition on \(S_B\): \(\partial _1 p|_{x_1=0} = 0\), that is, \(\partial _1 \hat{p}|_{x_1=0} = 0\). We thus deduce from the two boundary data that
Since the second equation (1.2) yields
we deduce from the additional boundary condition \(\partial _1^3 u |_{x_1=\frac{\pi }{2}} =0\) in (3.4) that the function \(z_1(x',t)=v_1(\frac{\pi }{2}, x', t)\) satisfies
On the other hand, we can see from (3.18) with the pressure (3.17) that \(\hat{v}_1(x_1,n',t)\) is a solution to the IBVP of the 1-D heat equation:
for \((x_1,n',t) \in [0,\hbox { }\ \frac{\pi }{2}] \times \mathbb {Z}^2 \times [0,\infty )\) subject to the boundary conditions
and the initial condition \(\hat{v}_1|_{t=0} = \partial _1 \hat{a}_1\). In particular, it follows from (3.19) with the pressure (3.17) that
for \((n',t) \in \mathbb {Z}^2 \times [0,\infty )\), which yields
Therefore, we obtain (3.8).
Set \(\hat{w}_1:= \hat{v}_1 -\hat{z}_1\). We deduce from (3.20)–(3.22) with (3.17) that
subject to the boundary conditions
and the initial condition \(\hat{w}_1|_{t=0} = \partial _1 \hat{a}_1 - \partial _1 \hat{a}_1|_{x_1=\frac{\pi }{2}}\). By using the Fourier method and the Duhamel formula, we get
since
and
On the other hand, we have
since \(\partial _1 \hat{u}_1=\hat{v}_1=\hat{z}_1 + \hat{w}_1\) and \(\hat{u}_1|_{x_1=0} =0\). We compute
We apply \(\mathcal {F}'^*[\, \cdot \,]\) to the both sides in (3.26) to obtain (3.12).
We deduce from the first component in equation (1.2) with the pressure (3.17) that
Thus \(\hat{z}_0(n',t)=\hat{u}_1(\frac{\pi }{2},n',t)\) is governed by
since \(\partial _1^2 \hat{u}_1|_{x_1=\frac{\pi }{2}}=0\) (cf. (3.4)). That is,
and we get Eq. (3.7). Furthermore, Eq. (3.11) follows from the first equation (1.1).
Let us derive integral representations of \(\hat{u}_i=\hat{u}_i(t)\) for \(i=2,3\) similarly as above. For \(\hat{v}_i\) for each \(i=2,3\), we deduce from (3.18) with the pressure (3.17) the governing equation
for all \((x_1,n',t) \in [0,\frac{\pi }{2}] \times \mathbb {Z}^2 \times [0,\infty )\) and from (1.5) and (3.4) the boundary conditions
and from (1.7) the initial condition
Set
From (3.27)–(3.31), we have the IBVP of the 1–D heat equation on \([0,\hbox { }\ \frac{\pi }{2}]\): for \(i=2,3\),
subject to the boundary conditions
and the initial condition
Again, by using the Fourier method and the Duhamel formula, we get
since
On the other hand, we have
since \(\partial _1 \hat{u}_i=\hat{v}_i= -\sqrt{-1} n_i \hat{z}_0 + \hat{w}_i\) and \(\hat{u}_i|_{x_1=0} =0\). We compute
Therefore, (3.13) is obtained by applying \(\mathcal {F}'^*\) to the both sides in (3.32). \(\square \)
4 Unique Solvability
In this section, we establish the unique solvability of the IBVP (1.1)–(1.7) under the condition \(0< \alpha <3/2\).
Theorem 4.1
Assume \(0< \alpha < 3/2\). Let an integer \(m \ge 2\) and \(T>0\) be arbitrary. Suppose that \(\eta _0 \in H^{m+1} (\mathbb {T}^2)\) and that \(a=(a_i)_{1 \le i \le 3}\), \(a_i \in W_{s_1}^m(\Omega )\) satisfies the compatibility conditions:
Then there exist unique functions \(h, z_0,z_1\) and \(Q=(Q_i)_{1 \le i \le 3}\) in (3.6)–(3.10) such that
for \(i=2,3\), and the velocity field \(u=(u_i(x,t))_{1\le i \le 3}\) given by
is a solution with \(\eta (x',t)= h(x',t) + \frac{\pi }{2}\) to the problem (1.1)–(1.7).
Proof
Let us redefine the \(H^r\)-norm of a function \(z(x')=z(x_2,x_3):\mathbb {T}^2 \rightarrow \mathbb {R}\) for \(r \ge 0\) by
We have from Eq. (1.1) that \(\{ \hat{h}(n',t)\}_{n' \in \mathbb {Z}^2, t>0}\) is governed by
where
associated with
corresponding to equations (3.28) and (3.23) respectively.
For \(n'=0\), we deduce from (4.10) with (4.11) that
For every \(n' \ne 0\), by substituting (4.11) into (4.10) and integrating by parts, we obtain that
Substituting (4.12) into the above equation, we get
where
Let
Differentiating the Eq. (4.14), we obtain that \(\hat{h}=\hat{h}(n',t)\) satisfies
with \(D_t= d/dt\). Multiplying the above equation by \(e^{3 \nu |n'|t}\), we get by differentiation,
Thus we obtain the governing ODE of \(\{\hat{h}(n',t)\}_{n' \ne 0, t>0}\):
with the initial conditions
where
Since the characteristic equation \(\xi ^2 + 3\nu |n'|^2\xi + \mu ^\alpha (n')=0\), we set
with
Then we can find a number \(N_0 \in \mathbb {N}\), a sum of two squares such that
since \(\tanh \frac{\pi }{2} |n'|/|n'|^{3-2\alpha }\) is a decreasing function in terms of \(|n'|\) for every \(\alpha < 3/2\). Furthermore, thanks to the discreteness of \(n'\), there exists a constant \(\varepsilon _0 = \varepsilon _0(\nu ,g,\sigma ,\alpha ) \in (0,1)\) independent of \(n'\) such that
In the case (i), we set
with the Wronskian
By the solution formula for the 2nd order linear ODE:
we obtain the general solution to (4.15) given by
We also have that the initial conditions of \(\hat{h}(n',t)\) imply \(C_1+C_2 =\hat{h}_0(n')\) and \(-\omega _1(n')C_1 -\omega _2(n')C_2 + J_0(n',0)= \hat{a}_1(\hbox { }\ \frac{\pi }{2},n')\), that is,
Since
and
we deduce from (4.19) that
for all \(|n'| \ge \sqrt{N_0}\). Therefore, it follows from the above estimate that
with a constant \(C_3=C_3(\nu ,g,\sigma ,\alpha )>0\). If \(N_0=1\), then we have
since (4.13). On the other hand, if \(N_0 > 1\), we must divide the case (ii) in (4.17) into the two cases separately:
In the case (a), a set of solutions \(\{\hat{h}(n',t)\}_{1 \le |n'| \le \sqrt{N_0-1},t>0}\) is obtained by the solution formula (4.19) with
associated with the Wronskian \(W(n',t)\). Similarly as above, since there exists a constant \(\varepsilon _1=\varepsilon _1(\nu ,g,\sigma ,\alpha )>0\) independent of \(n'\) such that
we can find a constant \(C_4=C_4(\nu , g, \sigma , \alpha )>0\) such that
In the case (b), a set of solutions \(\{\hat{h}(n',t)\}_{|n'| = \sqrt{N_0-1}, t>0}\) is obtained by the solution formula (4.19) with \(g_1(n', t):= e^{-\frac{3}{2} \nu |n'|^2 t}\) and \(g_2(n', t):= te^{-\frac{3}{2} \nu |n'|^2 t}\). We also have
with a constant \(C_5=C_5(\nu , g, \sigma , \alpha )>0\). Therefore, we get
with \(C_6=C_6(\nu , g, \sigma , \alpha ):= \max \{C_3, C_4,C_5\}\). Hence, it follows that
for all \(0< t < T\).
For the Eq. (4.12), we have
which yields that for all \(0< t < T\),
Therefore, we deduce from (4.25) with \(\hat{z}_1(0,t)= \partial _1 \hat{a}_1(\frac{\pi }{2},0)\) that
for all \(0<t <T\) with a constant \(C_7 =C_7(\nu ,g,\sigma ,\alpha )>0\). Furthermore, one can see that such a solution \(z_1\) lies in \(C([0,T), H^{m-2}(\mathbb {T}^2))\).
Substituting (4.12) into (4.11), we have that for every \(n' \ne 0\),
which implies
for all \(0<t <T\). Thus,
Therefore, we have that for all \(0< t < T\),
with a constant \(C_8=C_8(\nu , g, \sigma , \alpha )>0\), since \(\hat{z}_0(0,t) = \hat{a}_1(\frac{\pi }{2},0)\). Furthermore, one can verify that \(z_0\) belongs to \(C([0,T), H^{m-1}(\mathbb {T}^2))\).
As for \(Q_1\) in (3.9), we estimate for any \(\rho < 3/2\),
for all \(0< t < T\), where \(C_\rho , C'_\rho \) are positive constants depending only on \(\rho ,\nu , g, \sigma , \alpha \). Similarly, we can see that that for any \(\rho ' <1/2\), \(Q_i\) for \(i=2,3\) satisfies
for all \(0< t < T\), where \(C_{\rho '}, C'_{\rho '}\) are positive constants depending only on \(\rho ', \nu , g, \sigma , \alpha \). Hence, we complete the proof of Theorem 4.1. \(\square \)
Data availability
Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.
References
Ashurov, R.R.: Almost everywhere convergence of multiple trigonometric Fourier series of functions from Sobolev classes. (Russian); translated from Mat. Zametki 109 (2021), no. 2, 163–169. Math. Notes 109(1–2), 157–162 (2021)
Beale, J.T.: Large-time regularity of viscous surface waves. Arch. Ration. Mech. Anal. 84(4), 307–352 (1983/84)
Carleson, L.: On convergence and growth of partial sums of Fourier series. Acta Math. 116, 135–157 (1966)
Hunt, R.A.: On the Convergence of Fourier Series. Orthogonal Expansions and their Continuous Analogues (Proc. Conf., Edwardsville, Ill., 1967) pp. 235–255. Southern Illinois University Press, Carbondale (1968)
Lions, J.-L., Magenes, E.: Non-homogeneous Boundary Value Problems and Applications. Vol. I. Translated from the French by P. Kenneth. Die Grundlehren der mathematischen Wissenschaften. Springer, New York (1972–1973)
Nishida, T., Teramoto, Y., Yoshihara, H.: Global in time behavior of viscous surface waves: horizontally periodic motion. J. Math. Kyoto Univ. 44(2), 271–323 (2004)
Sjölin, P.: Convergence almost everywhere of certain singular integrals and multiple Fourier series. Ark. Mat. 9, 65–90 (1971)
Tevzadze, N.R.: The convergence of the double Fourier series at a square summable function. Sakharth. SSR Mecn. Akad. Moambe 58, 277–279 (1970). (Russian)
Tice, I., Zbarsky, S.: Decay of solutions to the linearized free surface Navier–Stokes equations with fractional boundary operators. J. Math. Fluid Mech. 22(4), 1–28 (2020)
Wehausen, J.V., Laitone, E.V.: Surface Waves. Handbuch der Physik, Part 3, vol. 9, pp. 446–778. Springer, Berlin (1960)
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The author declares that he has no conflict of interest.
Additional information
Communicated by Y. Giga.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Hirata, D. On a Stokes System Arising in a Free Surface Viscous Flow of a Horizontally Periodic Fluid with Fractional Boundary Operators. J. Math. Fluid Mech. 26, 16 (2024). https://doi.org/10.1007/s00021-023-00850-3
Accepted:
Published:
DOI: https://doi.org/10.1007/s00021-023-00850-3