ISSN 1064-5624, Doklady Mathematics, 2013, Vol. 87, No. 2, pp. 148–152. © Pleiades Publishing, Ltd., 2013.
Original Russian Text © M.E. Bogovskii, 2013, published in Doklady Akademii Nauk, 2013, Vol. 449, No. 2, pp. 136–140.
148
INTRODUCTION
If the nonlinear nonstationary Navier–Stokes
problem has a strong solution, it can be constructed
using an iterative method in which each iteration step
involves solving a problem with convective terms lin-
earized about the solution obtained at the previous
iteration. The global convergence of this iterative
method was targeted in [1], for which reason the
method itself is below referred to for brevity as the
method [1]. It was shown in [1] that, for any initial
approximation, the global convergence to the sought
solution in the norm of the class of Hopf weak solu-
tions is faster than a geometric progression rate with a
ratio arbitrarily close to zero. However, the proof of the
global convergence in the norm of the class of strong
solutions for the method [1] faces rather significant
though not insurmountable technical difficulties. In
this paper, a modification of the method [1] is pro-
posed that substantially facilitates the proof of its glo-
bal convergence from any initial approximation in the
norm of the class of strong solutions assuming that the
sought strong solution does exist. More specifically, for
iterated convective terms, certain cutoff factors are
introduced that formally damp the hypothetical
growth of the strong norms of the successively approx-
imated solutions with the number of iterations
increasing.
When the sought strong solution does exist, the
iterative method thus modified is almost identical to
the original method [1], since the cutoff factors are
taking effect only in a number of first iteration steps.
After several iterations, whose number is estimated
from above in terms of the strong norm of the sought
solution, the cutoff factors become identically equal to
unity and hence disappear, letting the modified itera-
tions take their original unmodified form [1].
It is established below that the existence of a strong
solution to the nonlinear Navier–Stokes problem with
any initial approximation from the class of strong solu-
tions is equivalent to the global convergence of the
modified iteration sequence to the sought solution in
the norm of the class of strong solutions at a conver-
gence rate higher than a geometric progression rate
with a ratio arbitrarily close to zero. Interestingly, the
sought strong solution is not involved in the construc-
tion of the modified iteration sequence and, in fact,
the global boundedness of this sequence in the norm of
the class of strong solutions becomes a criterion for the
global solvability of the nonlinear Navier–Stokes
problem in the class of strong solutions (for more
detail, see Section 2 below).
In a bounded domain Ω ⊂
3
with the time-inde-
pendent smooth boundary ∂Ω ∈ C
2
, we consider the
nonlinear evolution Navier–Stokes problem, i.e., the
initial–boundary value problem for the nonstationary
Navier–Stokes equations
(1)
describing the dynamics of an incompressible fluid with a
kinematic viscosity ν > 0, velocity field v: Q
T
→
3
, and
hydrodynamic pressure p: Q
T
→ .
To define a strong solution of the nonlinear initial–
boundary value problem (1), we need to introduce
suitable function spaces. Let L
r
(Q
T
;
3
) denote the
Lebesgue space of vector fields v: Q
T
→
3
that are
integrable to the power r ∈ (1, ∞) over the cylinder Q
T
.
Let (Q
T
;
3
) denote the anisotropic Sobolev
space of vector fields v: Q
T
→
3
, where the super-
scripts l and m designate, as usual, the orders of
smoothness in x and t, while the corresponding deriv-
atives belong to L
r
(Q
T
;
3
). The symbol H
T
stands for
the (Q
T
;
3
)-closed subspace of x-solenoidal
v
t
v ∇ , ( ) v νΔ v – ∇p + + fxt , ( ) , =
div v 0 , xt , ( ) Q
T
∈ Ω 0 T , ( ) , × = =
v
t 0 =
ax () , div a 0 , x Ω, ∈ = =
v
∂Ω
0 , t 0 T , ( ) , a
∂Ω
∈ 0 , = =
def
W
rxt , ,
lm ,
W
2 xt , ,
21 ,
Global Convergence in the Strong Norm of an Iterative Method
for the Nonstationary Navier–Stokes Problem
M. E. Bogovskii
Presented by Academician V.P. Dymnikov October 2, 2012
Received October 10, 2012
DOI: 10.1134/S1064562413020075
Dorodnicyn Computing Center, Russian Academy of Sciences,
ul. Vavilova 40, Moscow, 119333 Russia
e-mail: bogovskii@ccas.ru
MATHEMATICS