Acta Mathematica Scientia 2018,38B(6):1759–1766

http://actams.wipm.ac.cn

LOCAL EXISTENCE AND BLOW-UP CRITERION

OF 3D IDEAL MAGNETOHYDRODYNAMICS
EQUATIONS∗

Jae-Myoung KIM
Center for Mathematical Analysis & Computation, Yonsei University, Seoul 03722,
Republic of Korea
E-mail : [email protected]

Abstract We investigate the local existence of smooth solutions of a 3D ideal magneto-

hydrodynamics (MHD) equations in a bounded domain and give a blow-up criteria to this
equations with respect to vorticists.
Key words ideal magnetohydrodynamics equations; blow-up criterion; bounded domain
2010 MR Subject Classification 35Q35; 76D03; 76W05

1 Introduction
We consider the existence for solutions (u, b, π) : QT → R3 ×R3 ×R to the three-dimensional
incompressible ideal magnetohydrodynamic (MHD) equations


 ut + (u · ∇)u − (b · ∇)b + ∇π = 0,

bt + (u · ∇)b − (b · ∇)u = 0, in QT := Ω × [0, T ), (1.1)



div u = 0 and div b = 0
|b|2
here u is the flow velocity, b is the magnetic field and π = p + 2 is the total pressure. We
consider the initial value problem of (1.1), which requires initial

u(x, 0) = u0 (x) and b(x, 0) = b0 (x), x∈Ω (1.2)

together with the boundary conditions defined as follows

u·n=0 and b · n = 0, (1.3)

where n is the outward unit normal vector along boundary ∂Ω.

In [4, 7, 13], it is well known that for a smooth initial data, there exists local in time
(smooth) solution to the equations (1.1)–(1.2) with the boundary conditions (1.3). In contrast
to the Euler equation, the global in time smooth solutions to the inviscid MHD equations is not
∗ Received
August 10, 2017; revised April 3, 2018. Jae-Myoung Kim was supported by NRF-
2015R1A5A1009350 and the National Research Foundation of Korea Grant funded by the Korean Government
(NRF-2016R1D1A1B03930422).
1760 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

known even in two dimensions. On the other hand, the blow-up criterion of smooth solutions
to this equations were studied by many researcher (see [1–4, 11, 12, 15]).
In this paper, we prove the existence and uniqueness of the local in time (smooth) solution
to (1.1)–(1.3) in a bounded domain in R3 via the energy estimate in [7] and [13]. Moreover,
we give a blow-up criteria to this equations with respect to vorticists, motivated by [12]. More
speaking, we are to obtain a similar extension criterion for the 3D ideal MHD equations.
The organization of the paper is as follows. In Section 2, we introduce some notation and
well-known results, and state our main theorems. In Sections 3 and 4, we prove the theorems.

2 Preliminaries and Main Results

2.1 Main Results

In this subsection, we introduce some notations and main results. Let Ω be a bounded
domain with smooth boundary ∂Ω in R3 . For 1 ≤ q ≤ ∞, we denote the usual Sobolev spaces
by W k,q (Ω) = {u ∈ Lq (Ω) : Dα u ∈ Lq (Ω), 0 ≤ |α| ≤ k}. As usual, W0k,q (Ω) is the completion
′
of C0∞ (Ω) in the W k,q (Ω) norm. We also denote by W −k,q (Ω) the dual space of W0k,q (Ω), where
q and q ′ are Hölder conjugates. When q = 2, we write W k,q (Ω) (or W0k,q (Ω)) as H k (Ω) (or
H0k (Ω)).
Now we state our main results.
Theorem 2.1 Let u0 , b0 ∈ H m (Ω) with ∇ · u0 = 0 and ∇ · b0 = 0. Suppose that
u0 , b0 ∈ H m , div u0 = div b0 = 0 in Ω and u · n = b · n = 0 on ∂Ω. Then there exists a
positive time Tmax > 0 such that problem (1.1)–(1.3) has a unique solution (u, b) satisfying
u, b ∈ L∞ (0, Tmax ; W m,p ) for m > 1 + p3 with p ≥ 1.
Remark 2.2 In Theorem 2.1, the proof of the uniqueness of local (smooth) solution is
omitted because the argument of the proof is standard (e.g. [13]). It seems that a proof of local
existence of a smooth solution to (1.1)–(1.3) is well known result and thus we give a sketch of
proof.
Remark 2.3 Theorem above is also hold in higher dimensions with the modification on
the assumption m > 1 + Np .
Theorem 2.4 Let Ω be a bounded domain with smooth boundary ∂Ω. Assume that u
is a solution to the ideal 3D MHD equations (1.1)–(1.3) in the class C([0, T ); H m (Ω) for some
integer m ≥ 3 and for any 0 < T < ∞. If T is maximal, i.e., (u, b) cannot be continued to the
solution in the class C([0, T ′ ); H m (Ω) for any T ′ > T , then
Z T
(kω(t)kbmoβ (Ωε ) + kj(t)kbmoβ (Ωε ) ) + (kω(t)kMΘ (Ωε ) + kj(t)kMΘ (Ωε ) )dt = ∞, (2.1)
0

where ω := curl u, j := curl b and for all ε > 0 and Ωε := {x ∈ Ω; dist(x, ∂Ω) < ε}, ε is a small
positive constant depending only on Ω.
Now let us introduce the function space of bmo type used in [12]. We state the necessary
Definitions using the notation borrowing [12].
Definition 2.5 Let β(r) be a positive function on (0, 1] and Ω ⊂ R3 be a domain with
∂Ω ∈ C ∞ .
No.6 J.M. Kim: LOCAL EXISTENCE AND BLOW-UP CRITERION 1761

(1) bmoβ (R3 ) is defined as the set of functions f in L1loc (R3 ) such that
Z
1
kf kbmoβ (R3 ) := sup |f (y) − f¯B(x,r) |dy
0<r<1,x∈R3 |B(x, r)|β(r) B(x,r)
Z
1
+ sup |f (y)|dy ≤ ∞,
x∈R3 |B(x, 1)| B(x,1)
R
where f¯B := |B|
1
B f (y)dy.
(2) On Ω ⊂ R3 we define bmoβ as restrictions of the above space bmoβ (R3 ):

bmoβ (Ω) := {f |Ω ; f ∈ bmoβ (R3 )},

where f |Ω is the restriction of f on Ω. The norm of this space is defined by


kf kbmoβ (Ω) := inf kf˜kbmoβ (R3 ) ; f˜ ∈ bmoβ (R3 ) with f˜ = f in Ω .

In particular if β(r) = 1, we write bmoβ (R3 ) = bmo(R3 ) and bmoβ (Ω) = bmo(Ω).
Definition 2.6 Let Θ(α)(≥ 1) be a nondecreasing function on [1, ∞).

YΘ (Ω) := {f ∈ L1 (Ω) : kf kYΘ (Ω) < ∞},

where
kf kLp
kf kYΘ (Ω) := sup .
p≥1 Θ(p)

MΘ (Ω) := {f ∈ L1 (Ω) : kf kMΘ (Ω) < ∞},

where  Z 
1 3
kf kMΘ(Ω) := sup sup r−3+ p |f (y)|dy .
p≥1 Θ(p) 0<r<1,x∈R 3 B(x,r)∩Ω

We note that these spaces have the following relations.

kf kMΘ (Ω) ≤ Ckf kYΘ(Ω) ≤ Ckf kbmo(Ω) . (2.2)

Let
Θ(log(e + 1r ))
β(r) := .
log(e + 1r )
In this paper, we assume that the following hypotheses.
(H1) Θ(α) is a positive and nondecreasing function on [1, ∞) satisfying
Z ∞
dα
= ∞, Θ(α) ≥ α. (2.3)
1 Θ(α)
(H2) For all s ≥ 1 there exists C(s) such that

Θ(sα) ≤ C(s)Θ(α) for all α ≥ 1.

(H3) β(r) is a non-increasing function on (0, 1].

Since β(r) ≥ 1, we note that kf kbmoβ (Ω) ≤ kf kbmo(Ω) by the definition. Using this inequal-
ity (2.2) and condition (2.1) in Theorem 2.4, we have
Corollary 2.7 Assume the same hypotheses on Theorem 2.4. If Tmax is maximal, then
Z T
kω(t)kbmo(Ω) + kj(t)kbmo(Ω) dt = ∞. (2.4)
0
1762 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

2.2 Useful Inequalities

We review some useful estimates for the proof of theorems.
We state the bilinear commutator and the product estimate in [4].
Lemma 2.8 Let Ω ⊂ R3 be a bounded domain with ∂Ω ∈ C ∞ and s ≥ 1.
(1) If f, g ∈ H s (Ω) ∩ C(Ω), then


kf gkH s (Ω) ≤ C kf kH s (Ω) kgkL∞ (Ω) + kf kL∞ (Ω) kgkH s (Ω) .

(2) If f ∈ H s (Ω) ∩ C 1 (Ω) and g ∈ H s−1 (Ω) ∩ C(Ω), then for |α| ≤ s,


kDα (f g) − f Dα gkL2 (Ω) ≤ C kf kH s (Ω) kgkL∞(Ω) + kf kW 1,∞ (Ω) kgkH s−1 (Ω) .

Lemma 2.9 (see [12]) Let Ω ⊂ R3 be a bounded domain with ∂Ω ∈ C ∞ . For all ε > 0,
we have


k∇ukL∞ (Ω) ≤ C 1 + kukL2 (Ω) + k curl ukbmoβ (Ω) + k curl ukMΘ (Ωε) Θ(log(e + kukH 3 (Ω) ))

for all u ∈ H 3 (Ω) with div u = 0 in Ω and u · ν = 0 on ∂Ω.

We state the well-known Osgood type lemma in [5] or [12].
RT
Lemma 2.10 Let ψ be nonnegative function on (0, T ) with 0 ψ(t)dt < ∞, let Θ(α) be
R +∞ dα
a positive and nondecreasing for α ≥ 1 and Θ(α) = ∞. Assume that v ∈ C([0, T )) and
Z t
0 ≤ v(t) ≤ v(0) + ψ(s)Θ(v(s))ds for all 0 ≤ t < T.
0

Then sup v(t) < ∞.

0<t<T
Lemma 2.11 If u, b and π satisfy equations (1.1)–(1.2) with the boundary condition
(1.3), then for each t > 0, for m > 25 ,

k∇πkH m (Ω) ≤ C(kuk2H m (Ω) + kbk2H m (Ω) )

and for m > 1 + p3 ,

k∇πkW m,p (Ω) ≤ C(kuk2W m,p (Ω) + kbk2H m,p (Ω) ).

Proof This proof is almost similar to that in [13, Lemma 1.2] and so we omit the proof.

Lemma 2.12 Let u, b ∈ H m (Ω) and π ∈ H m+1 (Ω) with m ≥ 3. Then

k∇πkH m (Ω) ≤ C(kukH m k∇ukW 1,∞ (Ω) + kbkH m k∇bkW 1,∞ (Ω) ).

Proof This proof is almost similar to that in [4, Lemma 2] and so we omit the proof. 

3 Proof of Theorem 2.1

Proof of Theorem 2.1 Before the proof of Theorem 2.1, we need the following the
estimates for the nonlinear operator F (u, v) := u · ∇v for the divergence free vector u, v (see [7,
Section 4] and [13, pp 36]):
(a)
kF (u, v)kH m ≤ CkukH m kvkH m+1 , m ≥ 2, u ∈ Hσm , v ∈ Hσm+1 .
No.6 J.M. Kim: LOCAL EXISTENCE AND BLOW-UP CRITERION 1763

(b)
k(F (u, v), v)kH m ≤ CkukH m kvk2H m , m ≥ 3, u, v ∈ Hσm .

(c)
k(F (u, v), v)kH 2 ≤ CkukH 3 kvk2H 2 , u, v ∈ Hσ3 .

• (In case of p = 2): next, we take the scalar product of equations (1.1)1 and (1.1)2 with
D u and Dα b, respectively, add them together and then sum the result over |α| ≤ m and use
α

the integration by parts on space, we obtain

1 d
(kuk2H m + kbk2H m ) = −((u · ∇)u, u)m + ((b · ∇)b, u)m
2 dt
−((u · ∇)b, b)m + ((b · ∇)u, b)m − (∇π, u)m
:= K1 + K2 + K3 + K4 + K5 . (3.1)

⊲ (Estimate of K1 ,K2 ,K3 ,K4 ): by the inequality above (b) and Young’s inequalities with
the divergence free conditions and boundary conditions (1.3), we have

K1 , K3 ≤ C(kuk3H m + kbk3H m ).

Also, by the Leibniz rule, we have

K2 + K4 = ((b · ∇)b, u)m + ((b · ∇)u, b)m

X  
= (Dα ((b · ∇)b), Dβ u) + (Dα ((b · ∇)u), Dβ b)
|α|≤m
X  X
= (b · ∇)Dα b, Dβ u) + Cα,β ((Dβ b · ∇)Dα−β b, Dβ u)
|α|≤m 0<β≤α
X  X
α β
+ (b · ∇)D u, D b) + Cα,β ((Dβ b · ∇)Dα−β u, Dβ b)
|α|≤m 0<β≤α
X X
= Cα,β ((Dβ b · ∇)Dα−β b, Dβ u) + Cα,β ((Dβ b · ∇)Dα−β u, Dβ b)
0<β≤α 0<β≤α

≤ C(kuk3H m + kbk3H m ) (3.2)

(see [13, pp. 36–37] or [7] for a detailed proof).

⊲ (Estimate of K5 ): by Lemma 2.11, this term is estimated as follows:

K5 ≤ C(kuk2H m + kbk2H m ).

With aid of the estimate K1 –K5 , estimate (3.1) becomes for m > 52 ,
d
(ku(t)k2H m + kb(t)k2H m )dx ≤ C(ku(·, t)k3H m + kb(·, t)k3H m ) + C(ku(·, t)k2H m + kb(·, t)k2H m ).
dt
Set X(t) := ku(·, t)k2H m (Ω) + kb(·, t)k2H m (Ω) . So, we have
d
X(t) ≤ CX(t) + CX 3/2 (t).
dt
Therefore, due to non linear Gronwall’s inequality, one can get the local existence result based
on the analysis in [10]. Therefore, we complete the proof in this case.
• (In case of p 6= 2): applying the operator Dα to the first and second equation (1.1) by
|D u| D u and |Dα b|p−2 Dα b, respectively, and we integrate them over Ω, and add these
α p−2 α
1764 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

equations for |α| ≤ m, we obtain

1 d X
(kukpW m,p + kbkpW m,p ) = − (Dα [(u · ∇)u], |Dα u|p−2 Dα u)
p dt
|α|≤m
X
+ (Dα [(b · ∇)b], |Dα u|p−2 Dα u)
|α|≤m
X
− (Dα [(u · ∇)b], |Dα b|p−2 Dα b)
|α|≤m
X
+ (Dα [(b · ∇)u], |Dα b|p−2 Dα b)
|α|≤m
X
− (Dα ∇π, |Dα u|p−2 Dα u)
|α|≤m

≤ C(kuk3W m,p + kbk3W m,p ),

here for the convention terms, we use the following the estimate type:
X
(Dα [(u · ∇)u], |Dα u|p−2 Dα u) ≤ Ckuk3W m,p .
|α|≤m

Similarly, we get
X
(Dα [(b · ∇)b], |Dα u|p−2 Dα u) ≤ C(kuk3W m,p + kbk3W m,p ),
|α|≤m
X
(Dα [(u · ∇)b], |Dα u|p−2 Dα b) ≤ C(kuk3W m,p + kbk3W m,p ),
|α|≤m

and
X
(Dα [(b · ∇)b], |Dα u|p−2 Dα b) ≤ C(kuk3W m,p + kbk3W m,p )
|α|≤m

(see e.g. [13, pp. 38–39] for a detailed proof). Also, for the total pressure, we use Lemma 2.11

k∇πkW m,p (Ω) ≤ C(kuk2W m,p (Ω) + kbk2H m,p (Ω) ).

As like as the argument of previous case, we set Z(t) := ku(·, t)k2W m,p (Ω) + kb(·, t)k2W m,p (Ω) . So,
we have
d
Z(t) ≤ CZ 3/2 (t) + CZ(t).
dt
Therefore, due to Gronwall’s inequality with the analysis in [10], we get the local existence
result. Therefore, we complete the proof in this case. 

4 Proof of Theorem 2.4

Proof of Theorem 2.4 Following the argument in [12], using the standard argument
of continuation of local solutions, it suffices to prove that
Z T
kω(t)kbmoβ (Ωε ) + kω(t)kMΘ (Ωε ) dt < ∞, (4.1)
0

then
sup (ku(t)kH m (Ω) + kb(t)kH m (Ω) ) < ∞.
0<t<T
No.6 J.M. Kim: LOCAL EXISTENCE AND BLOW-UP CRITERION 1765

From (3.1), we know that

1 X d
(kuk2H m + kbk2H m ) + (u · Dα ∇u, Dα u)L2 − (b · Dα ∇b, Dα u)L2
2 dt
|α|≤m

+(u · Dα ∇b, Dα b)L2 − (b · ∇Dα u, Dα b)m

= (u · Dα ∇u − Dα (u · ∇u), Dα u)L2 − (b · Dα ∇b − Dα (u · ∇b), Dα u)L2
+(u · Dα ∇b − Dα (u · ∇b), Dα b)L2 − (b · Dα ∇u − Dα (b · ∇u), Dα b)L2 − (Dα ∇π, Dα u)L2 .

Since the boundary condition and divergence free condition, we have

1 X d
(kuk2H m + kbk2H m ) ≤ kDα ukL2 [ku · Dα ∇u − Dα (u · ∇u)kL2 + kDα ∇πkL2 ]
2 dt
|α|≤m

+kDα ukL2 kb · Dα ∇b − Dα (b · ∇b)kL2

+kDα bkL2 (ku · Dα ∇b − Dα (u · ∇b)kL2
+kb · Dα ∇u − Dα (b · ∇u)kL2 .

Using Lemmas 2.8, 2.12 and Young’s inequality, we have

1 X d
(kuk2H m + kbk2H m ) ≤ CkukH m [kukH m kukW 1,∞ + kbkH m kbkW 1,∞ + k∇πkH m ]
2 dt
|α|≤m

+CkbkH m (kukH m kbkW 1,∞ + kbkH m kukW 1,∞ )

≤ C(kuk2H m + kbk2H m )(kukW 1,∞ + kbkW 1,∞ )
+CkukH m k∇πkH m
≤ C(kuk2H m + kbk2H m )(kukW 1,∞ + kbkW 1,∞ ).

By Grownwall’s inequality, we have

 Z t 
(ku(t)k2H m + kb(t)k2H m ) ≤ (ku0 k2H m + kb0 k2H m ) exp C kukW 1,∞ + kbkW 1,∞ dτ
0

which yields
Z t
log(e + ku(t)k2H m + kb(t)k2H m ) ≤ ku0 kH m + kb0 kH m + C kukW 1,∞ + kbkW 1,∞ dτ.
0

Hence, it follows from Lemma 2.10 with hypothesis (2.3) and Lemma 2.9, (4.1) implies

sup (log(e + ku(t)k2H m + kb(t))k2H m ) < ∞.

0<t<T

This complete the proof. 

5 Conclusion Remark

Our arguments also allow to prove Theorems 2.1 and 2.4 for the incompressible ideal vis-
coelastic equations (see e.g. [6, 16]). This system arises in the Oldroyd model for ideal vis-
coelastic flow has the property of materials that exhibit both viscous and elastic characteristics
1766 ACTA MATHEMATICA SCIENTIA Vol.38 Ser.B

when undergoing deformation (see e.g. [8], [9] and references therein for physical background):

 XN



 ∂t u + u · ∇u + ∇P = Fk · ∇Fk ,



 k=1
∂t Fk + u · ∇Fk = Fk · ∇u, in QT := Ω × [0, T ) (5.1)



 div u = 0 and div Fk = 0,




 u(x, 0) = u0 (x) and Fk (x, 0) = F0k (x)
with k = 1, · · · , N , here u = u(x, t) ∈ Ω represents the fluid’s velocity, p = p(x, t) ∈ Re
represents the fluid’s pressure, and F = F (x, t) ∈ Ω × Ω represents the local deformation tensor
of the fluid. Assume that u and F satisfy the initial conditions

u(x, 0) = u0 (x) and F (x, 0) = F0 (x), x∈Ω

together with the boundary conditions defined as follows:

u·n=0 and F · n = 0,
Fij
where n is the outward unit normal vector along boundary ∂Ω. We denote (∇ · F )i = ∂xj for
a matrix F , in the (i, j)-th entries, i, j = 1, 2, 3.

References
[1] Cannone M, Chen Q, Miao C. A losing estimate for the ideal MHD equations with application to blow-up
criterion. SIAM J Math Anal, 2007, 38(6): 1847–1859
[2] Caflisch R E, Klapper I, Steele G. Remarks on singularities, dimension and energy dissipation for ideal
hydrodynamics and MHD. Comm Math Phys, 1997, 184(2): 443–455
[3] Du Y, Liu Y, Yao Z. Remarks on the blow-up criteria for three-dimensional ideal magnetohydrodynamics
equations. J Math Phys, 2009, 50(2): 023507, 8 pp
[4] Ferrari A B. On the blow-up of solutions of 3-D Euler equations in a bounded domain. Comm Math Phys,
1993, 155(2): 277–294
[5] Hartman Ph. Ordinary Differential Equations. 2nd ed. Boston, Basel, Stuttgart: Birkhäuser, 1982
[6] Hu X, Hynd R. A blowup criterion for ideal viscoelastic flow. J Math Fluid Mech, 2013, 15(3): 431–437
[7] Kato T. Nonstationary flows of viscous and ideal fluids in R3 . J Funct Anal, 1972, 9: 296–305
[8] Gurtin M E. An Introduction to Continuum Mechanics, Mathematics in Science and Enineering. Vol 158.
New York: Academic Press, 1981
[9] Larson R G. The Structure and Rheology of Complex Fluids. New York: Oxford University Press, 1995
[10] Majda A. Compressible fluid flow and systems of conservation laws in several space variables. New York:
Springer, 1984
[11] Mahalov A, Nicolaenko B, Golse F. Non blow-up of the 3D ideal magnetohydrodynamics equations for a
class of three-dimensional initial data in cylindrical domains. Zap Nauchn Sem S -Peterburg Otdel Mat
Inst Steklov (POMI), 2004, 318: 203–219; translation in J Math Sci (NY), 2006, 136(2): 3768–3777
[12] Ogawa T, Taniuchi Y. On blow-up criteria of smooth solutions to the 3D Euler equations in a bounded
domain. J Differential Equations, 2003, 190(1): 39-63
[13] Temam R. On the Euler equations of incompressible perfect fluids. J Funct Anal, 1975, 20(1): 32–43
[14] Vishik M. Incompressible flows of an ideal fluid with unbounded vorticity. Comm Math Phys, 2000, 213(3):
697–731
[15] Zhang Z-F, Liu X-F. On the blow-up criterion of smooth solutions to the 3D ideal MHD equations. Acta
Math Appl Sin, Engl Ser, 2004, 20(4): 695–700
[16] Zhu L, Wang F. A new blowup criterion for ideal viscoelastic flow. J Partial Differ Equ, 2015, 28(1): 23–29