CC BY
30
MSC 47B10, 34L40; secondary 35M10
ON FACTORIZATION OF A DIFFERENTIAL OPERATOR ARISING IN FLUID DYNAMICS
M. Chugunova, Institute of Mathematical Sciences, Claremont Graduate University, Claremont, USA, marina.chugunova@cgu.edu,
V. Strauss, Department of Pure & Applied Mathematics, Simon Bolivar University, Caracas, Venezuela, str@usb.ve
uniform coating steady state results into the operator which domain couples two subspaces spanned by positive and negative Fourier exponents which are not invariant subspaces of the operator. We prove that the operator admits factorization and use this new representation of the operator to prove compactness of its resolvent and to find its domain.
Keywords: factorization, lubrication approximation, fluid mechanics, forward-backward heat equation.
Introduction
Depending on the parameters involved, the dynamics of the film of viscous fluid can be
liquid on the inner surface of a cylinder rotating in a gravitational field was based on the lubrication approximation and examined by Benilov, O’Brien, and Sazonov [2, 3]. The related Cauchy problem has the following form:
Eigenmode solutions are very important in stability analysis, because even a single growing mode can destabilize an otherwise stable system. In case when all modes are bounded in time and the corresponding eigenfunctions form a complete set, the system normally regarded as a stable one. Because, an arbitrary initial condition can be represented as a series of these eigenmodes; and since all of them are stable, so expected to be the solution to the initial- value problem.
There are however counterexamples to the arguments above when each term of the series is bounded but the series as a whole diverges and the solution develops a singularity in a finite time. This effect was observed by Benilov, O’Brien, and Sazonov [2] for the problem 0.1 when parameter in (0.2) a = 0. For this case when the effect of gravitational drainage was neglected because of infinitesimally thin film they studied stability of the problem asymptotically and numerically. It was shown that even for infinitely smooth initial values numerical solutions blow up after a small number of iterations.
The spectrum of the linear operator L that is defined by the operation l[.] and periodic boundary conditions y(-n) = y(n) for the special case when the parameter a = 0 was studied
Spectral properties of linear operators are very important in stability analysis of dynamical systems. The paper studies the non-selfadjoint second order differential operator that originated from a steady state stability problem in dynamic of viscous Newtonian fluid on the inner surface of horizontally rotating cylinder in the presence of gravitational field. The linearization of the thin liquid film flow in the lubrication limit about the
described by different asymptotic equations. Under the assumption that the film is thin enough for viscous entrainment to compete with gravity, the time evolution model of a thin film of
yt + %] = 0, y(0,x) = yo, y(-n,t) = y(n,t), x £ \-n,n], t> 0 (0.1)
where
(0.2)
rigorously in [8, 6, 9]. Using different approaches they justified that if the parameter b restricted to the interval [0, 2] then the operator L is well defined in the sense that it admits closure in L2( —n, n) with non-empty resolvent set without breaking the boundary conditions y(—n) = y(n). The spectrum of the operator L is discrete and consists of simple pure imaginary eigenvalues only. As a result all eigenfunctions have the following symmetry y\(—x) = y\(x). The more general operator with the function sin(x) replaced by the arbitrary 2n-periodic functions was studied in [4] and it was proved that this operator multiplied by i belongs to a wide class of PT-symmetric operators which are not similar to self-adjoint but nevertheless possesses purely real spectrum due to some obvious and hidden symmetries.
The phenomenon of the coexistence of the neutrally stable modes with explosive instability of the numerical solutions [2] (which correspond to drops of fluid forming on the ceiling of the cylinder where the effect of the gravity is the strongest) was studied analytically and explained in terms of the absence of the Riesz basis property of the set of eigenfunctions in [5]. The question of a conditional basis property of the set of eigenfunction is still open.
For the case when a = 0, as it was discussed in [3], the spectral properties of the operator L are not expected to differ a lot from the case a = 0.
The goal of this paper is to find a factorization of the operator L (under some restrictions on parameters a and b) that would be in some sense similar to one we constructed for the special case a = 0 in [7] (in this case the operator L is J-self-adjoint with the operator J defined as a shift J(f (x)) = f (n — x)) and to examine some properties of the operator L using this factorization. The main difficulty to overcome here is an existing coupling between two subspaces spanned by positive and negative Fourier exponents which are not invariant subspaces of the operator L if a = 0. We also prove that the non-self-adjoint differential operator L has compact resolvent and as result spectrum of L is discrete with the only accumulating point at infinity.
1. Factorization of the non-self-adjoint operator L
We denote by D(T) and R(T) the domain and the range of linear operator T respectively. The notation L2 is used for the standard Lebesgue space of scalar functions defined on the interval (—n,n). From here on L is the indefinite convection-diffusion operator L :
, (Ly)(x) = d— ((1 — a cos x)y(x) + b sin x ■ \ , b = 0
d ^ \ d x j
with the domain of all absolutely continuous 2n-periodic functions y(x) such that (Ly)(x) G L2. The latter means, in particular, that we consider y(x) such that the function
(1 — a cos x)y(x) + b sin x ■ \ ' ' |
dx I
dy(x)\
x
can be defined at zero as a continuous function and by this way it is converted in an absolutely continuous function.
In addition, we define the operator S :
L2 ^L2, (Sy)(x) = y’(x),
where y'(x) G L2, y(—n) = y(n), and the operator M : L2 ^ L2,
dy(x)
(My)(x): = (1 — a cosx)y(x) + b sinx ■
dx
with the domain of all functions y(x) G L2 absolutely continuous on (—n, 0) U (0, n) and such that (My)(x) G L2. Note, that, for example, y(x) = x-1/3 G D(M). The operator M can also be
represented by the following expression
(My)(x) = (1 — (a + b) cos x)y(x) + b (sinx ■ y(x))'.
Theorem 1. If the parameters a and b satisfy the inequality 2a + b < 2, then L is a closed operator with a closed range and L = SM.
Proof. Let us consider the operator A :
L2 ^ L2, (Ay)(x) = (sin(x)y(x))>
with D(A) = {y(x) | y(x), (Ay)(x) G L2}. Then D(A) = D(M) and a function y(x) G D(A) can be written as
y(x) = . 1 ) ■ f(c(1 + Signx)/2 + ci(1 — Signx)/2) + f d(t)dt\ d(t) G L2. (1.1)
sin(x)0
If x > 0 then
rx
| / 6(t)dt\ < a(x) ■ x1/2, Jo
where a(x) = (fo ld(x)l2dx1/2. Since the two summands in (1.1) have different growth orders as x 0 this implies that if y(x) G L2 then c = 0 and
1 fx
y(x) = nnx) i mdt (L2)
Moreover,
x1/2
ly(x)l £ snx) ■a(x) (L3)
A small modification of the same reasoning leads to the following estimation for every x G (—n, n)
\r\ 1/2
|y(x)l - Timex)! ■a(x)- (L4)
with a(x) = | fX l&(x)l2dx\1/2.
Alternatively the same function y(x) can be written as
1 fn
y(x) = . ( ) ■ (c — 0(t)dt), d(t) G L2. (1.5)
sin(x) x
with the same Q(x) as in (1.1). Representation (1.5) yields the following relations
-1 r
Vix) = lMx) ' Jx. ^ (L6)
and
(17)
with f3(x) = | fX ld(x)l2dxl1/2.
It follows from (1.2) and (1.6) that
/ Q(t)dt = 0. (1.8)
/о
Starting from the point —n one can also obtain that
1x
y(x) = snx' L mdL (L9)
< {~и+(х)\2'lix> (L10)
with y(x) = (f-n \d(x)\2dx\1/2 and
/ 9(t)dt = 0. (1.11)
Now we are ready to calculate M*. Using smooth functions y(x) such that y(x) = 0 in some neighborhoods of the points —n, 0 and n (neighborhoods depend of y(x)) it is easy to show that
(M*z)(x) = (1 — a cos x)z(x) — b (sin x ■ z(x))'
for every z(x) G D(M*). Since the condition z(x) G D(M*) yields
z(x) G L2 and (M*z)(x) G L2, (1.12)
z(x) G D(A) and for z(x) the conditions of the type (1.4), (1.7) and (1.10) are satisfied. So,
lim y(x)z(x)sin x = lim y(x)z(x)sin x = lim y(x)z(x)sin x = 0 .
x^O x^—n+0 x^n—0
Taking into account the latter one can check that
(My, z) = (y, M#z)
for every y(x),z(x) G D(M) = D(A), where D(M#) = D(A) and M# is defined by the same differential expression as M*. Thus, D(M) = D(M*). The same reasoning shows that M** = M, so M is closed.
Let
(1 — a cos x)y(x) + b sinx ■ dy(x') = u(x), y(x), u(x) G L2. dx
Our aim is to express y(x) via u(x). Let x G (—n, n), x = 0. Then y(x) =
(c(1 + Signx)/2 + c1(1 — Signx)/2) ■ (sin lx\)a/b ■ (cot lxl/2)l/b +
1 f x
-(sin lxl)a/b ■ (cot lxl/2)1/b u(t)(sin ltl)-% ■ (sint)—1 ■ (tan \tl/2)l/bdt,
b0 where c and c1 are constants. The estimations that follow closely depend of a relation between a and b. We assume that 2a + b < 2. Then for x > 0
rx rx
u(t)(sint)-(b+1) ■ (tant/2)1/bdt = v(t)(t)—b+1) ■ (t/)1/bdt,
Jo Jo
where v(t) = u(t)(sint)~(ь+1) ■ (tant/2)1/b(t)(b+1) ■ (t)~1/b, so,
x
\ I u(t)(sint)-(b+1) ■ (tant/2)1/bdt\ <
Jo
Thus, the first summand (if c = 0) for y(x) has the order x % and the second one has the order
x—1/2a(x) with lim a(x) = 0. Since y(x) G L2, c = 0. The same reasoning shows that c1 = 0.
x—0
Thus,
1 f x
y(x) =-(sin lxl)a/b ■ (cot lxl/2)1/b u(t)(sin ltl)-% ■ (sin t)—1 ■ (tan ltl/2)1/bdt. (1.13)
b0
In particular, for u(x) = 1 we have 1
b jo
1 f x
yo(x): =-(sin\x\)a/b ■ (cot \x\/2)1/b (sin\t\)—b ■ (sint)—1 ■ (tan\t\/2)1/bdt.
b0
Some elementary estimations show that there are finite limits lim yo(x), lim yo(x) and lim yo(x)
x—0 x———n x——n
with lim yo(x) = lim yo(x). Let us show these relations. First, for t > 0 we define w(t): =
w(t) = (1/2)l/b. Moreover, for x
b 1 ■ fta,n{t/2)^i/b. Then liniow(t) = (1/2)l/b. Moreover, for x > 0
y0(x): = b (sin x)a/ ■ (cot x/2)l/ J w(t)t— %— 1+1/bdt,
so
yo(x) = 1 (sinx)a/b ■ (cot x/2)1/bY^-w(Cx)(x)—%+1/b, where £x G (0, x). The latter yields y(0): = lim y(x) = . Second, for t < n we define
x—+0 1 a
w+(t): = () —b — 1 ■ (■ tan(t/2))1/b. Then lim w+(t) = 1 and for z(x): = (1 + a) ■
t——+n—0
2
y0(x) ■ (sin(x/2))b we have
z(x) = 1 + a(sinx)~+~ ■ f w+(t)(n — t) — %—1—1/bdt. bo
Let us fix e > 0 Then there is 5 > 0 such that 1 — e < w+(x) < 1 + e for every x G (n — 5, n). Next, for the same x
z(x) = (sinx)■ ( ( w+(t)(n — t) — %—1—1/bdt+
bo
r x
w+(t)(n — t)—%—1—1/bdt).
J n—S
n—S
Since
[ w+(t)(n — t)—%—1—1/bdt) = —+—w(vx,s )((n — x) — — 5—^
Jn—S 1 +a
with vx,s G (n — 5,n),
—b— (1 — e)((n — x) — ~+~ — 5—~+~) < f w+(t)(n — t) — %—1—1/bdt) <
1 + a Jn-S
1 + a
Moreover, for fixed 5
rn—S
1 + a rn—o
lim —-— (sin x)ab ■ I w+(t)(n — t) — b—1—1/bdt = 0
t^+n—0 b Jq
and
a+1 1 + a
lim (sinx) % ■ w(vxs)5 % = 0,
t—+n—0 ’
so the equality lim yo(x) = 1++a is practically evident. Note also that yo(x) is even.
Thus, the function yo(xf is continuous on [—n,n] and satisfies the periodic conditions.
Now let u(x) = c + J! $(t)dt, where c = const and $(x) G L2. Then for x > 0:
l IX $(t)dtl < x1/2( fx l$(t)l2)1/2. The latter estimation and (1.13) yield lim y(x) = c ■ y0(0). The same function can be re-written as following u(x) = c+ + fX $(t)dt or u(x) = c — + fxn $(t)dt. Then the estimations l fX $(t)dtl < (n — x)l/2( — fX l$(t)l2)1/2, x G (o,n) and l f-n ^(t)dtl < (n + x)1/2( f'x l$(t)l2)1/2, x G (—n,o) together with Representation (1.13) yield lim y(x) = c+ ■ y0(n)
n x—n
and lim y(x) = c — ■ yo(—n). Thus, y(x) satisfies the periodic condition if and only u(x) satisfies.
x——n
The latter yields the equality
L = SM. (1.14)
Moreover, we have shown that for every u(x) = c + fX $(t)dt with $(x) G L2 there is absolutely continuous on [—n,n] function y(x), such that (My)(x) = u(x), so R(M) is dense in L2 or, equivalently, Ker(M*) = {0}.
Note, that M is boundedly invertible. Indeed, M = D + iC, where C: L2 ^ L2,
(Cy)(x): = ^ 2 cos xy(x) — b (sin x ■ y(x))'^, D(C) = D(M)
and D: L2 ^ L2,
(Dy)(x): = (1 — (a + 2) cosx)y(x).
Since D is bounded, D(C) = {y(x) l y(x), (Cy)(x) G L2} and a similar reasoning shows that D(C) = D(A). Thus, if y(x),z(x) G D(C), then for y(x) and z(x) Conditions (1.7) and (1.10) hold true, so
lim sin(n — e)y(n — e)z(n — e) = lim sin(—n + e)y(—n + e)z(—n + e) = 0.
e\0 e\0
Since C* is defined by the same differential expression as C, the above equalities show that D(C*) = D(A). Thus, C is self-adjoint. Moreover, D is positive and boundedly invertible, so the problem of invertibility of M is equivalent to the problem of regularity of non-real numbers for a self-adjoint operator (for a more detail reasoning see, for instance, [7]).
Now let us prove that L is closed. The operator S restricted to the subspace L1 C L2 of functions orthogonal to constants has a bounded inverse. Let us find M—1(L1). If (My)(x) G L1, then f-n(My)(x)dx = 0, but f-n(y(x)sinx)'dx = 0, so y(x) G M—1(L1) if and only if
y(x) G D(M) and f-n (1 — (a + b) cosx)y(x)dx = 0. Let L2 : = |(1 — (a + b) cos. Since
for y0(x) we have 2n = (My0)(x)dx = y0(x){1 — (a + b) cos x)dx, y0(x) G L2 and
L2 = L2+{^ ■ y0(x)}^c. Now let a sequence {yk(x)} be such that yk(x) ^ y(x) and zk(x) = (Lyk)(x) ^ z(x). Then (Myk)(x) = Vk(x) + ck, where Vk(x) = ((S^)— 1z)(x), ck = const, k = 1, 2,... . Since (S\c1 )— 1 is bounded, the sequence {vk (x)} has a limite v(x). In turn, in virtue of similar reasons the sequence wk(x) = (M—1Vk)(x) also has a limite. Simultaneously yk(x) = (M— 1(vk + Ck))(x) = wk(x) + ckyo(x). Thus, the sequence {ck} has a limite. The rest is straightforward.
□
Corollary 1. If the parameters a and b satisfy the inequality 2a + b < 2, then the set D(L) is the linear sub-manifold H of the Sobolev space H 1(—n,n):
D(L) = H : f G H 1(—n,n), f (n) = f (—n,) sin(x)f' G H 1(—n,n)
and is a Hilbert space with the norm defined as:
llf ll2 = llf ll2H 1 + ll sin(x)ff(x)ll2Hi.
The reasoning of this corollary is the same as the reasoning of the corresponding proposition in
[7].
For the next step we need the following simple remark.
Lemma 1. Let H be a Hilbert space, x1 ,x2 GH, (x1,x2) = 0. Let H1: = {x1 }\ H2 : = {x2}±. Let P1 and P2 be ortho-projections onto the subspaces H1 and H2 respectively. Then P2\h1 is one-to-one mapping onto H2.
Proof. Let H3: = H1 HH2. Without loss of generality we can assume that ||x1|| = \\x2W = 1, (x1, x2) = a> 0. Then H1 = {^ ■ (x2 — a ■ x^^ec ©H3, H2 = {^ ■ (x1 — a ■ x2)}^ec ©H3. Since P2H3 = Ih3 and P2(x2 — a ■ x1) = a ■ (x1 — a ■ x2), the rest is evident.
□
Theorem 2. If the parameters a and b satisfy the inequality 2a + b < 2, then the resolvent of L is compact of the Hilbert-Schmidt type.
Proof. Let Lo C L2 be the subspace of constants, L1: = L$ . Since Lo C D(L) and R(L) = L1, the operator L has the following matrix representation
L
0
L1Q
0
L11
with respect to the decomposition L2 = Lo ©L1, where the operator L10: 1 ^ a ■ sinx is bounded and D(Ln) = {f (x) lf G Hl(—n,n), f (n) = f (—n), sin(x)f'(x) G Hl(—n,n), — f (x)dx = 0}. Let us analyze the properties of L11. From Theorem 1 we have L11 = SM\l1 . Let L2 = M(D(M) n L1), y0(x) = (M—HXx), z0(x) = ((M*)—HXx). Then for y(x) G D(M) n L1 we have 0 = (y, 1) = (y,M*z0) = (My,z0), so L2 = {z0}x. From the other hand, (1,z0) = (My0,z0) = (yo, M*zo) = (yo, 1). Since (see the proof of Theorem 1) (yo, 1) = 0, the pair {1, zo} is under the conditions of Lemma 1. Let P1 be the ortho-projection onto L1. Then L11 = S ■ (P1L2) ■ (MlL1),
so L—11 = (M 1\l2) ■ (P1\l1 ) 1 ■ (51^) 1. Thus, L—11 is an operator of the Hilbert-Schmidt type.
1
Since
the rest is straightforward.
Rx(L) =
R\(0) 0
L JRX(L11)L1Q RX(L11)
□
Acknowledgement. This work was partially supported by a grant from the Simons Foundation (# 275088 to Marina Chugunova)
References
1. Azizov T.Ya., Iokhvidov I.S. Linear Operators in Spaces with Indefinite Metric. N.Y., Wiley, 1989.
2. Benilov E.S., O’Brien S.B.G., Sazonov I.A. A New Type of Instability: Explosive
Disturbances in a Liquid Fild Inside a Rotating Horizontal Cylinder. J. Fluid Mech., 2003, vol. 497, pp. 201-224.
3. Benilov E.S., Kopteva N., O’Brien S.B.G. Does Surface Tension Stabilise Liquid Films Inside a Rotating Horizontal Cylinder. Q. J. Mech. Appl. Math., 2005, vol. 58, pp. 158-200.
4. Boulton L., Levitin M., Marletta M. On a Class of Non-self-adjoint Periodic Eigenproblems with Boundary and Interior Singularities. J. of Differential Equations, 2010, vol. 249, no. 12, pp. 3081-3098.
5. Chugunova M., Karabash I.M., Pyatkov S.G. On the Nature of Ill-posedness of the Forward-Backward Heat Equation. Integral Equations and Operator Theory, 2009, vol. 65, pp. 319-344.
6. Chugunova M., Pelinovsky D. Spectrum of a Non-Self-Adjoint Operator Associated with the Periodic Heat Equation. J.Math.Anal.Appl., 2008, vol. 342, pp. 970-988.
7. Chugunova M., Strauss V. Factorization of the Indefinite Convection-Diffusion Operator. C. R. Math. Rep. Acad. Sci. Canada, 2008, vol. 30, no. 2, pp. 40-47.
8. Davies E.B. An Indefinite Convection-Diffusion Operator. LMS J. Comp. Math., 2007, vol. 10, pp. 288-306.
9. Weir J. An Indefinite Convection-Diffusion Operator with Real Spectrum. Appl. Math. Lett., 2008, vol. 22, pp. 280-283.
УДК 517.984
О ФАКТОРИЗАЦИИ ОДНОГО ДИФФЕРЕНЦИАЛЬНОГО ОПЕРАТОРА, ВОЗНИКАЮЩЕГО В ГИДРОДИНАМИКЕ
М.В. Чугунова, В.А. Штраус
Спектральные свойства линейных операторов играют важную роль в анализе устойчивости динамических систем. В заметке исследуются свойства несамосопряженного дифференциального оператора второго порядка, связанного с исследованием проблемы устойчивости стационарного динамического состояния тонкой пленки, образованной вязкой ньютоновской жидкостью и расположенной на внутренней поверхности вращающегося цилиндра, при наличии гравитационного поля. Линеаризация по малому параметру (отношению толщины потока к размеру цилиндра) в этом случае порождает дифференциальный оператор с областью определения, вложенной в прямую сумму двух подпространств, натянутых, соответственно, на базисы {етх} и {е-тх} (п > 0), причем указанные подпространства не являются инвариантными по отношению к оператору, и одномерного подпространства констант. Доказывается, что этот оператор допускает представление в виде произведения двух дифференциальных операторов первого порядка. Полученное представление используется для доказательства компактности резольвенты исследуемого оператора и непосредственного описания его области определения.
Ключевые слова: спектральный анализ дифференциального оператора, факторизация, гидродинамика, прямое/обратное ура,внение теплопроводности.
Марина Васильевна Чугунова, PhD, профессор, математический институт, Кларемонт-ский университет, г. Кларемонт, США, marina.chugunova@cgu.edu.
Владимир Абрамович Штраус, доктор физико-математических наук, профессор, университет Симон Боливар, г. Каракас, Венесуэла, str@usb.ve.
Поступила в редакцию 28 мая 2013 г.
