INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS. I.

HYPOELLIPTIC DIFFUSION AND SUBDIFFUSION EQUATIONS

MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

Abstract. A class of inverse problems for restoring the right-hand side of a par-
abolic equation for a large class of positive operators with discrete spectrum is
considered. The results on existence and uniqueness of solutions of these problems
as well as on the fractional time diffusion (subdiffusion) equations are presented.
Consequently, the obtained results are applied for the similar inverse problems
arXiv:1812.11483v2 [math.AP] 8 Nov 2019

for a large class of subelliptic diffusion and subdiffusion equations (with continu-
ous spectrum). Such problems are modelled by using general homogeneous left-
invariant hypoelliptic operators on general graded Lie groups. A list of examples is
discussed, including Sturm-Liouville problems, differential models with involution,
fractional Sturm-Liouville operators, harmonic and anharmonic oscillators, Landau
Hamiltonians, fractional Laplacians, and harmonic and anharmonic operators on
the Heisenberg group. The rod cooling problem for the diffusion with involution
is modelled numerically, showing how to find a “cooling function”, and how the
involution normally slows down the cooling speed of the rod.

Contents
1. Introduction 2
2. Inverse problem for the heat equation 4
2.1. Statement of the problem 4
2.2. Proof of the existence result 5
2.3. Proof of the uniqueness result 6
3. Inverse problem for the time-fractional diffusion equation 7
3.1. Preliminaries 7
3.2. Proof of Theorem 3.6 9
4. Inverse time-fractional diffusion problem for the hypoelliptic operators 12
4.1. Graded Lie groups 12
4.2. Proof of Theorem 4.2 14
5. Appendix: Non-harmonic analysis of operators with discrete spectrum 16
6. Examples 17
7. Numerical illustrations 20
Date: November 12, 2019.
2010 Mathematics Subject Classification. 35K90, 42A85, 44A35.
Key words and phrases. Heat equation, time-fractional diffusion equation, inverse problem, self–
adjoint operator, Rockland operator.
The first author was supported in parts by the FWO Odysseus Project, EPSRC grant
EP/R003025/1 and by the Leverhulme Grant RPG-2017-151. The second author was supported
by the Ministry of Education and Science of the Republic of Kazakhstan Grant AP05130994. The
third author was supported by the Ministry of Education and Science of the Republic of Kazakhstan
Grant AP05131756. No new data was collected or generated during the course of research.

1
2 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

7.1. Conclusion 22
References 22

1. Introduction
Many instances are known in which the practical needs lead to the problems of
determining the coefficients or the right-hand-side of a differential equation from
some available data about the solution. These are called the inverse problems of
mathematical physics. Inverse problems arise in various areas of human activity such
as seismology, mineral exploration, biology, medicine, quality control of industrial
goods, etc. All these circumstances place inverse problems among the important
problems of modern mathematics.
The first purpose of this paper is to study inverse problems for the heat equation.
We consider the heat equation
(1.1) ut (x, t) + Lu(x, t) = f (x),
with Cauchy condition
(1.2) u(x, 0) = ϕ(x), x ∈ Q̄,
for (x, t) ∈ Ωt = {x ∈ Q ⊂ Rd , d ≥ 1, t ∈ [0, T ]}, where L is a linear self-adjoint
positive operator with a discrete spectrum {λξ > 0 : ξ ∈ I} on a separable Hilbert
space H. Here Q is a bounded domain with a smooth boundary or unbounded domain.
Respectively λξ , the operator L has the system of orthonormal eigenfunctions {eξ :
ξ ∈ I} on the separable Hilbert function space H.
The problem of determination of temperature at interior points of a region when
the initial and boundary conditions along with diffusion source term are specified are
known as direct diffusion conduction problems. In many physical problems, deter-
mination of coefficients or right hand side (the source term, in case of the diffusion
equation) in a differential equation from some available information is required; these
problems are known as inverse problems. These kind of problems are ill posed in
the sense of Hadamard. A number of articles address the analytical and numerical
solvability of the inverse problems for the diffusion and anomalous diffusion equa-
tions (see [CD98, CNYY09, JR15, KS10, OS12a, OS12b, TT17, FIK14, IC16, KM11,
KST17, NLN16, ZX11, ZW13, WYH13] and references therein).
The setting of a general operator L as in this paper allows one to include many
models. A number of physical examples are discussed in Section 6, including Sturm-
Liouville problems, differential models with involution, fractional Sturm-Liouville
operators, harmonic and anharmonic oscillators, Landau Hamiltonians, fractional
Laplacians, and harmonic and anharmonic operators on the Heisenberg group.
Section 3 is dedicated to finding the couple of functions (u(x, t), f (x)) satisfying
the equation
α
(1.3) D0+,t u (x, t) + Lu(x, t) = f (x) ,
in the domain (x, t) ∈ Ωt = {x ∈ Q ⊂ Rd , d ≥ 1, t ∈ [0, T ]}, under the conditions
u (x, 0) = ϕ (x) , x ∈ Q̄,
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 3

u (x, T ) = ψ (x) , x ∈ Q̄,

α
where D0+,t is a Caputo fractional derivative of order 0 < α ≤ 1, ϕ (x) and ψ (x)
are sufficiently smooth functions, and L is a linear positive operator with a discrete
spectrum.
In many contexts, for example, in the sub-Riemannian settings, the (fractional)
time diffusion equation for subelliptic operators arises naturally. However, such oper-
ators have a non-discrete, but continuous spectrum. Following Rothschild and Stein
[RS76] and further developments, many of such operators can be modelled by the
so-called Rockland operators (homogeneous left-invariant hypoelliptic differential op-
erators) on graded Lie groups. By using the proceeding analysis, we will employ the
group Fourier transform to reduce such problems to those with the discrete spectrum,
for which the above established results would be applicable.
Thus, let R be a positive self–adjoint Rockland operator acting on L2 (G), where
G is a graded Lie group of homogeneous dimension Q ≥ 3. In a domain Ω = {(t, x) :
(0, T ) × G} we seek a couple of functions (u(t), f ) satisfying the equation
α
(1.4) D0+,t u (x, t) + Ru(x, t) = f (x) ,
under the conditions
(1.5) u (x, 0) = ϕ (x) , x ∈ G,

(1.6) u (x, T ) = ψ (x) , x ∈ G,

α
where D0+,t is a Caputo fractional derivative, 0 < α ≤ 1.
We seek a solution u ∈ C([0, T ]; L2 (G)) of the problem (1.4)–(1.6) such that
α
D0+,t u ∈ C([0, T ]; L2 (G)), Ru ∈ C([0, T ]; L2 (G)), and f ∈ C([0, T ]; L2 (G)). We
are able to solve this problem by employing the natural global Fourier analysis on
G, allowing one to use the solution to the problem (1.3) applied to the infinitesi-
mal representations of the operator R, which is in turn known to have the discrete
spectrum.
We note that the inverse source problems for Rockland operators (1.4)–(1.6) cover
also the Heisenberg case, namely, the equation
α
D0+,t u (x, t) − ∆Hn u(x, t) = f (x) ,
with the conditions
u (x, 0) = ϕ (x) , x ∈ Hn ,
u (x, T ) = ψ (x) , x ∈ Hn ,
α
where D0+,t is a Caputo fractional derivative, 0 < α ≤ 1. Here, ∆Hn is the sub-
Laplacian on the Heisenberg group Hn . As the physical application, the relevance of
the Heisenberg group Hn for quantum mechanics has been established. In 1931 Weyl
[Wey31] recognized that the Heisenberg algebra generated by the momentum and
position operators originates from the representation of the Lie algebra associated
with the corresponding group, namely, the Heisenberg group, or the Weyl group as
the physicists call it. For the recent studies of the heat equation on the Heisenberg
groups, we refer to the publications [BC17, DM05, Jui14, TW18, RTT19].
Thus, let us briefly summarise the results of this paper:
4 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

• Existence and uniqueness for the inverse diffusion problem

ut (x, t) + Lu(x, t) = f (x),

for general positive operators L with discrete spectrum and the basis of eigen-
functions.
• Existence and uniqueness for the inverse time-fractional subdiffusion problem
α
D0+,t u (x, t) + Lu(x, t) = f (x).

• Existence and uniqueness for the inverse diffusion and subdiffusion problems
α
D0+,t u (x, t) + Ru(x, t) = f (x),

for 0 < α ≤ 1, and for general homogeneous left-invariant hypoelliptic differ-

ential operators R on graded Lie groups.
• Numerical analysis of the obtained formulae in the case of the inverse heat
problems for differential operators with involution.

2. Inverse problem for the heat equation

2.1. Statement of the problem. The Section is concerned with inverse problem for
the heat equation (1.1). We obtain existence and uniqueness results for this problem,
based on the L–Fourier method. An introduction and some basic definitions of the
L–Fourier analysis are given in [KRT17, DRT17, RT16, RT17a, RT18b].

Problem 2.1. Find the couple of functions (u(x, t), f (x)) satisfying the Cauchy prob-
lem (1.1)-(1.2), under the condition

(2.1) u(x, T ) = ψ(x), x ∈ Q̄.

A generalised solution of Problem 2.1 is the pair of functions (u(x, t), f (x)), where
u = u(x, t) ∈ C 1 ([0, T ], H) ∩ C([0, T ], H1 ), and f = f (x) ∈ H. Here Hs is the closure
of HL∞ (see, Appendix 5) under the norm

kvkHs := k(I + L)s vkH ,

for all s ∈ R+ .
For the considered Problem 2.1, the following theorem holds true.

Theorem 2.2. Let ϕ, ψ ∈ H1 . Then the generalised solution of Problem 2.1 exists,
is unique, and can be written in the form
X [(ϕ, eξ )H − (ψ, eξ )H ](e−λξ t − 1)eξ (x)
u(x, t) = ϕ(x) + ,
ξ∈I
(1 − e−λξ T )

X [(Lϕ, eξ )H − (Lψ, eξ )H ]eξ (x)

f (x) = Lϕ(x) − .
ξ∈I
(1 − e−λξ T )
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 5

2.2. Proof of the existence result. We want to find a generalised solution by

the Fourier method, we have the eigenvalues λξ and eigenfunctions system eξ (x) of
operator L on the space H. Eigenfunctions system eξ (x) is an orthonormal basis in
H, the functions u(x, t) and f (x) can be expanded as follows:
X
(2.2) u(x, t) = uξ (t)eξ (x),
ξ∈I

and
X
(2.3) f (x) = fξ eξ (x),
ξ∈I

where fξ , uξ (t) are unknown. Substituting Equations (2.2) and (2.3) into Equation
(1.1), we obtain the following equation for the function uξ (t) and the constant fξ :
0
uξ (t) + λξ uξ (t) = fξ .
Solving this equation, we obtain
fξ
uξ (t) = + Cξ e−λξ t ,
λξ
where the constants Cξ , fξ are unknown. To find these constants, we use conditions
(1.2) and (2.1). Let ϕξ , ψξ be the coefficients of the expansions of ϕ(x) and ψ(x)

ϕξ = (ϕ, eξ )H ,
and
ψξ = (ψ, eξ )H .
We first find Cξ :
fξ
uξ (0) = + Cξ = ϕ ξ ,
λξ
and
fξ
uξ (T ) = + Cξ e−λξ T = ψξ ,
λξ
then
ϕξ − ψξ
Cξ = .
1 − e−λξ T
The constant fξ is represented as
fξ = λξ ϕξ − λξ Cξ .
Substituting fξ , uξ (t) into formulas (2.2) and (2.3), we find
X
u(x, t) = ϕ(x) + Cξ (e−λξ t − 1)eξ (x).
ξ∈I

Using the self-adjointness property of the operator L, we have

(Lϕ, eξ )H = (ϕ, Leξ )H
we know that Leξ = λξ eξ using this we obtain
(Lϕ, eξ )H
(ϕ, eξ )H = ,
λξ
6 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

and for ψ(x) we can write it in a similar way. Substituting these equality into formula
of Cξ we can get that
(Lϕ, eξ )H − (Lψ, eξ )H
Cξ = .
λξ (1 − e−λξ T )
Then
X [(Lϕ, eξ )H − (Lψ, eξ )H ](e−λξ t − 1)eξ (x)
u(x, t) = ϕ(x) + −λξ T )
.
λ ξ (1 − e
ξ∈I
Similarly,
X [(Lϕ, eξ )H − (Lψ, eξ )H ]eξ (x)
f (x) = Lϕ(x) − .
ξ∈I
(1 − e−λξ T )
Now, for the convergence of the series, using keξ kH = 1 and ϕ, ψ ∈ H1 , we have the
following estimate
X |(ϕ, eξ )H − (ψ, eξ )H |2 · |e−λξ t − 1|2 · keξ k2
H
max ku(x, t)k2H ≤ Ckϕ(x)k2H + C max −λ T 2
t∈[0,T ] t∈[0,T ]
ξ∈I
|1 − e ξ |
X
≤ Ckϕ(x)k2H + C |(ϕ, eξ )H |2 + |(ψ, eξ )H |2 < ∞, C = const > 0,
ξ∈I

and
X |(Lϕ, eξ )H − (Lψ, eξ )H |2 · | − λξ · e−λξ t |2 · keξ k2
H
max kut (x, t)kH |2 ≤ max
t∈[0,T ] t∈[0,T ]
ξ∈I
|λξ |2 · |1 − e−λξ T |2
X
≤C |(Lϕ, eξ )H | + |(Lψ, eξ )H | < ∞, C = const > 0.
ξ∈I
1
From this and ϕ, ψ ∈ H , we obtain
ku(x, t)kC 1 ([0,T ],H) = max ku(x, t)kH + max kut (x, t)kH < ∞.
t∈[0,T ] t∈[0,T ]

Similarly for f (x), we obtain the estimate

X
kf k2H ≤ CkLϕk2H + C |(Lϕ, eξ )H |2 + |(Lψ, eξ )H |2 < ∞.
ξ∈I

Existence of the solution of Problem 2.1 is proved.

2.3. Proof of the uniqueness result. Suppose that there are two solutions {u1 (x, t), f1 (x)}
and {u2 (x, t), f2 (x)} of Problem 2.1. Denote
u(x, t) = u1 (x, t) − u2 (x, t)
and
f (x) = f1 (x) − f2 (x).
Then the functions u(x, t) and f (x) satisfy Equation (1.1) and homogeneous condi-
tions (1.2) and (2.1). We also have
(2.4) uξ (t) = (u(x, t), eξ (x))H , ξ ∈ I,
and
(2.5) fξ = (f (x), eξ (x))H , ξ ∈ I.
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 7

d
Applying the operator dt
to (2.4), we have
duξ (t)
= (ut (x, t), eξ (x))H = (Lu(x, t) + f (x), eξ (x))H = λξ uξ (t) + fξ .
dt
Thus we get
duξ (t)
(2.6) = λξ uξ (t) + fξ ,
dt
an ordinary first-order differential equation. The general solution of this equation is:
fξ
uξ (t) = Aξ eλξ t −
,
λξ
where Aξ and fξ are unknown constants. Using the homogeneous conditions (1.2)
and (2.1) we obtain following conditions:
uξ (0) = uξ (T ) = 0.
Using this we can find unknown constants Aξ and fξ . We first find Aξ :
fξ fξ
uξ (0) = Aξ − = 0, then Aξ = .
λξ λξ
Similarly, from
fξ
uξ (T ) = Aξ eλξ T − =0
λξ
we obtain
f ξ λξ T
(e − 1) = 0,
λξ
this implies
fξ = 0, Aξ = 0 =⇒ uξ (t) ≡ 0.
Further, by the completeness of the system eξ in H, we obtain f (x) ≡ 0, u(x, t) ≡ 0.
Uniqueness of the solution of the Problem 2.1 is proved.

3. Inverse problem for the time-fractional diffusion equation

The section deals with an inverse problem concerning the time-fractional diffusion
equation.
3.1. Preliminaries. Now, to formulate the problem, we need to define fractional
differentiation operators.
Definition 3.1 ([KST06]). The left and right Riemann–Liouville fractional integrals
α α
Ia+ and Ib− of order α ∈ R (α > 0) are given by
Zt
1
α
Ia+ [f ] (t) = (t − s)α−1 f (s)ds, t ∈ (a, b],
Γ (α)
a
and
Zb
1
α
Ib− [f ] (t) = (s − t)α−1 f (s)ds, t ∈ [a, b),
Γ (α)
t
8 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

respectively. Here Γ denotes the Euler gamma function.

α
Definition 3.2 ([KST06]). The left Riemann–Liouville fractional derivative Da+ of
order α ∈ R (0 < α < 1) is defined by
α d 1−α
Da+ [f ] (t) = I [f ] (t) , ∀t ∈ (a, b].
dt a+
α
Similarly, the right Riemann–Liouville fractional derivative Db− of order α ∈ R (0 <
α < 1) is given by
α d 1−α
Db− [f ] (t) = − I [f ] (t) , ∀t ∈ [a, b).
dt b−
Definition 3.3 ([KST06]). The left and right Caputo fractional derivatives of order
α ∈ R (0 < α < 1) are defined by
α α
Da+ [f ] (t) = Da+ [f (t) − f (a)] , t ∈ (a, b],
and
α α
Db− [f ] (t) = Db− [f (t) − f (b)] , t ∈ [a, b),
respectively.
Now, we are in a way to state our problem.
Problem 3.4. Find the couple of functions (u (x, t) , f (x)) satisfying the equation
α
(3.1) D0+,t u (x, t) + Lu(x, t) = f (x) ,
in the domain (x, t) ∈ Ωt = {x ∈ Q ⊂ Rd , d ≥ 1, t ∈ [0, T ]}, under the conditions
(3.2) u (x, 0) = ϕ (x) , x ∈ Q̄,

(3.3) u (x, T ) = ψ (x) , x ∈ Q̄,

where ϕ (x) and ψ (x) are sufficiently smooth functions, L is a linear self-adjoint
operator with a discrete spectrum.
If α = 1, then equation (3.1) coincides with the classical heat equation. The
heat equation also describes the diffusion process. So, the equation of the form
(3.1) with fractional derivatives with respect to the time variable is called the sub-
diffusion equation [Uch13]. This equation describes the slow diffusion. When α = 12
the equation was interpreted by Nigmatullin [Nig86] within a percolation (pectinate)
model. The solution (in an unbounded domain in the space variable) was investigated
by Mainardi [Mai00] and others by means of integral transformations.
Definition 3.5 ([CF18]). Let X be a Banach space. We say that u ∈ C α ([0, T ], X)
if u ∈ C([0, T ], X) and Dtα u ∈ C([0, T ], X).
Observe that, viewed as a subspace of C([0, T ], X), the space C α ([0, T ], X) is a
Banach space.
A generalised solution of Problem 3.4 is the pair of functions (u(x, t), f (x)), where
u = u(x, t) ∈ C α ([0, T ], H) ∩ C([0, T ], H1 ) and f = f (x) ∈ H.
For the considered Problem 3.4, the following theorem holds true.
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 9

Theorem 3.6. Let ϕ, ψ ∈ H1 . Then the generalised solution of the Problem 3.4,
exists, is unique, and can be written in the form
X [(ϕ, eξ )H − (ψ, eξ )H ](Eα,1 (−λξ tα ) − 1)eξ (x)
u(x, t) = ϕ(x) + ,
ξ∈I
(1 − Eα,1 (−λξ tα ))
X [(Lϕ, eξ )H − (Lψ, eξ )H ]eξ (x)
f (x) = Lϕ(x) − ,
ξ∈I
(1 − Eα,1 (−λξ tα ))
where Eα,1 (z) is the Mittag-Leffler function:
∞
X zm
Eα,1 (z) = .
m=0
Γ(αm + 1)

For different properties of the Mittag-Leffler function Eα,1 (z) see e.g. [KST06].
3.2. Proof of Theorem 3.6. We give the full proof for Problem 2.1.
3.2.1. Existence of solution. We want to find a generalised solution by Fourier method,
we have the eigenvalues λξ and eigenfunctions system eξ (x) of operator L on the space
H. Eigenfunctions system eξ (x) is an orthonormal basis in H, so that the functions
u(x, t) and f (x) can be expanded as follows:
X
(3.4) u (x, t) = uξ (t) eξ (x),
ξ∈I

X
(3.5) f (x) = fξ eξ (x),
ξ∈I

where uξ (t) , fξ are unknown. Substituting (3.4) and (3.5) into (3.1), we obtain the
following equations for the unknown functions uξ (t) and the constants fξ :
α
D0+ uξ (t) + λξ uξ (t) = fξ , ξ ∈ I.
By solving this equation (see [KST06]), we obtain
fξ
uξ (t) = + Cξ Eα,1 (−λξ tα ) ,
λξ
where the constants Cξ and fξ are unknown and Eα,1 (z) is the Mittag-Leffler function
[KST06]:
∞
X zm
Eα,1 (z) = .
m=0
Γ(αm + 1)
To find these constants, we use conditions (3.2). Let ϕξ , ψξ be the coefficients of the
expansions of ϕ (x) and ψ (x)
ϕξ = (ϕ, eξ )H , ψξ = (ψ, eξ )H ,
where (·, ·)H is the inner product of the Hilbert space H.
Then for Cξ we have
fξ
uξ (0) = + Cξ = ϕ ξ ,
λξ
10 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

fξ
uξ (T ) = + Cξ Eα,1 (−λξ T α ) = ψξ ,
λξ
ϕξ − Cξ + Cξ Eα,1 (−λξ T α ) = ψξ .
Thus, we get
ϕξ − ψξ
Cξ = .
1 − Eα,1 (−λξ T α )
The unknowns fξ can be represented as
fξ = λξ (ϕξ − Cξ ).
Substituting uξ (t) , fξ into (3.4) and (3.5), we find
X
u (x, t) = ϕ (x) + Cξ (Eα,1 (−λξ tα ) − 1) eξ (x).
ξ∈I

Using the self-adjointness property of the operator L, we have

(Lϕ, eξ )H = (ϕ, Leξ )H
we know that Leξ = λξ eξ using this we obtain
(Lϕ, eξ )H
(ϕ, eξ )H = ,
λξ
and for ψ(x) we can write it in a similar way. Substituting these equalities into
formula of Cξ we can get that
(Lϕ, eξ )H − (Lψ, eξ )H
Cξ = .
λξ (1 − Eα,1 (−λξ T α ))
Then
X [(Lϕ, eξ )H − (Lψ, eξ )H ](Eα,1 (−λξ tα ) − 1)eξ (x)
u(x, t) = ϕ(x) + .
ξ∈I
λξ (1 − Eα,1 (−λξ T α ))

Similarly,
X [(Lϕ, eξ )H − (Lψ, eξ )H ]eξ (x)
f (x) = Lϕ(x) − .
ξ∈I
(1 − Eα,1 (−λξ T α ))
The following Mittag-Leffler function’s estimate is known [Sim14]:
1 1
≤ Eα,1 (−z) ≤ , z > 0.
1 + Γ(1 − α)z 1 + Γ(1 − α)−1 z
From this inequality it follows that
0 < Eα,1 (−z) < 1, z > 0.
Now, for the convergence of the series, using keξ kH = 1 and ϕ, ψ ∈ H1 , we have the
following estimate
X
max ku(x, t)k2H ≤ Ckϕ(x)k2H + C |(ϕ, eξ )H |2 + |(ψ, eξ )H |2 < ∞.
t∈[0,T ]
ξ∈I
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 11

We also have
X |(Lϕ, eξ )H − (Lψ, eξ )H |2 · | − λξ · Eα,1 (−λξ tα )|2 · keξ k2
α H
max kD0+,t u(x, t)k2H ≤ max
t∈[0,T ] t∈[0,T ]
ξ∈I
λ2ξ · |1 − Eα,1 (−λξ T α )|2
X
≤C |(Lϕ, eξ )H − (Lψ, eξ )H |2
ξ∈I
X
≤C |(Lϕ, eξ )H |2 + |(Lψ, eξ )H |2 < ∞.
ξ∈I

From this and ϕ, ψ ∈ H1 , we obtain

ku(x, t)k2C α ([0,T ],H) = max ku(x, t)k2H + max kD0+,t α
u(x, t)k2H
t∈[0,T ] t∈[0,T ]
X
≤ Ckϕ(x)k2H + C |(ϕ, eξ )H |2 + |(ψ, eξ )H |2
ξ∈I
X
+C |(Lϕ, eξ )H |2 + |(Lψ, eξ )H |2 < ∞.
ξ∈I

Similarly for f (x), we obtain the estimate

X
kf k2H ≤ CkLϕk2H + C |(Lϕ, eξ )H |2 + |(Lψ, eξ )H |2 < ∞.
ξ∈I

Existence of the solution of the Problem 3.4 is proved.

3.2.2. Uniqueness of solution. Hence the obtained solution satisfies the equation (3.1)
point-wise; by construction, it satisfies the conditions (3.2)-(3.3).
Suppose that there are two solutions {u1 (x, t) , f1 (x)} and {u2 (x, t) , f2 (x)} of
Problem 3.4. Denote
u (x, t) = u1 (x, t) − u2 (x, t)
and
f (x) = f1 (x) − f2 (x) .
Then the functions u (x, t) and f (x) satisfy (3.1) and homogeneous conditions (3.2)
and (3.3).
Let
(3.6) uξ (t) = (u(x, t), eξ (x))H , ξ ∈ H,
and
(3.7) fξ = (f (x), eξ (x))H , ξ ∈ I.
α
Applying the operator D0+ to the equation (3.6), we have
α α


D0+ uξ (t) = D0+,t u(x, t), eξ (x) H = − hLu(x, t), eξ (x)iH + fξ .
By self-adjointness and taking into account the homogeneous conditions (3.2) and
(3.3), we obtain
α
D0+ uξ (t) + λξ uξ (t) = fξ , uξ (0) = 0 , uξ (T ) = 0.
Consequently, fξ ≡ 0, uξ (t) ≡ 0.
12 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

Further, by the completeness of the system {eξ (x)}ξ∈I in H we obtain

f (t) ≡ 0, u (x, t) ≡ 0, 0 ≤ t ≤ T, x ∈ Q̄.
Hence, uniqueness of the solution of Problem 3.4 is proved.

4. Inverse time-fractional diffusion problem for the hypoelliptic

operators
In this section we show how the obtained results can be applied for the analysis
of the corresponding inverse problems for a large class of hypoelliptic diffusions and
subdiffusions.
4.1. Graded Lie groups. We start by recalling following Folland and Stein [FS82]
or [FR16, Section 3.1] and give some definitions and notations. A Lie algebra g is
graded if it is endowed with a vector space decomposition
M∞
g= Vj such that [Vi , Vj ] ⊂ Vi+j ,
j=1

where all but finitely many of Vj ’s are zero. Consequently, we call that a connected
simply connected Lie group G is graded if its Lie algebra g is graded. When the first
stratum V1 generates g as an algebra, we get a stratified case of G.
Graded Lie groups are homogeneous Lie groups with dilations. Define the operator
A by setting AX = νj X for X ∈ Vj . Then the dilations on g are defined by
Dr := Exp(A ln r), r > 0.
The homogeneous dimension Q of the graded Lie group G is defined by
Q := ν1 + . . . + νn = Tr A.
In what follows, we assume that G is a graded Lie group. Rockland operators are
firstly defined in [Roc78] through the representations. By following [FR16, Definition
4.1.1], we call that R is a Rockland operator on the graded Lie group G if R is a
left-invariant differential operator which is homogeneous of a positive order ν ∈ N
and satisfies the Rockland condition: for all representations π ∈ G, b excluding the
∞
trivial one, π(R) is injective on Hπ , i.e., from
π(R)v = 0
it follows that v = 0 for arbitrary v ∈ Hπ∞ .
We denote by G b the unitary dual of G, H∞ is the space of smooth vectors of the
π
representation π ∈ G,b and π(R) is the infinitesimal representation of R, see [FR16,
Definition 1.7.2]. For more information on graded Lie groups and Rockland operators
the readers are referred to [FR16, Chapter 4].
Let π be a representation of the graded Lie group G on the separable Hilbert space
Hπ . We say that v ∈ Hπ is smooth if the function
G 3 x 7→ π(x)v ∈ Hπ
is of class C ∞ . We denote by Hπ∞ the space of all smooth vectors of a representation
π.
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 13

In the paper [HJL85] Hulanicki, Jenkins and Ludwig showed that the spectrum of
π(R) is purely discrete and positive. Hence we can choose an orthonormal basis for
Hπ such that the infinite matrix related to the (self-adjoint) operator π(R) has the
following form
 2 
π1 0 . . . . . .
 0 π22 0 . . .
(4.1) π(R) =  .. .. ,
 
. 0 . 
.. .. ..
. . .

where π ∈ G\{1}
b and πj ∈ R>0 .
1
For f ∈ L (G) and π ∈ G,
b let us define the group Fourier transform of f at π by
Z
FG f (π) ≡ f (π) ≡ π(f ) :=
b f (x)π(x)∗ dx,
G

with the integration against the biinvariant Haar measure on the graded Lie group
G, which implies that a linear mapping fb(π) from Hπ to itself can be represented by
an infinite matrix once we choose a basis for Hπ . Consequently, we obtain
FG (Rf )(π) = π(R)fb(π).

From now on, when we write fb(π)m,k , we will be using the same basis in Hπ as the
one giving (4.1).
In [CG90] by Kirillov’s orbit method, the authors showed that the Plancherel mea-
sure µ on Gb can be constructed explicitly. This means that we can have the Fourier
inversion formula. Furthermore, π(f ) = fb(π) is the Hilbert-Schmidt operator, i.e.
kπ(f )k2HS = Tr(π(f )π(f )∗ ) < ∞,

and G b 3 π 7→ kπ(f )k2 is an integrable function with respect to µ. Moreover, the

HS
Plancherel formula holds (see e.g. [CG90] or [FR16]):
Z Z
2
(4.2) |f (x)| dx = kπ(f )k2HS dµ(π).
G G
b

In [HN79] Helffer and Nourrigat showed that a left-invariant differential operator

R of homogeneous positive degree ν ∈ N satisfies the Rockland condition if and only
if it is hypoelliptic. Such operators we call Rockland operators.
s
The Sobolev spaces HR (G), s ∈ R, associated to positive Rockland operators R
have been analysed in [FR17] (see also [FR16]). One of the definitions of Sobolev
spaces is
H s (G) := HRs
(G) := f ∈ D0 (G) : (I + R)s/ν f ∈ L2 (G) ,


with the norm kf kHRs (G) := k(I + R)s/ν f kL2 (G) , for a positive Rockland operator of
homogeneous degree ν. It is known that these spaces do not depend on a particular
choice of a Rockland operator used in the above definition. We refer to [FR16] for
details of the Fourier analysis on graded Lie groups.
Now we state the main problem of this section.
14 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

Problem 4.1. Let G be a graded Lie group of homogeneous dimension Q ≥ 3

and let R be a positive self–adjoint Rockland operator acting on L2 (G). In a domain
Ω = {(t, x) : (0, T )×G} we seek a couple of functions (u(t), f ) satisfying the equation
α
(4.3) D0+,t u (x, t) + Ru(x, t) = f (x) ,
under the conditions
(4.4) u (x, 0) = ϕ (x) , x ∈ G,

(4.5) u (x, T ) = ψ (x) , x ∈ G,

α
where is a Caputo fractional derivative, 0 < α ≤ 1.
D0+,t
We seek a solution u ∈ C([0, T ]; L2 (G)) of the problem (4.3)–(4.5) such that
α
D0+,t u ∈ C([0, T ]; L2 (G)), Ru ∈ C([0, T ]; L2 (G)), and f ∈ C([0, T ]; L2 (G)).
Theorem 4.2. Assume that R is a positive Rockland operator of homogeneous order
ν. Let ϕ, ψ ∈ H ν (G). Then there exists a unique solution (u, f ) such that f ∈ L2 (G),
u ∈ C([0, T ]; H ν (G)), and D0+,t
α
u ∈ C([0, T ]; L2 (G)) of Problem 4.1, and can be
written in the form
Z
(4.6) u(t, x) = Tr[K(t,
b π)π(x)]dµ(π),
G
b

and Z
f (x) = Tr[Fb(π)π(x)]dµ(π),
G
b
where
b l,k ](Eα,1 (−π 2 tα ) − 1)
b l,k − ψ(π)
[ϕ(π) l
K(t,
b π)l,k = ϕ(π)
b l,k + ,
(1 − Eα,1 (−πl2 T α ))
and
2 πl2 [ϕ(π)
b l,k − ψ(π)
b l,k ]
b b l,k −
F (π)l,k = πl ϕ(π) ,
1 − Eα,1 (−πl2 T α )
for all π ∈ G
b and l, k ∈ N, where Eα,1 (z) is the Mittag-Leffler function.

4.2. Proof of Theorem 4.2.

4.2.1. Proof of the existence result. We give a full proof of Problem 4.1. Let us take
the group Fourier transform of (4.3) with respect to x ∈ G for all π ∈ G,b that is,
α
(4.7) D0+,t u
b(t, π) + π(R)b
u(t, π) = fb(π).
Taking into account (4.1), we rewrite the matrix equation (4.7) componentwise as an
infinite system of equations of the form
(4.8) Dα u b(t, π)l,k + π 2 u
0+,t b(t, π)l,k = fb(π)l,k ,
l

for all π ∈ G,
b and any l, k ∈ N. Now let us decouple the system given by the matrix
equation (4.7). For this, we fix an arbitrary representation π, and a general entry
(l, k) and we treat each equation given by (4.8) individually.
According to [LG99], the solutions of the equations (4.8) satisfying initial conditions
(4.9) u
b(0, π)l,k = ϕ(π)
b l,k , u
b(T, π)l,k = ψ(π)
b l,k ,
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 15

can be represented in the form

b l,k ](Eα,1 (−π 2 tα ) − 1)
b l,k − ψ(π)
[ϕ(π) l
(4.10) u
b(t, π)l,k = ϕ(π)
b l,k + 2 α
,
(1 − Eα,1 (−πl T ))
π 2 [ϕ(π)
b l,k − ψ(π)
b l,k ]
b l,k − l
fb(π)l,k = πl2 ϕ(π) ,
1 − Eα,1 (−πl2 T α )
for all π ∈ G
b and any l, k ∈ N, where Eα,1 (z) is the Mittag-Leffler function [KST06]:
∞
X zm
Eα,1 (z) = .
m=0
Γ(αm + 1)
Then there exists a solution of Problem 4.1, and it can be written as
Z
(4.11) u(t, x) = Tr[K(t,
b π)π(x)]dµ(π),
Gb
Z
f (x) = Tr[Fb(π)π(x)]dµ(π),
G
b
where
b l,k ](Eα,1 (−π 2 tα ) − 1)
b l,k − ψ(π)
[ϕ(π) l
K(t,
b π)l,k = ϕ(π)
b l,k + ,
(1 − Eα,1 (−πl2 T α ))
π 2 [ϕ(π)
b l,k − ψ(π)
b l,k ]
b l,k − l
Fb(π)l,k = πl2 ϕ(π) ,
1 − Eα,1 (−πl2 T α )
for all π ∈ G
b and l, k ∈ N.
We note that the above expressions ares well-defined in view of ϕ, ψ ∈ Hν (G).
Finally, based on (4.10), we rewrite our formal solution as (4.11).
4.2.2. Convergence of the formal solution. Here, we prove convergence of the obtained
α
infinite series corresponding to functions u(x, t), D0+,t u(x, t), and Ru(x, t).
Thus, since for any Hilbert-Schmidt operator A one has
X
kAk2HS = |(Aφl , φk )|2
l,k

for any orthonormal basis {φ1 , φ2 , . . .}, then we can consider the infinite sum over l, k
of the inequalities provided by (4.10), we have
(4.12) u(t, π)k2HS ≤ C(kϕ(π)k
kb b 2 2
HS + kψ(π)kHS ),
b

α
(4.13) kD0+,t b(t, π)k2HS ≤ C(kπ(R)ϕ(π)k
u b 2 2
HS + kπ(R)ψ(π)kHS ),
b
and
(4.14) c π)k2HS ≤ C(kπ(R)ϕ(π)k
kRu(t, b 2 2
HS + kπ(R)ψ(π)kHS ).
b

Thus, integrating both sides of (4.14) against the Plancherel measure µ on G,

b then
using the Plancherel identity (4.2) we obtain
kukC([0,T ];L2 (G)) ≤ C(kϕkL2 (G) + kψkL2 (G) ),
α
kD0+,t ukC([0,T ];L2 (G)) ≤ C(kRϕkL2 (G) + kRψkL2 (G) ),
16 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

and
kRukC([0,T ];L2 (G)) ≤ C(kRϕkL2 (G) + kRψkL2 (G) ).
Similarly for f , we obtain the estimate
kf kL2 (G) ≤ C(kϕkHν (G) + kRψkHν (G) ).
The uniqueness result can be proved in analogy to the previous arguments.

5. Appendix: Non-harmonic analysis of operators with discrete

spectrum
In this section we recall the necessary elements of the global Fourier analysis that
has been developed in [RT16], and its applications to the spectral properties of op-
erators in [DRT17]. The space
HL∞ := Dom (L∞ )
is called the space of test functions for L. Here we define
∞
\
Dom (L∞ ) := Dom Lk ,


k=1
k


where Dom L is the domain of the operator Lk , in turn defined as
Dom Lk := f ∈ H : Lj f ∈ Dom (L) , j = 0, 1, 2, ..., k − 1 .



The Fréchet topology of HL∞ is given by the family of semi-norms

(5.1) kϕkHL∞ := max Lj ϕ H
, k ∈ N0 , ϕ ∈ HL∞ .
j≥k
∗
Analogously to the operator L (H-conjugate to L), we introduce the space
HL∞∗ := Dom ((L∗ )∞ )
of test functions for L∗ , and we define
∞
\  
∗ ∞
Dom ((L ) ) := Dom (L∗ )k ,
k=1
 
where Dom (L∗ )k is the domain of the operator (L∗ )k , in turn defined as
  n o
Dom (L∗ )k := f ∈ H : (L∗ )j f ∈ Dom (L∗ ) , j = 0, 1, 2, ..., k − 1 .
The Fréchet topology of HL∞∗ is given by the family of semi-norms
(5.2) kϕkHL∞∗ := max (L∗ )j ϕ , k ∈ N0 , ϕ ∈ CL∞∗ .
j≥k H

Now the space

HL−∞
∗ := L (HL∞ , C)
of linear continuous functionals on HL∞ is called the space of L∗ -distributions. We
can understand the continuity here in terms of the topology (5.1). For w ∈ HL−∞∗ and
w ∈ HL∞ , we shall also write
w(ϕ) = hw, ϕi.
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 17

For any ψ ∈ HL∞∗ , the functional

HL∞ 3 ϕ 7→ (ψ, ϕ)
is an L∗ -distribution, which gives an embedding ψ ∈ HL∞∗ ,→ HL−∞ ∗ .

Since the system of eigenfunctions {eξ : ξ ∈ I} of the operator L is a Riesz basis in

H then its biorthogonal system {e∗ξ : ξ ∈ I} is also a Riesz basis in H (see e.g. Bari
[Bar51], as well as Gelfand [Gel63]). Note that the function e∗ξ is an eigenfunction of
the operator L∗ corresponding to the eigenvalue λ̄ξ for each ξ ∈ I. They satisfy the
orthogonality relations
(eξ , e∗η ) = δξ,η ,
where δξ,η is the Kronecker delta.
If L is self-adjoint, we clearly have e∗ξ = eξ .

6. Examples
Now as an illustration we give several examples of the settings where our inverse
problems are applicable. Of course, there are many other examples, here we collect
the ones for which different types of partial differential equations have particular
importance.
• Sturm-Liouville problem.
First, we describe the setting of the Sturm-Liouville operator. Let l be an
ordinary second order differential operator in L2 (a, b) generated by the differ-
ential expression
(6.1) l(u) = −u00 (x), a < x < b,
and one of the boundary conditions
(6.2) a1 u0 (b) + b1 u(b) = 0, a2 u0 (a) + b2 u(a) = 0,
or
(6.3) u(a) = ±u(b), u0 (a) = ±u0 (b),
where a21 + a22 > 0, b21 + b22 > 0 and αj , βj , j = 1, 2 some real numbers.
It is known [Nai68] that the Sturm-Liouville problem for equation (6.1) with
boundary conditions (6.2) or with boundary conditions (6.3) is self-adjoint in
L2 (a, b). It is known that the self-adjoint problem has real eigenvalues and
their eigenfunctions form a complete orthonormal basis in L2 (a, b).
• Differential operator with involution.
As a next example, we consider the differential operator with involution in
L2 (0, π) generated by the expression
(6.4) l(u) = u00 (x) − εu00 (π − x), 0 < x < π,
and homogeneous Dirichlet conditions
(6.5) u(0) = 0, u(π) = 0,
where |ε| < 1 some real number.
18 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

The nonlocal functional-differential operator (6.4)-(6.5) is self-adjoint [KST17,

TT17]. For |ε| < 1, the operator (6.4)-(6.5) has the following eigenvalues
λ2k = 4(1 + ε)k 2 , k ∈ N and λ2k+1 = (1 − ε)(2k + 1)2 , k ∈ N ∪ {0}
and corresponding eigenfunctions
r
2
u2k (x) = sin 2kx, k ∈ N,
π
r
2
u2k+1 (x) = sin (2k + 1)x, k ∈ N ∪ {0}.
π
• Fractional Sturm-Liouville operator.
We consider the operator generated by the integro-differential expression
α α
(6.6) `(u) = Da+ Db− u, a < x < b,
and the conditions
1−α 1−α
(6.7) Ib− u(a) = 0, Ib− u(b) = 0,
α α
where Da+ is the left Caputo derivative of order α ∈ (1/2, 1] , Db− is the
α
right Riemann-Liouville derivative of order α ∈ (1/2, 1] and Ib− is the right
Riemann-Liouville integral of order α ∈ (1/2, 1] (see. Subsection 3.1 and
[KST06]). The fractional Sturm-Liouville operator (6.6)-(6.7) is self-adjoint
and positive in L2 (a, b) (see [TT16]). The spectrum of fractional Sturm-
Liouville operator (6.6)-(6.7) is discrete, positive and real valued, and the
system of eigenfunctions is a complete orthogonal basis in L2 (a, b). For more
properties of the operator generated by the problem (6.6)-(6.7) we refer to
[TT18, TT18d, TT18a].
• Harmonic oscillator.
For any dimension d ≥ 1, let us consider the harmonic oscillator,
L := −∆ + |x|2 , x ∈ Rd .
L is an essentially self-adjoint operator on C0∞ (Rd ). It has a discrete spectrum,
consisting of the eigenvalues
d
X
λk = (2kj + 1), k = (k1 , · · · , kd ) ∈ Nd ,
j=1

and with the corresponding eigenfunctions

d
Y |x|2
ϕk (x) = Pkj (xj )e− 2 ,
j=1

which are an orthogonal basis in L2 (Rd ). We denote by Pl (·) the l–th order
Hermite polynomial, and
 l
|ξ|2 d |ξ|2
Pl (ξ) = al e 2 x − e− 2 ,
dξ
where ξ ∈ R, and
al = 2−l/2 (l!)−1/2 π −1/4 .
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 19

For more information, see for example [NR10].

• Anharmonic oscillator.
Another class of examples – anharmonic oscillators (see for instance [HR82]),
operators on L2 (R) of the form
d2k
L := − + x2l + p(x), x ∈ R,
dx2k
for integers k, l ≥ 1 and with p(x) being a polynomial of degree ≤ 2l − 1 with
real coefficients. Other examples are a large class of harmonic and anharmonic
oscillators in all dimension, see e.g. [CDR18].
• Landau Hamiltonian in 2D.
The next example is one of the simplest and most interesting models of the
Quantum Mechanics, that is, the Landau Hamiltonian.
The Landau Hamiltonian in 2D is given by
 2  2 !
1 ∂ ∂
(6.8) L := i − By + i + Bx ,
2 ∂x ∂y
acting on the Hilbert space L2 (R2 ), where B > 0 is some constant. The
spectrum of L consists of infinite number of eigenvalues (see [Foc28, Lan30])
with infinite multiplicity of the form
(6.9) λn = (2n + 1)B, n = 0, 1, 2, . . . ,
and the corresponding system of eigenfunctions (see [ABGM15, HH13]) is
 s
 n! k+1
 B(x2 + y 2 ) 
1
(x + iy)k Ln(k) (B(x2 + y 2 )), 0 ≤ k,

 e
 k,n
 (x, y) = B 2 exp −
(n − k)! 2

s
 j! n−1
 B(x2 + y 2 ) 
 e2j,n (x, y) = (n)
(x − iy)n Lj (B(x2 + y 2 )), 0 ≤ j,

B 2 exp −

(j + n)! 2


(α)
where Ln are the Laguerre polynomials given by
n k
n−k t
X
(α)
Ln (t) = (−1)k Cn+α , α > −1.
k=0
k!
Note that in [RT17b, RT18, RT19a] the wave equation for the Landau Hamil-
tonian with a singular magnetic field was studied.
The reader is referred to [MRT19a, MRT19b, RT17c, RT18c, RT19b] for
more examples of operators L and its applications.
• The restricted fractional Laplacian.
On the other hand, one can define a fractional Laplacian operator by using the
integral representation in terms of hypersingular kernels already mentioned
g(x) − g(ξ)
Z
s
(−∆Rn ) g(x) = Cd,s P.V. dξ,
|x − ξ|n+2s
Rn

where s ∈ (0, 1).

In this case we materialize the zero Dirichlet condition by restricting the op-
erator to act only on functions that are zero outside bounded domain Ω ⊂ Rn .
20 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

Caffarelli and Siro [CS17] called the operator defined in such a way the re-
stricted fractional Laplacian (−∆Ω )s . As defined, (−∆Ω )s is a self-adjoint op-
erator on L2 (Ω), with a discrete spectrum λs,k > 0, k ∈ N. The corresponding
set of eigenfunctions {Vs,k (x)}k∈N , normalized in L2 (Ω).
• Hesienberg Harmonic and Anharmonic Oscillators.
In [RR18] the authors studied classes of operators yielding harmonic and an-
harmonic oscillators on the Heisenberg group Hn . These operators share one
main feature: Every gradable Lie algebra equipped with a specific gradation
admits a non-empty set of positive Rockland forms, consequently, one can
associate to each positive Rockland form P ∈ u(hn,2 ) a family of anharmonic
oscillators {AkHn }k∈R\{0} ; the family of harmonic oscillators {OkHn }k∈R\{0} was
included as the special case in which the Dynin-Folland Lie algebra hn,2 was
equipped with the canonical homogeneous structure related to the natural
stratification and P = −(X12 + . . . X2n 2 2
+ Y2n+1 ). Such operators have a dis-
crete spectrum and also fall within the scope of this paper.

7. Numerical illustrations
In this section we provide some numerical simulations. We deal with the case of
operator (6.4)-(6.5). Namely, in L2 (0, π) consider the inverse source problem for the
heat equation
(7.1) ut (x, t) − u00 (x, t) + εu00 (π − x, t) = f (x), 0 < x < π,
with boundary conditions
(7.2) u(0, t) = 0, u(π, t) = 0, t ∈ [0, T ],
and with the Cauchy problem
(7.3) u(x, 0) = ϕ(x), x ∈ [0, π],
with some extra information
(7.4) u(x, T ) = ψ(x), x ∈ [0, π],
where −1 < ε < 1 is some real number.
By Theorem 2.2, from the Cauchy data ϕ ∈ H2 := Ẇ24 (0, π), that is, Ẇ24 (0, π) :=
{ϕ, ϕ(4) ∈ L2 (0, π) with ϕ(0) = ϕ(π) = ϕ00 (0) = ϕ00 (π) = 0}, and from the obser-
vation function ψ ∈ Ẇ24 (0, π) we restore a unique solution u ∈ C 2 ([0, T ], L2 (0, π)) ∩
C([0, T ], Ẇ24 (0, π)) and a source term f ∈ L2 (0, π) by formulas
X π l2 (ψ(s) − ϕ(s))vk (s)ds
R
(7.5) u(x, t) = ϕ(x) + 0
2 −λ T
(1 − e−λk t )vk (x),
k∈N
λk (1 − e k )
and
Rπ
X
0
l2 (ψ(s) − ϕ(s))vk (s)ds
(7.6) f (x) = l(ϕ(x)) + vk (x).
k∈N
λk (1 − e−λk T )
Here, we denote
l(ϕ(x)) := −ϕ00 (x) + εϕ00 (π − x).
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 21

From Example (6.4)-(6.5), we have a representation for the eigenvalues

λk = (1 + (−1)k ε)k 2 , k ∈ N,
and for the eigenfunctions
r
2
vk (x) = sin kx, k ∈ N.
π

Figure 1. In this picture the temperature function at the initial point

0 is drawn. The function is given by the formula (7.7).

In what follows, we consider and compare the cases ε = 0 and ε = 0.9. For
numerical simulations, we choose
ϕ(x) = x3 (π − x)3 , ψ = 0,
for all x ∈ [0, π]. Our problem is to cool (make u(x, T ) = 0 at some time T ) a rod
with the initial temperature
(7.7) u(x, 0) = x3 (π − x)3 , x ∈ [0, π].
Now, a question stands as follows: how should we choose a cooling function (a source
term) f to succeed the goal?
For our calculations we use the following intermediate formulas:
l Rπ 2
X l (ψ(s) − ϕ(s))vk (s)ds
(7.8) u(x, t) = ϕ(x) + 0
2 −λk T )
(1 − e−λk t )vk (x),
k=1
λk (1 − e
and
l Rπ
X
0
l2 (ψ(s) − ϕ(s))vk (s)ds
(7.9) f (x) = l(ϕ(x)) + vk (x).
k=1
λk (1 − e−λk T )
Here l ∈ N means a number of terms taken into account in the series of the formulas
(7.5)–(7.6). Here, we analyse the convergence of the series in (7.8)–(7.9) for different
l.
In Figures 2, 3, 4 we compare the pair of solutions (u, f ) of the inverse problem
(7.1)–(7.4) in the cases ε = 0 and ε = 0.9.
22 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

Figure 2. In the plots we illustrate the graphics of the solution u of

the inverse problem (7.1)–(7.4) for parameters ε = 0 and ε = 0.9 at t =
0.5, 2.5, 4.5. Here the graphics of f are given for different l = 7, 10, 20,
and colored in blue, red, yellow, respectively.

Figure 3. In these plots we compare the graphics of the source term

f of the inverse problem (7.1)–(7.4) for parameters ε = 0 and ε = 0.9.

7.1. Conclusion. By analysing Figures 2, 3, 4, we observe that in the case of the

problem with an involution (when ε 6= 0) we need more energy (see, Figure 3) to cool
the rod with the initial temperature shown in Figure 1. In Figure 3 we can see the
“cooling speed” of the rod. As the result, the absence of the involution guarantees
relatively rapid cooling.

References
[ABGM15] L. D. Abreu, P. Balazs, M. de Gosson and Z. Mouayn. Discrete coherent states for higher
Landau levels. Ann. Physics, 363: 337–353, 2015.
[Bar51] N. K. Bari. Biorthogonal systems and bases in Hilbert space. Moskov. Gos. Univ. Uchenye
Zapiski Matematika, 148(4):69–107, 1951.
[BC17] K. Beauchard, P. Cannarsa. Heat equation on the Heisenberg group: Observability and
applications. J. Diff. Eq. 262(8): 4475–4521, 2017.
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 23

Figure 4. In the plots the graphics of the solution u of the inverse

problem (7.1)–(7.4) are shown at x = π/2 in the cases ε = 0 and
ε = 0.9. Here the graphics of f are given for different l = 7, 10, 20, and
colored in blue, red, yellow, respectively.

[CDR18] M. Chatzakou, J. Delgado and M. Ruzhansky. On a class of anharmonic oscillators.

https://arxiv.org/abs/1811.12566
[CS17] L. A. Caffarelli, Y. Sire. On Some Pointwise Inequalities Involving Nonlocal Operators. In:
Chanillo S., Franchi B., Lu G., Perez C., Sawyer E. (eds) Harmonic Analysis, Partial Differential
Equations and Applications. Applied and Numerical Harmonic Analysis. Birkhäuser, Cham.
2017.
[CD98] J. R. Cannon, P. Du Chateau. Structural identification of an unknown source term in a heat
equation. Inverse Problems. 14: 535–551, 1998.
[CF18] P. M. de Carvalho-Neto, R. Fehlberg Junior. Conditions for the Absence of Blowing Up
Solutions to Fractional Differential Equations. Acta Applicandae Mathematicae. 154(1): 15–29,
2018.
[CNYY09] J. Cheng, J. Nakagawa, M. Yamamoto, T. Yamazaki. Uniqueness in an inverse problem
for a one-dimensional fractional diffusion equation. Inverse Problems. 25: 115002, 2009.
[CG90] L. J. Corwin and F. P. Greenleaf. Representations of nilpotent Lie groups and their applica-
tions. Cambridge Studies in Advanced Mathematics, Cambridge University Press, Cambridge,
18. Basic theory and examples, 1990.
[DRT17] J. Delgado, M. Ruzhansky and N. Tokmagambetov. Schatten classes, nuclearity and non-
harmonic analysis on compact manifolds with boundary. J. Math. Pures Appl., 107:758–783,
2017.
[DM05] B. Driver, T. Melcher. Hypoelliptic heat kernel inequalities on the Heisenberg group. J.
Funct. Anal. 221(5): 340–365, 2005.
[FR16] V. Fischer and M. Ruzhansky. Quantization on nilpotent Lie groups, volume 314 of Progress
in Mathematics. Birkhäuser/Springer, [Open access book], 2016.
[FR17] V. Fischer and M. Ruzhansky. Sobolev spaces on graded groups. Ann. Inst. Fourier,
67(4):1671–1723, 2017.
[Foc28] V. Fock, Bemerkung zur Quantelung des harmonischen Oszillators im Magnetfeld. Z. Phys.
A, 47(5–6): 446–448, 1928.
[FS82] G. B. Folland and E. M. Stein. Hardy spaces on homogeneous groups, volume 28 of Math-
ematical Notes. Princeton University Press, Princeton, N.J.; University of Tokyo Press, Tokyo,
1982.
[FIK14] K.M. Furati, O.S. Iyiola, M. Kirane. An inverse problem for a generalized fractional diffu-
sion. Applied Mathematics and Computation. 249: 24–31, 2014.
24 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

[Gel63] I. M. Gelfand. Some questions of analysis and differential equations. Am. Math. Soc. Transl.,
26:201-219, 1963.
[HH13] A. Haimi and H. Hedenmalm. The polyanalytic Ginibre ensembles. J. Stat. Phys., 153(1):10–
47, 2013.
[HN79] B. Helffer and J. Nourrigat. Caracterisation des opérateurs hypoelliptiques homogènes in-
variants à gauche sur un groupe de Lie nilpotent gradué. Comm. Partial Differential Equations,
4(8):899–958, 1979.
[HR82] B. Helffer and D. Robert. Asymptotique des niveaux d’énergie pour des hamiltoniens à un
degr é de liberté. Duke Math. J., 49(4):853–868, 1982.
[HJL85] A. Hulanicki, J. W. Jenkins, and J. Ludwig. Minimum eigenvalues for positive, Rockland
operators. Proc. Amer. Math. Soc., 94:718–720, 1985.
[IC16] M. I. Ismailov, M. Cicek. Inverse source problem for a time-fractional diffusion equation with
nonlocal boundary conditions. Applied Mathematical Modelling. 40(7): 4891–4899, 2016.
[JR15] B. Jin, W. Rundell. A tutorial on inverse problems for anomalous diffusion processes. Inverse
Problems. 31(3): 035003, 2015.
[Jui14] N. Juillet. Diffusion by optimal transport in Heisenberg groups. Calc. Var. Partial Differ.
Equ. 50(3-4): 693–721, 2014.
[KS10] I. A. Kaliev, M.M. Sabitova. Problems of determining the temperature and density of heat
sources from the initial and final temperatures. Journal of Applied and Industrial Mathematics.
4(3): 332–339, 2010.
[KRT17] B. Kanguzhin, M. Ruzhansky, and N. Tokmagambetov. On convolutions in Hilbert spaces.
Funct. Anal. Appl., 51(3):221–224, 2017.
[KST06] A. A. Kilbas, H. M. Srivastava, J.J. Trujillo. Theory and Applications of Fractional Dif-
ferential Equations, Elsevier, North-Holland, Mathematics studies, 2006.
[KM11] M. Kirane, A.S. Malik. Determination of an unknown source term and the temperature dis-
tribution for the linear heat equation involving fractional derivative in time. Applied Mathematics
and Computation. 218(1):163–170, 2011.
[KST17] M. Kirane, B. Samet, B. T. Torebek. Determination of an unknown source term tempera-
ture distribution for the sub-diffusion equation at the initial and final data. Electronic Journal
of Differential Equations. 2017: 1–13, 2017.
[Lan30] L. Landau. Diamagnetismus der Metalle. Z. Phys. A, 64(9–10): 629–637, 1930.
[LG99] Y. Luchko, R. Gorenflo. An operational method for solving fractional differential equations
with the Caputo derivatives. Acta Math. Vietnam., 24:207–233, 1999.
[Mai00] F. Mainardi. Waves and Stability in Continuous Media. Singapore: World Scientific, 2000.
[MRT19a] J. C. Munoz, M. Ruzhansky, and N. Tokmagambetov. Acoustic and Shallow Water Wave
Propagation with Irregular Dissipation. Functional analysis and its applications, 53(2): 153-156,
2019.
[MRT19b] J. C. Munoz, M. Ruzhansky, and N. Tokmagambetov. Wave propagation with irregular
dissipation and applications to acoustic problems and shallow waters. Journal de mathematiques
pures et appliquees, 123: 127-147, 2019.
[Nai68] M. A. Naı̆mark. Linear differential operators. Part II: Linear differential operators in Hilbert
space. With additional material by the author, and a supplement by V. È. Ljance. Translated
from the Russian by E. R. Dawson. English translation edited by W. N. Everitt. Frederick Ungar
Publishing Co., New York, 1968.
[NLN16] H. T. Nguyen, D. L. Le, V. T. Nguyen. Regularized solution of an inverse source problem
for a time fractional diffusion equation. Applied Mathematical Modelling. 40(19): 8244–8264,
2016.
[NR10] F. Nicola and L. Rodino. Global pseudo-differential calculus on Euclidean spaces, volume 4
of Pseudo-Differential Operators. Theory and Applications. Birkhäuser Verlag, Basel, 2010.
[Nig86] R. R. Nigmatullin. The realization of the generalized transfer equation in a medium with
fractal geometry, Phis. Stat. Sol., 133:299–318, 1986.
[OS12a] I. Orazov, M.A. Sadybekov. One nonlocal problem of determination of the temperature
and density of heat sources. Russian Mathematics. 56(2):60–64, 2012.
 INVERSE SOURCE PROBLEMS FOR POSITIVE OPERATORS 25

[OS12b] I. Orazov, M.A. Sadybekov. On a class of problems of determining the temperature and den-
sity of heat sources given initial and final temperature. Siberian Mathematical Journal. 53(1):146–
151, 2012.
[RS80] M. Reed and B. Simon. Methods of Modern Mathematical Physics, volume 1 of Functional
Analysis, revised and enlarged edition. Academic Press, 1980.
[Roc78] C. Rockland. Hypoellipticity on the Heisenberg group-representation-theoretic criteria.
Trans. Amer. Math. Soc., 240:1–52, 1978.
[RR18] D. Rottensteiner, M. Ruzhansky. Harmonic and Anharmonic Oscillators on the Heisenberg
group. https://arxiv.org/abs/1812.09620
[RS76] L. P. Rothschild and E. M. Stein. Hypoelliptic differential operators and nilpotent groups.
Acta Math., 137:247–320, 1976.
[RT16] M. Ruzhansky, N. Tokmagambetov. Nonharmonic analysis of boundary value problems. Int.
Math. Res. Notices, 2016(12): 3548–3615, 2016.
[RT17a] M. Ruzhansky, N. Tokmagambetov. Nonharmonic analysis of boundary value problems
without WZ condition. Math. Model. Nat. Phenom., 12(1):115–140, 2017.
[RT17b] M. Ruzhansky, N. Tokmagambetov. Very weak solutions of wave equation for Landau
Hamiltonian with irregular electromagnetic field. Lett. Math. Phys., 107:591–618, 2017.
[RT18] M. Ruzhansky and N. Tokmagambetov. On a very weak solution of the wave equation for a
Hamiltonian in a singular electromagnetic field. Math. Notes, 103(5–6):856–858, 2018.
[RT18b] M. Ruzhansky and N. Tokmagambetov. Convolution, Fourier analysis, and distributions
generated by Riesz bases. Monatsh. Math., 187(1):147-170, 2018.
[RT17c] M. Ruzhansky, N. Tokmagambetov. Wave equation for operators with discrete spectrum
and irregular propagation speed. Arch. Ration. Mech. Anal., 226:1161–1207, 2017.
[RT18c] M. Ruzhansky and N. Tokmagambetov. Nonlinear damped wave equations for the sub–
Laplacian on the Heisenberg group and for Rockland operators on graded Lie groups. Journal
of Differential Equations, 265(10):5212–5236, 2018.
[RTT19] M. Ruzhansky, N. Tokmagambetov, and B.T. Torebek. Bitsadze-Samarskii type problem
for the integro-differential diffusion-wave equation on the Heisenberg group, Integral transforms
and special functions, 2019, to appear.
[RT19a] M. Ruzhansky and N. Tokmagambetov. Wave Equation for 2D Landau Hamiltonian. Ap-
plied and computational mathematics, 18(1): 69-78, 2019.
[RT19b] M. Ruzhansky and N. Tokmagambetov. On nonlinear damped wave equations for positive
operators. I. Discrete spectrum. Differential and integral equations, 32(7-8): 455-478, 2019.
[SY11] K. Sakamoto, M. Yamamoto. Inverse source problem with a final overdetermination for
afractional diffusion equation. Math. Control Relat. Fields. 1: 509–518, 2011.
[Sim14] T. Simon. Comparing Frechet and positive stable laws. Electron. J. Probab. 19:1–25, 2014.
[TT16] N. Tokmagambetov, B.T. Torebek. Fractional Analogue of Sturm–Liouville Operator. Doc-
umenta Mathematica. 21: 1503–1514, 2016.
[TT18] N. Tokmagambetov, B.T. Torebek. Well–posed problems for the fractional Laplace equation
with integral boundary conditions. Electronic Journal of Differential Equations. 2018: 1–10,
2018.
[TT18d] N. Tokmagambetov, B. T. Torebek. Green’s formula for integro–differential operators. J.
Math. Anal. Appl., 468(1):473–479, 2018.
[TT18a] Tokmagambetov N., Torebek B.T. Fractional Sturm–Liouville Equations: Self–Adjoint
Extensions. Complex Analysis and Operator Theory, 13(5):2259–2267, 2019.
[TT17] B.T. Torebek, R. Tapdigoglu. Some inverse problems for the nonlocal heat equation with
Caputo fractional derivative. Mathematical Methods in the Applied Sciences. 40(18): 6468–6479,
2017.
[TW18] J. Tyson, J. Wang. Heat content and horizontal mean curvature on the Heisenberg group.
Communications in Partial Differential Equations. 43(3): 467–505, 2018.
[Uch13] V.V. Uchaikin. Fractional Derivatives for Physicists and Engineers. V. 1, Background and
Theory. V. 2, Application. Springer, 2013.
26 MICHAEL RUZHANSKY, NIYAZ TOKMAGAMBETOV, AND BERIKBOL T. TOREBEK

[ZX11] Y. Zhang, X. Xu. Inverse source problem for a fractional diffusion equation. Inverse Prob-
lems. 27: 035010, 2011.
[ZW13] Z. Q. Zhang, T. Wei. Identifying an unknown source in time-fractional diffusion equation
by a truncation method. Appl. Math. Comput. 219: 5972–5983, 2013.
[WYH13] W. Wang, M. Yamamoto, B. Han. Numerical method in reproducing kernel space for an
inverse source problem for the fractional diffusion equation. Inverse Problems. 29: 095009, 2013.
[Wey31] H. Weyl, The Theory of Groups and Quantum Mechanics, Methuen, 1931.

Michael Ruzhansky:
Department of Mathematics: Analysis, Logic and Discrete Mathematics
Ghent University, Belgium
and
School of Mathematical Sciences
Queen Mary University of London
United Kingdom
E-mail address michael.ruzhansky@ugent.be

Niyaz Tokmagambetov:
Al–Farabi Kazakh National University
71 Al–Farabi ave., Almaty, Kazakhstan
and
Department of Mathematics: Analysis, Logic and Discrete Mathematics
Ghent University, Belgium
and
Institute of Mathematics and Mathematical Modeling
125 Pushkin str., Almaty, Kazakhstan
E-mail address niyaz.tokmagambetov@ugent.be

Berikbol T. Torebek:
Al–Farabi Kazakh National University
71 Al–Farabi ave., Almaty, Kazakhstan
and
Department of Mathematics: Analysis, Logic and Discrete Mathematics
Ghent University, Belgium
and
Institute of Mathematics and Mathematical Modeling
125 Pushkin str., Almaty, Kazakhstan
E-mail address berikbol.torebek@ugent.be