Applied Mathematics and Computation 293 (2017) 423–437

Contents lists available at ScienceDirect

Applied Mathematics and Computation

journal homepage: www.elsevier.com/locate/amc

Analytical solutions for heat diffusion beyond Fourier law

K.V. Zhukovsky a,∗, H.M. Srivastava b,c
a
Faculty of Physics, Moscow State University, Leninskie Gory, Moscow 119991, Russia
b
Department of Mathematics and Statistics, University of Victoria, Victoria, British Columbia V8W 3P4, Canada
c
Department of Medical Research, China Medical University Hospital, China Medical University, Taichung 40402, Taiwan, Republic of
China

a r t i c l e i n f o a b s t r a c t

Keywords: We obtain solutions for differential equations, describing a broad range of physical prob-
Inverse operator lems by the operational method with recourse to inverse differential operators, integral
Schrödinger equation
transforms and operational exponent. Generalized families of orthogonal polynomials and
Guyer–Krumhansl equation
special functions are also employed with recourse to their operational definitions. The evo-
Hermite polynomials
lutional type problems for heat transfer in various heat conduction models are studied.
Exact analytical solutions for Guyer–Krumhansl hyperbolic heat equation are obtained and
compared with those of Fourier and Cattaneo equations. Modelling heat pulse propaga-
tion from a laser source is performed in the framework of Fourier, Cattaneo and Guyer–
Krumhansl heat transfer models. Compliance of obtained solutions with the maximum
principle is studied.
© 2016 Published by Elsevier Inc.

1. Introduction

Differential equations (DE) play very important role in mathematics. Moreover, they are of great importance also in
physics since they describe many physical processes. Sometimes, the same DE describes more than one physical process
in different branches of physics. Thus, obtaining the solutions for differential equations is of paramount importance. Few
types of differential equations have explicit analytical solutions. During the last decade the numerical calculations and com-
puter methods have advanced greatly. They facilitate equations solving. Numerous numerical methods for solving differential
equations exist (see, for example, [1–6]. However, understanding of the obtained solutions and of the interplay of various
parameters in them can be best done in analytical form. To this end expansion in series of generalized forms of orthog-
onal Hermite, Laguerre and others polynomials are employed (see, for example, [7–11]). Operational approach allows easy
solving of a broad class of differential equations and relevant physical problems. It has been successfully applied for treat-
ment of the problems, related to propagation and radiation of accelerated charges, undulator radiation (UR) [12–15], heat
and mass transfer [16–19]. In these studies generalized forms of special functions and polynomials naturally arise [20–23].
Some examples of solutions of the heat diffusion and other equations by the inverse derivative method were considered
in [16,24,25]. The studies of hyperbolic heat conduction phenomena were reviewed in [26,27]. Extensive study is presented
in the book [28]. Other important studies were conducted, for example, in [29–32]. In what follows we shall apply opera-
tional approach, combined with integral transforms and extended forms of orthogonal polynomials to obtain exact analytical
solutions for heat diffusion equations beyond Fourier model.

∗
Corresponding author.
E-mail addresses: zhukovsk@physics.msu.ru (K.V. Zhukovsky), harimsri@math.uvic.ca (H.M. Srivastava).

http://dx.doi.org/10.1016/j.amc.2016.08.038
0 096-30 03/© 2016 Published by Elsevier Inc.
424 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437

2. Fourier and non-Fourier heat conduction

The Fourier’s law [33] is an explicit scheme of heat conduction modelling within continuum physics as it provides an
excellent agreement between theory and experiment for more than 90% of heat propagation cases. It relates linearly the
cause—the temperature gradient, to the effect—the heat flux:

∂t T = α∇ 2 T , (1)

where α = kT denotes heat diffusivity. The solution of the above equation with the initial condition T (x, 0 ) = f (x ) is given
by the Gauss-type integral:
  
1 ∞
(x − ξ )2
T (x, t ) ≡ Sˆ f (x ) = √ exp − f ( ξ )d ξ , (2)
2 π αt −∞ 4α t

where the heat operator Sˆ reads as follows:

Sˆ = exp(αt ∂x2 ). (3)

However, Fourier’s law has some contradictions and shortcomings, among which, for example, infinite speed of the heat
signal propagation. Indeed, it was noted by L. Onsager in 1931 that the Fourier’s model was in contradiction with the
principle of microscopic reversibility [34] and this contradiction “is removed when we recognize that is only an approximate
description of the process of conduction, neglecting the time needed for acceleration of the heat flow”. The Fourier’s law
is unsuitable for description of many low temperature (<25°K) heat phenomena. Moreover, it has the following unphysical
property: if an instant temperature disturbance is applied at a point in the solid, it is felt instantly at any distance in
the body. To describe the so-called second sound, when temperature disturbance propagates like damped waves, Cattaneo
[35] proposed a time-dependent relaxational model, which yielded in one dimension the following equation:

(τ ∂t2 + ∂t )T = kT ∇ 2 T , (4)

where kT is the thermal diffusivity and the relaxation time τ in heat conduction is extremely small (τ ≈ 10–13 s) at
roomtemperature. Cattaneo’s eq. (4) describes heat transfer in the form of wave propagation at finite velocity: the ratio
vt = kT /τ is indeed a velocity like quantity, representing the speed of the heat wave in the medium. The material pa-
rameter τ is an intrinsic thermal property, representing the build-up period for the start of a heat flow after a temperature
disturbance was imposed at the boundary of the domain. In Cattaneo’s model the heat flow starts with a delay time τ after
the application of the temperature gradient. This relaxation time τ is associated with the linkage time of phonon–phonon
collision, needed for the heat to start flowing, and in a sense, it measures the thermal inertia of the matter. Eq. (4) is
the simplest model of the second sound phenomenon observed first in liquid Helium [36]. Later on, the analysis of the
theoretical background resulted in the observation of second sound also in solid crystals [37], via properly designed experi-
ments [38–40]. In these tests the heat pulse technology was crucial for the sensitive detection of the thermal diffusivity. In
what follows we will demonstrate how to analytically solve Cattaneo’s equation and its extended forms by the operational
method. However, even Cattaneo’s eq. (4) does not eliminate all the problems. Although it leads to a finite value for the
wave velocity, the latter is not in agreement with ultrasonic wave propagation in dilute gases and it does not describe heat
pulse propagation in non-metallic very pure crystals, like Bi or Na F at very low temperature. So-called ballistic transport
should be taken into account; it occurs when the mean free path of a particle significantly exceeds the dimension of the
medium in which it travels. For example, for electrons in a medium with negligible electric resistance, their motion is al-
tered by collisions with the walls. The mean free path can be increased by reducing the number of impurities in a crystal
or by lowering its temperature. Ballistic conduction is not limited to electrons but can also apply to phonons. The ballistic
conditions are typical for very thin films and wires. It is typically observed in such quasi one- or two-dimensional structures
as graphene, carbon nanotubes, silicon nanowires etc. (see, for example, [41,42–44,45]). In the ballistic conditions, when the
mean free path of a particle significantly exceeds the dimension of the medium in which it travels, neither Casimir phonon
[46] nor Fourier diffusion [33] theories are accurate and thermal transfer is influenced by both internal and boundary scat-
tering. Most known further generalization of the Fourier law is based on the Guyer–Krumhansl (GK) equation [47], which in
one dimension reduces to the following form:
 
∂t2 + ε ∂t − δ∂t,x,x
3
T (x, t ) = α∂x2 T (x, t ). (5)

Guyer and Krumhansl solved the linearized Boltzmann equation for a phonon field in dielectric crystals at low tempera-
ture and derived a non-local extension of Cattaneo’s equation. It is well adapted to the description of phonon gases, where
heat transport is not only governed by diffusion (like in Fourier’s description) and second sound (like in Cattaneo’s model),
but, in addition, by ballistic transport. Moreover, some recent heat propagation experiments at room temperature on micro-
scopic [48] and on macroscopic samples [49–53] also demonstrate deviation from Fourier and Cattaneo’s law. The observed
phenomenon was considered typical for the GK relation [54], despite there was no phonon transport there. In what follows
we shall address these equations and obtain analytical solutions for them using operational approach. Other extensions and
modifications of the GK equation predict more terms and they will be discussed elsewhere.
 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437 425

3. Operational study of laser pulse heat diffusion with Fourier equation

Before we address second order of time partial differential equations we consider the example of the Fourier heat diffu-
sion. It is instructive because of it underlines that many equations in mathematical physics may describe different physical
processes at the same time. For example, one-dimensional Schrödinger equation for a charge in constant electric field

i ∂τ (x, τ ) = −∂x2 (x, τ ) + b x (x, τ ) (6)

was studied by the operational method [16,18]. For particle propagation under a potential barrier it transforms upon t → iτ
change into the following Fourier type heat equation with linear coordinate term:

∂t F (x, t ) = α∂x2 F (x, t ) + β x F (x, t ). (7)

Eq. (7) does make sense in the imaginary time formalism, i.e. in the Euclidean picture in quantum mechanics (see, e.g.
[55]). This equation is important in the studies of quantum chromodynamics vacuum where it describes tunnelling of a
particle through a region, where potential energy is greater than the particle energy. Its two-dimensional generalization
describes charged particle diffusion under a potential barrier in constant electric field, when the particles energy is lower
than the potential energy and it reads as follows:
 
∂t F (x, y, t ) = α∂x2 + β∂x ∂y + γ ∂y2 + bx + cy F (x, y, t ), min (α , β , γ ) > 0, (8)
where the initial condition isF (x, y, 0 ) = f (x, y ). Following the lines of [18], we obtain the two-dimensional generalization of
the solution for Eq. (7) as sequential operational action:

F (x, y, t ) = e 
ˆ x
ˆ y Eˆ f (x, y ), (9)
where

= (α b2 + γ c2 + β bc )t 3 /3 + t (bx + cy ) (10)
is the phase and
∂

ˆ x = et 2
(α b+β c/2)∂x , 
ˆ y = et 2
(γ c+β b/2)∂y , ˆ = e 2α t Dx ≡ e 2α t ∂ x ,
 ˆ f ( x ) = f ( x + 2α t )
 (11)
are the well-known operators of translation. The heat diffusion is governed by the operator

 
Eˆ = exp t α∂x2 + β∂x ∂y + γ ∂y2 . (12)

The explicit double integral form of the operator Eˆ was obtained in [24]; it is the Gauss type integral, which we omit
here for the sake of conciseness. If β = 0, the heat diffusion is executed by the one-dimensional operators (3) Sˆx Sˆy . Direct
usage of the operational definition (12) with the initial function f (x, y ) = xm yn yields the solution of (8):
  
F (x, t ) = e Hm,n x + t 2 (α b + β c/2 ), t α ; y + t 2 (γ c + β b/2 ), t γ t β , (13)
where generalized Hermite polynomials Hm,n (x, α ; y, γ |β ) reduce to Hk (x, y ) as follows:
min (m,n )
βs
Hm,n ( x, α ; y, γ |β ) = m!n! Hm−s (x, α )Hn−s (y, γ ). (14)
s ! ( m − s )! ( n − s )!
s=0

The Hermite polynomials Hn (x, y ) can be defined operationally, by sums or by generating function as follows:
 
∂2 n
[n/2]
xn−2r yr
∞
tn  
Hn (x, y ) = exp y x = n ! , Hn (x, y ) = exp xt + yt 2 . (15)
∂x 2
r=0
(n − 2r )! r! n=0 n!
They can be reduced to the well-known forms of Hermite polynomials of single variable:
   
ix x
Hn (x, y ) = (−i ) y n n/2
Hn √ = i ( 2y )
n n/2
H en  . (16)
2 y i 2y

While the physical meaning of the above solution is tunnelling of a charge in electric field through a region, where
potential energy exceeds the particle energy, it can be also interpreted as two-dimensional Fourier heat diffusion, where
the linear coordinate term is not that important. For sensitive detection in thermal diffusivity tests heat pulse technology
has been established [56], thus the study of intense instant heat surge propagation is valuable. In view of that we consider
the example of the initial δ -function f (x ) = δ (x ), which models a very short point heat burst by a laser beam. It is easy to
demonstrate that the action of the heat diffusion operator on δ -function results Gauss distribution
  1  
Sˆδ (x ) ≡ exp αt ∂x2 δ (x ) = √ exp −x2 /4at . (17)
2 π αt
426 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437

Fig. 1. Solution F(x, t) of the Fourier heat equation with initial δ (x) function for α = 1, β = 10 for the interval of time t ∈ [0.01, 0.15].

Then the one-dimensional solution of (7) for the initial condition f (x ) = δ (x ) simply reads

(x+αβt 2 )
2

−
e
4t
F (x, t ) = e √ , (18)
2 π αt
where (x, t; α , β ) = 13 αβ 2 t 3 + β t x. Observe slight asymmetry of the solution for α = 1 due to β = 10 in Fig. 1.
For long periods of time the solution rises exponentially due to β
= 0 and
= 0 (see [57]). Two dimensional δ -function
transforms upon the action of the heat operators Sˆ on it into
   
exp − x2 /α + y2 /γ /4t
f (x, y, t ) = Sˆx Sˆy δ (x, y ) = , (19)
4π t αγ
operators 
ˆ x and 
ˆ y produce space shift and yield the particular solution, which evolves into Gaussian as follows:
  2  2 
x + t 2 α b /α + y + t 2 γ c /γ

F (x, y, t ) = e  ˆ y Sˆx Sˆy δ (x, y ) = e
ˆ x exp − . (20)
4π t αγ 4t

It is easy to extend the obtained results to three dimensions too. We omit it for brevity.

4. Hyperbolic heat equation (HHE)—telegraph equation

The following hyperbolic heat conduction equation, in the case of κ

= 0 also known as telegraph type equation, was
proposed for more accurate description of the heat conduction phenomena at low temperatures:
   2 
∂2 ∂ ∂
+ ε F ( x, t ) = α + κ F (x, t ), F (x, 0 ) = f (x ), F (x, ∞ ) < ∞. (21)
∂t2 ∂t ∂ x2
The operational method for solution of differential equations can be successfully applied for the second order of time
partial differential equations. Among them the equations of the following type:
 
∂2 ∂
+ ε
ˆ ( x ) F (x, t ) = Dˆ (x )F (x, t ), (22)
∂t2 ∂t
where Dˆ (x ) is a differential operator, acting over the coordinate, are of particular importance. The l.h.s. of these equations
cover a variety of species, customized for description of a broad range of physical processes. Perhaps the simplest example is
given by εˆ (x ) = ε and D(x ) = ∂x2 ; this set forms the well known hyperbolic heat equation HHE. More complicated operators
can be considered; importantly, we may also include the extensions, containing ∂t , which yield HHE of Guyer–Krumhansl
type. To begin with we recall that the general solution of Eq. (22) has the following form:
tε
 t√ √ 
F (x, t ) = e− 2 e− 2 ε +4D(x )C1 (x ) + e 2 ε +4D(x )C2 (x ) ,
2 ˆ t 2 ˆ
(23)

where C1,2 (x ) are to be determined from the initial or boundary conditions. Let us denote the initial condition F (x, 0 ) = f (x )
and require that the solution of Eq. (22) is bounded at t → ∞ times, which is a reasonable supposition for many physical
 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437 427

applications. Other boundary conditions are certainly possible; they will be considered elsewhere. The above choice sets
C2 (x ) = 0 and the remaining branch of the solution is subject to the Laplace transforms

√ t ∞
dξ t2
e−t V
= √  e − 4 ξ −ξ V , t > 0 . (24)
2 π 0 ξ ξ
Thus, we obtain for Eq. (22) the following bounded at infinite time solution:

εt t ∞
dξ t2
F (x, t ) = e− 2  e− 16ξ e−ξ ε e−4ξ Dˆ (x) f (x ),
2
√ (25)
4 π 0 ξ ξ
provided the integral converges. The particular form of the solution depends on the operator Dˆ (x ) and on the initial function
f (x ). With account for (25) the bounded solution of (21) for the initial function f (x )reads as follows:

t ∞
dξ
 e− 16ξ −ξ (ε ) Sˆ f (x ),
t2
εt +4κ
Sˆ f (x ) = e−4αξ ∂x f (x )
2
F (x, t ) = e− 2
2
√ (26)
4 π 0 ξ ξ
and according to the operational identity for the exponential of a squared operator pˆ [58]:
   ∞  
1
exp pˆ2 = √ exp −ξ 2 + 2ξ pˆ dξ (27)
π −∞

we obtain the following integral form of the solution:

εt t
 ∞
du − t 2 −u(ε2 +4κ ) 1
 ∞   
F (x, t ) = e− 2 e −v f x + 2 i v ζ dv, ζ = 4uα .
2
√ √ e 16u √ (28)
4 π 0 u u π −∞

The example of the initial function f (x ) = xn produces Hermite polynomials upon the action of the operator Sˆ on it (see
(15)). Then the solution of Eq. (21) for f (x ) = xn reads
  
−t ε /2 t ∞
Hn (x, −4α u ) t2  
F (x, t )| f (x )=xn = e √ √ exp − − u ε 2 + 4κ du
4 π 0 u u 16u
 √  r
tε t ε 2 + 4κ
[n/2]
(−α )r (x )n−2r t
t  
= e− 2 n! √ K 1 −r ε 2 + 4κ . (29)
π r=0
(n − r )!r! ε 2 + 4κ 2 2

For κ = 0, n = 2, 3, 4 we easily obtain simple forms, such as

F (x, t )|n=2 = e−t ε (x2 − 2t α /ε ), F (x, t )|n=3 = e−t ε x(x2 − 6t α /ε ) etc (30)
The other physical meaning of Eq. (21) is related to the phenomenon, occurring with heat propagation at low temper-
atures: the second sound, observed first in liquid Helium [36]. To address this phenomenon, Cattaneo’s equation [35] (4)
for the temperature was introduced, which reduces in one dimension to the well-known telegraph eq. (21) with κ = 0.
Possible presence of κ
= 0 in the context of the heat conduction may be interpreted as radiative loss of heat due to small
temperature excess T over the surroundings. The operational solution of the hyperbolic heat eq. (21) in three dimensions
generalizes (26) as follows:

t ∞
dξ
 e− 16ξ −ξ (ε ) Sˆx Sˆy Sˆz f (x, y, z ),
t2
εt 2
+4κ
F (x, y, z, t ) = e− 2 √ f (x, y, z ) = F (x, y, z, 0 ) (31)
4 π 0 ξ ξ
and consists in the action of the heat diffusion operators for each coordinate on the initial function in complete analogy
with (28), expressed via multiple integrals of Gauss type.
To model the established experimental technique of an instant laser heat burst we consider the example of the initial
δ -function f (x ) = δ (x ) for HHE Eq. (21). δ -function√transforms into a Gaussian function upon the action of Sˆ = e−4αξ ∂x (17)
2

and we complete the integration in (26) for t > x/ α to obtain the following one-dimensional solution:
   
− ε2t t ∞
dξ   t 2 − x2 /α − ε2t t ω √ 
F (x, t )| f (x )=δ (x ) = e exp −ξ ε + 4κ
2
− =e K ωυ ,
16π α 1/2
0 ξ 2 16ξ 4π α 1/2 υ 1
ω = ε 2 + 4κ > 0, 4υ = t 2 − x2 /α >0. (32)
√
The condition t > x/ α can be interpreted in the sense of the time, greater than the time, within which the initial
√
perturbation by δ (x) surge reaches point x in space, α being the velocity of the surge propagation. For κ < 0 in (21)
the integral in (32) converges for ε > −4κ . Despite the solution (32) was obtained for ω > 0 and υ > 0, it is valid for any
2

ω
= 0 and υ
= 0. The values ω < 0 and υ < 0 yield F (x, t ) > 0, for ω < 0 and υ > 0 or for ω > 0 and υ < 0 we obtain complex
εt
values for the solution F (x, t ). For ω → 0, υ
= 0 we have F (x, t )|ω→0 = e− 2 4π α 1t/2 |υ| . For ω
= 0, υ → 0 the solution (32)
diverges. Some examples of solutions of Eq. (21), where ε = 1, κ = 1, α = 1 for initial δ (x) function at given times t = 0.001,
0.01, 0.05, 0.10 are shown in Fig. 2.
428 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437

Fig. 2. Solutions F(x) of hyperbolic heat equation with initial δ (x) function for t = 0.001, 0.01, 0.05, 0.10.

Observe that the behaviour of the solution of Eq. (21) with f (x ) = δ (x )(see Fig. 2) differs from that of the solution of
Fourier heat eq. (7), which exhibits the reduction of the amplitude and the spread of the initial function, as shown in Fig.
1 for α = 1, β = 1 (see (7)). Contrary to Fourier law, formula (32) describes propagation of the instant surge at a finite speed.
The surge is followed by the spread as seen in Fig. 2; its amplitude reduces first rapidly and then slowly vanishes to zero.
We also evaluate the solution for α = 105 , ε = 103 , κ = 102 in the moment t = 0.1: F (t = 0.1 )|α =105 ,ε=103 ,κ =102 < 10−20 . For
this set of parameters the Cattaneo’s second time-derivative plays minor role in (21) and the solution fades out rapidly and
at t = 0.1 it totally vanishes.
Consider three-dimensional δ (x, y, z ) initial condition, modelling a laser pulse heating in a volume in (0, 0, 0) point. Then
x2 +y2 +z 2
e 16αξ
Sˆx Sˆy Sˆz δ (x, y, z ) = and the solution of the hyperbolic heat conduction equation with the initial δ (x, y, z) condition
16 (π |αξ | )3/2
reads as follows:
  
εt t ∞
dξ   t 2 − (x2 + y2 + z2 )/α
F (x, y, z, t ) = e− 2 exp −ξ ε 2 + 4κ − . (33)
64π 2 α 3/2 0 ξ3 16ξ

The above integral converges for t > (x2 + y2 + z2 )/α and we obtain the following simple analytical expression, involv-
ing modified Bessel functions K2 :
εt t ω √ 
F (x, y, z, t ) = e− 2 K ωυ , ω = ε 2 + 4κ > 0, 4υ = t 2 − (x2 + y2 + z2 )/α > 0. (34)
8π 2 α 3/2 υ 2
The above solution also holds for ω
= 0 and υ
= 0. For ω → 0, υ
= 0 we obtain the limit
εt t 1
F (x, y, z, t )|ω→0 = e− 2 . (35)
4 π 2 α 3/2 υ 2
As well as in one-dimensional case (32) for ω < 0, υ < 0 we obtain positive values for F (x, t ), while for ω < 0, υ > 0 as
well as for ω > 0, υ < 0 we obtain complex values for the resulting function F (x, t ). For ω
= 0, υ → 0 the above solution
diverges. Note that hyperbolic eq. (21) in the context of the heat propagation has also the advantage of being reversible
within periods of time of the order of the thermal relaxation time τ , unlike the irreversible in time Fourier propagation
process.
Cattaneo’s relation was successful in qualitative description
 of the second sound and other phenomena, but quantita-
tive prediction for the finite speed of the heat wave kT /τ disagrees with experiments. Moreover, there are mathematical
contradictions, regarding this equation, such as it allowed negative solutions and the maximum principle was not satisfied.
According to the latter, for u(x, t), satisfying the heat equation in the rectangle R = {0 ≤ x ≤ l, 0 ≤ t ≤ T} in space-time, the
maximum value of u(x, t) over the rectangle is assumed either initially (t = 0), or on the lateral sides (x = 0, or x = l)), and
for the hyperbolic heat equation it does not hold [59–61]. This is possible from mathematical viewpoint since the maximum
principle is established for elliptic and parabolic DE. The essence of the maximum principle is related to the Second Law
of thermodynamics. The natural flow of heat is downwards and therefore without any external work heat flows from high
temperature places to low temperature places. It ensures uniqueness and stability of solutions for heat equation. Indeed,
for two solutions, satisfying the same initial and boundary conditions, their difference satisfies zero initial and boundary
conditions and consequently its maximum and minimum is zero, which means that the difference between two solutions
is identical zero. Moreover, some physical contradictions, related to Cattaneo’s equation, were also noted (see, for example,
[62] and [63]); new terms were added to the equation (see review in [64]).
The telegraph eq. (21) can be modified by adding commuting and non-commuting with ∂x2 terms and, accounting for the
arising in the exponential commutators, the solution can be obtained. Let us consider Eq. (22), where εˆ (x ) is the operator.
 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437 429

The bounded at t → ∞ solution, developing from f (x ), follows from (25):


t ∞
dξ t2
F (x, t ) = e− 2 εˆ(x ) √  e− 16ξ e−ξ εˆ (x ) e−4ξ Dˆ (x ) f
t 2
(x ) (36)
4 π 0 ξ ξ
Further calculations depend on the explicit form of the operators εˆ (x ) and Dˆ (x ) and on the commutator [εˆ2 , Dˆ ], which
may help to disentangle them.

5. Guyer–Krumhansl type extended HHE with mixed derivative

The hyperbolic heat eq. (21) predicts finite heat wave propagation speed, but the values do not agree with experimental
results of ultrasonic wave propagation in dilute gases and heat pulse propagation in non-metallic pure crystals of Bi or Na
F at low temperature <25°K. Guyer and Krumhansl derived a non-local extension [47] of Cattaneo’s equation. Indeed, Eq.
(22) becomes suitable for description of heat propagation if completed by the set of operators

εˆ = ε − δ∂x2 and Dˆ = α∂x2 + κ 2 (37)

which yields the following modified hyperbolic heat equation with the mixed derivative term:
   2 
∂2 ∂ ∂3 ∂
+ε −δ F (x, t ) = α 2 + κ F (x, t ), α , ε , δ, κ = const
2
(38)
∂t2 ∂t ∂ t ∂ x2 ∂x
Eq. (38) is the GK type heat conduction equation, frequently written as follows:
 2   
∂ ∂ ∂3 ∂2
τ 2+ F (x, t ) = kb + k + μ F (x, t ), (39)
∂t ∂t ∂ t ∂ x2 T
∂ x2
where

τ = 1/ε, μ = κ 2 /ε , kT = α /ε , kb = δ /ε (40)
Similar form of the heat conduction equation arises in the study of heat propagation in one-dimensional thin film, which
thickness may be of the same order of magnitude as, or even smaller than the mean free path of the phonons [65]. Note,
that for α = δε the solution of Fourier eq. (1) automatically satisfies GK eq. (5) or (39) with κ = 0. According to (36) we
now obtain the following solution for the extended GK equation with κ
= 0:

t ∞
dξ t2
( 2t δ+ξ (2εδ−4α ))∂x2 e−ξ δ2 ∂x4 f (x )
F (x, t ) = e− 2 ε √  e− 16ξ −ξ (ε +4κ 2 )+
t 2
(41)
4 π 0 ξ ξ
and with the help of identity (27) we reduce the exponential as follows:
 ∞ 
1
exp(−ξ δ 2 ∂x4 ) f (x ) = √ exp(−ζ 2 + 2iζ δ ξ ∂x2 ) f (x )dζ . (42)
π −∞

Grouping the terms for the 2nd order derivative in the exponential, yields the particular bounded solution of the modi-
fied hyperbolic heat conduction eq. (38) with initial function f (x ) as follows:
 
e− 2 ε t ∞
dξ ∞
t

 e− 16ξ −ξ (ε )
t2
+4κ 2
e−ζ Sˆ f (x )dζ ,
2 2
F (x, t ) = (43)
4π 0 ξ ξ −∞

where

Sˆ = eν∂x , ν = a + ibζ , a = t δ /2 − 4ξ α + 2ξ εδ, b = 2 ξ δ
2
(44)

The above solution consists in the integrated weighted action of the heat diffusion operator Sˆ on the initial functions
and it is valid even if the constant term in the r.h.s. of Eq. (38) is negative: κ 2 < 0 for ε > 2|κ|, which insures the integral
convergence. Equations with negative sign of κ 2 appear for ballistic heat propagation in thin films and wires (see [65]). In
what follows we shall consider few physical examples of initial functions.
For heat conductivity measurements it is particularly important to consider the heat surge propagation F (x, 0 ) = δ (x ),
modelling ultra short pulse laser heating in the point x = 0 at the moment of time t = 0. Employing operational method,
it is easy to obtain from Eqs. (17), (43) and (44) the following exact analytical solution of Guyer–Krumhansl type heat eq.
(38), (39) with initial δ -pulse:

e− 2 ε t ∞
dξ
t

 e− 16ξ −ξ (ε ) (x; a, b) ,
t2 2
+4κ 2
F (x, t ) = (45)
8 π 3/2 0 ξ ξ
where we denote the special function
430 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437

Fig. 3. Solution of Guyer–Krumhansl heat equation with initial δ (x) function for α = ε = δ = κ = 1.

 2
−ζ 2 − 4(ax+ibζ )
∞
e dζ
(x; a, b) ≡  ,
−∞ |a + ibζ |

a = t δ /2 − 4ξ α + 2ξ εδ , b = 2δ ξ. (46)

The function (x; a, b) reduces for some particular values of a, b to well-known special functions, such as hypergeometric
function 0 F2 (; {β1 , β2 }; z ) , this last being the particular case of p Fq ({α1 ...α p }; {β1 ...βq }; z ) . For example,
1 
(0; 0, b)=  / |b| (47)
4

also note that

 2            
e −ζ − 4xbζ
2
∞
dζ 1 1 3 x4 π 3 5 x4
 =  F
0 2 ; , ; − − |x | F
0 2 ; , ; − sgn b. (48)
−∞ |ζ | 4 2 2 4 64b 2 b 4 4 64b

Careful numerical calculations of the double integral (45), accounting for the point a = 0, i.e. t = 4ξ (2 αδ − ε ), yield Real
result. The evolution (45), (46) of the initial pulse F (x, 0 ) = δ (x )for α = δ = ε = κ = 1, when all four heat transfer contri-
butions equal each other, is demonstrated in Fig. 3. The plot in Fig. 3 looks rather like the solution of the Fourier heat eq.
(7) (see Fig. 1) than the solution of the hyperbolic heat eq. (21) (see Fig. 2). Higher values of the coefficient δ of the mixed
GK derivative cause more spread of the solution, i.e. higher effective heat conductivity, which is evidenced by the plot for
α = κ = 1, ε = 0.4, δ = 4 in Fig. 4, which practically does not change for α = 5, κ = 1, ε = 2, δ = 4.
The behaviour of the solution after the initial sharp drop is better illustrated in Fig. 5, where the spread is well visible.
Comparing Fig. 1 with Fig. 5 we conclude that the solution of the Fourier heat eq. (7) at short times has higher and sharper
peak than that of GK equation. Varying ε does not influence the solution so much; for ε = 0.1, δ = 1, α = 1 we obtain little
difference from the plot in Fig. 5, where α = ε = δ = 1. The increase of κ gradually boosts the spread.
Significantly smaller values of δ result in less spread as seen in Fig. 6 for δ = 0.05, ε = 2, α = 1. Such small value of the
coefficient of the 3rd order derivative in GK e. (38) sharpens the peak of the solution (see Fig. 6) and helps preserving the
shape of the pulse for prolonged period of time (see Fig. 6), exceeding that for Fourier heat propagation.

Suppose the initial function can be approximated by a polynomial: f (x ) = n cn xn . Then it is reasonable to consider the
example of the initial monomial f (x ) = xn . Then upon the action of the operator Sˆ = eν∂x and with account for (44) the
2

solution of Eq. (39) reads

  
e− 2 ε t ∞
dξ ∞
t

 e− 16ξ −ξ (ε )
t2
+4κ 2
e−ζ Hn (x, a + iζ b)dζ ,
2
b = 2δ ξ
2
F (x, t ) = (49)
4π 0 ξ ξ −∞
 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437 431

Fig. 4. Solution of Guyer–Krumhansl heat equation with initial δ (x) function for α = κ = 1, ε = 0.4, δ = 4.

Fig. 5. Solution of Guyer–Krumhansl heat equation with initial δ (x) function for α = ε = δ = κ = 1.

Fig. 6. Solution of Guyer–Krumhansl heat equation with initial δ (x) function for α = κ = 1, ε = 2, δ = 0.05.
432 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437

Fig. 7. Solution of GK equation with equal each other heat transport contributions: α = ε = δ = 1 and positive linear term κ 2 = 0.5 for the initial function
F (x, 0 ) = x4 e−x .

The integration with account for (15) can be easily accomplished. For example, for f (x ) = x3 we obtain the following
simple solution of Eq. (39):
√
 
F (x, t )| f (x )=x3 = 4π xe− 2 (
t
V +ε ) x2 + 3t δ + √3t (δε − 2α ) , V = ε 2 + 4κ 2 . (50)
V
Note, that the divergency in the solution appears for negative values of κ 2 as, for example, for V = 0, i.e. κ 2 = −ε 2 /4 in
(50). These cases require careful account, when met in some models, describing ballistic effects for heat propagation in thin
films [65]. Similarly we can obtain the solution for any given n; we omit it here for the sake of conciseness.
It is interesting to consider the evolution of the initial function f (x ) = eγ x xn . We make use of the operational identity of
the operational identity
exp(yD2x )xk eα x = e(α x+α y ) Hk (x + 2α y, y ),
2
(51)
to obtain for arbitrary values of n ∈ Integers, γ ∈ Reals the following integral:
 
e 2 (γ δ −ε )+γ xt ∞ dξ − 16t 2ξ −ξ d ∞ −ζ 2 +ibζ γ 2
t 2

F (x, t ) =  e e Hn (x + 2γ ν, ν )dζ , (52)

4π 0 ξ ξ −∞

where

ν = a + ibζ , a = t δ /2 − 4ξ α + 2ξ εδ, b = 2 ξ δ, d = ε 2 + 4κ 2 + 2γ 2 (2α − εδ ). (53)
The above integration yields elementary functions, but it is cumbersome and we omit this expression for conciseness. In
the particular case of given n and γ , which simplest example is given by n = −γ = 1(F (x, 0 ) = e−x x), we obtain the following
simple solution of GK equation:
√  √ 
F (x, t ) = e−x+t (δ −ε )/2 e− 2 x − δt + (2α + δ (δ − ε ))t / q ,
t
q
(54)
where
q = ε 2 + 4κ 2 + δ 2 + 4α − 2εδ. (55)
Below we present spectacular example of evolution of the initial function f (x ) = e−x x4 , i.e. n = 4, γ = −1. The solution
for equal each other heat transport contributions α = ε = δ = 1 and positive linear term κ 2 = 0.5 is plotted in Fig. 7. In this
case all terms in Eq. (39) and their respective effects—heat exchange with the environment, Cattaneo’s term, diffusion and
ballistic heat transfer—are of comparable with each other order.
The evolution of the initial wave-like function F (x, 0 ) = x4 e−x qualitatively changes, dependently on the sign of the lin-
ear term κ 2 . If κ 2 > 0 in the r.h.s. of GK eqs. (38, 39), then the solution fades (see Fig. 7), exhibiting rather diffusion-like
behaviour. For κ 2 < 0 in (38), (39), i.e. when κ 2 > 0 term is present in the l.h.s. of GK eqs. (38, 39), the solution rises on one
of the sides of the domain and slowly fades as seen in Fig. 8 for equal each other heat transport contributions α = ε = δ = 1
and negative linear term κ 2 = –0.5.
Very interesting solution appears when the ballistic transport is dominant, relatively to all others. Compare Fig. 9 for
distinct ballistic transport α = ε = 1, δ = 10 and negative linear term κ 2 = –0.5 with Fig. 10 for distinct ballistic transport
α = ε = 1, δ = 10 and positive linear term κ 2 = +0.5. In this case Cattaneo’s second order time derivative and the Fourier
 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437 433

Fig. 8. Solution of GK equation with equal each other heat transport contributions: α = ε = δ = 1 and negative linear term κ 2 = –0.5 for the initial function
F (x, 0 ) = x4 e−x .

Fig. 9. Solution of GK equation with distinct ballistic transport: α = ε = 1, δ = 10 and with negative linear term κ 2 = –0.5 for the initial function F (x, 0 ) =
x4 e−x .

Fig. 10. Solution of GK equation with distinct ballistic transport: α = ε = 1, δ = 10 and with positive linear term κ 2 = +0.5 for the initial function F (x, 0 ) =
x4 e−x .
434 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437

Fig. 11. Solution of GK equation for small values of α = ε = δ = 10, κ 2 = +1 for the initial function F (x, 0 ) = x4 e−x .

Fig. 12. Solution of GK equation with distinct ballistic transport: α = ε = 1, δ = 10 and with negative linear term κ 2 = –0.5 for the symmetric initial function
F (x, 0 ) = |x|e−|x| .

diffusion play minor role, but the effect of the linear term is very different. Indeed, for negative values of κ 2 = –0.5 in (38),
(39), the solution grows with time for long and then fades very slowly (see Fig. 9). The difference in the fading time, as
compared with the plot in Fig. 8, is more than one order of magnitude.
The change of the sign of the linear term changes the solution behaviour radically. Indeed, for κ 2 = 0.5, when the ballistic
term is dominant α = ε = 1, δ = 10, and κ 2 = 0.5, the solution exhibits bizarre behaviour, oscillating between positive and
negative values in time and then fading to zero from positive values as seen in Fig. 10. It is totally different from the
behaviour of the solution for κ 2 = –0.5 in Fig. 9. Interestingly, in this case the non-negativity of the solution is not preserved!
The characteristic fading time for α = ε = 1, δ = 10, κ 2 = 0.5 is τ fade ≈ 30 and it greatly exceeds τ fade ≈ 1 for α = ε = δ = 1,
κ 2 = 0.5 (see Fig. 7).
The behaviour of the solution for relatively small coefficients of the Cattaneo’s term: α = ε = δ = 10, κ 2 = ±1 is shown
in Fig. 11. It does not depend on the sign of the linear term κ and also exhibits oscillations; in this case the solution first
grows and then flips to negative values and eventually fades to zero from negative values.
Eventually, note that the solution for the initial function F (x, 0 ) = |x|e−|x| simply does not preserve
√ the maximum prin-
ciple. Direct substitution of the solution (54), subject to change x → |x| with account for |x| ≡ x2 , satisfies GK equation.
During the evolution of the initial function its maximum over the time-space domain appears in the middle of the do-
main and not on its edges (see Fig. 12). Thus, the maximum principle, which guarantees the uniqueness and the stability
of the solution for elliptic and for parabolic equations, is not fulfilled in the case of GK equation, at least in our example
in the case, when the ballistic term dominates. This is possible, since GK equation is hyperbolic and the maximum princi-
ple is established for elliptic and parabolic equations. Moreover, generally speaking, the heat should flow from high to low
 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437 435

temperature values, while the evolution of F (x, 0 ) = |x|e−|x| exhibits different from that behaviour, which apparently contra-
dicts the Second Law of thermodynamics; the latter established for Fourier heat propagation.
In conclusion we note that solutions for a variety of the above considered and other DE can be also obtained by us-
ing operational approach (see, for example, [66–70]). The above developed operational technique can be applied for other
equations, involving more complicated differential operators. The examples will be considered in forthcoming publications.

6. Conclusions

Exact analytical solutions were obtained by means of operational method for three types of heat equations: Fourier,
Cattaneo and Guyer–Krumhansl. The initial δ -function heat pulse propagation was considered, modelling heat pulse exper-
imental technique, common for thermal conductivity measurements. The comparison of the obtained analytical solutions
with the common Fourier heat equation solution evidenced finite speed pulse propagation in the Cattaneo’s hyperbolic
model. The decisive role of the 2nd time derivative in it was illustrated. This term being small, the solution fades much
faster. The pulse propagation, obeying Cattaneo’s heat equation, is in striking contrast with that obeying Fourier equation.
We have found for HHE solution that the instant front moves at constant finite speed, leaving a spread behind, contrary
to smooth and spreading around solution for the Fourier equation. The analytical solutions for Fourier and hyperbolic heat
equations in three dimensions were written explicitly in modified Hermite polynomials and in Bessel functions. The spread
of the pulse was demonstrated.
Exact bounded analytical solution for the Guyer–Krumhansl type heat equation was obtained for arbitrary initial function
in terms of integrals and special functions. Propagation of the initial heat surge was investigated. We evidenced how small
values for the third order mixed derivative in GK equation make the spread of the solution slower. Variation of the coeffi-
cient for the first order time derivative does not influence the solution so much. The role of the constant heat dissipative
term in this case is not decisive.
Exact analytical solutions for GK and Cattaneo’s type equations with initial exponential–monomial function F (x, 0 ) =
xn e−γ x were obtained. They allow modelling the dynamics of heat propagation not only from a laser heat surge, but prac-

tically any pulse: we can get the result for the initial function, which can be expanded in series f (x ) = n,γ cn eγ xn or ap-
proximated by them. Such approximation gives better fit with experimental data for solitary wave phenomena, than approx-
imation by power series or by harmonic functions. The analytical solutions in terms of the generalized Hermite polynomials
and modified Bessel functions were obtained for all three types of heat equations: Fourier, Cattaneo and Guyer–Krumhansl
types. Some of them reduce to elementary functions.
The linear term plays important role in GK-type equations. If the positive linear term κ 2 > 0 is present in the r.h.s. of
GK-type equation, then the solution fades, exhibiting rather diffusion-like behaviour. If the linear term is negative κ 2 > 0, in
the r.h.s. of GK-type equation, as occurs in some models of heat conduction in thin films, then the solution exhibits different
behaviour: it may rise and fade or oscillate, dropping to negative values. Stability of obtained solutions with respect to the
auxiliary conditions can be determined, for example, based on the maximum principle. We have demonstrated that the
maximum principle and non-negativity for solutions is not preserved when the ballistic transport term dominates in GK
equation.
Thus, the operational approach to the analytical study of the second order heat conduction and related equations yielded
analytical solutions with clear physical meaning, evidenced the differences between the solutions and distinguished the
role of various terms in the equations. The obtained analytical results can be evaluated with the initial conditions good for
experimental and laboratory check and are be useful as from purely theoretical as from practical viewpoint. The obtained
solutions being exact can also represent benchmark for numerical solutions of more complicated forms of heat transport
equations in limiting cases, studied above.

References

[1] Dale U. Von Rosenberg, Methods for the Numerical Solution of Partial Differential Equations, vol. 16, Society of Petroleum Engineers, 1969.
[2] Gordon D. Smith, Numerical Solution of Partial Differential Equations: Finite Difference Methods, Oxford University Press, 1985.
[3] U.K.N.G. Ghia, Kirti N. Ghia, C.T. Shin, High-resolutions for incompressible flow using the Navier-Stokes equations and a multigrid method, J. Comput.
Phys. 48 N3 (1982) 387–411.
[4] William F. Ames, Numerical Methods for Partial Differential Equations, Academic Press, 2014.
[5] Claes Johnson, Numerical Solution of Partial Differential Equations by the Finite Element Method, Courier Corporation, 2012.
[6] Brice Carnahan, Herbert A. Luther, James O. Wilkes, Applied Numerical Methods, John Wiley & Sons, 1969.
[7] A. Appèl, J. Kampé de Fériet, Fonctions Hypergéométriques et Hypersphériques; Polynômes d’Hermite, Gauthier-Villars, Paris, 1926.
[8] D.T. Haimo, C. Markett, A representation theory for solutions of a higher-order heat equation. I, J. Math. Anal. Appl. 168 (1992) 89.
[9] D.T. Haimo, C. Markett, A representation theory for solutions of a higher-order heat equation. II, J. Math. Anal. Appl. 168 (1992) 289.
[10] G. Dattoli, Generalized polynomials, operational identities and their applications, J. Comput. Appl. Math. 118 (20 0 0) 111.
[11] G. Dattoli, H.M. Srivastava, K. Zhukovsky, Оrthogonality properties of the Hermite and related polynomials, J. Comput. Appl. Math. 182 (2005) 165.
[12] G. Dattoli, V.V. Mikhailin, K.V. Zhukovsky, Influence of a constant magnetic field on the radiation of a planar undulator, Moscow Univ. Phys. Bull. 64
(2009) 507–512 c/c of Vestnik Moskovskogo Universiteta Seriya 3 Fizika Astronomiya. 5 (2009) 33–38.
[13] K. Zhukovsky, Analytical account for a planar undulator performance in a constant magnetic field, J. Electromagnet. Wave. 28 N15 (2014) 1869–1887,
doi:10.1080/09205071.2014.945664.
[14] K. Zhukovsky, High harmonic generation in the undulators for free electron lasers, Opt. Comm. 353 (2015) 35–41, doi:10.1016/j.optcom.2015.04.079.
[15] K. Zhukovsky, Generation by ultrarelativistic electrons in a planar undulator and the emission-line broadening, J. Electromagnet. Wave 29 N1 (2015)
132–142, doi:10.1080/09205071.2014.985854.
436 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437

[16] G. Dattoli, H.M. Srivastava, K.V. Zhukovsky, Operational methods and differential equations with applications to initial-value problems, Appl. Math.
Comput. 184 (2007) 979.
[17] G. Dattoli, V.V. Mikhailin, K. Zhukovsky, Undulator radiation in a periodic magnetic field with a constant component, J. Appl. Phys. 104 (2008)
124507–124507-8.
[18] K. Zhukovsky, Solution of some types of differential equations: operational calculus and inverse differential operators, Sci. World J. (2014) 8 (2014),
article ID 454865.
[19] G. Dattoli, V. Mikhailin, P.-L. Ottaviani, K. Zhukovsky, Two-frequency undulator and harmonic generation by an ultrarelativistic electron, J. Appl. Phys.
10 0 (20 06) 084507–084507-12.
[20] K.V. Zhukovsky, Harmonic radiation in a double-frequency undulator with account for broadening in Moscow, Univ. Phys. Bull. 70 N4 (2015) 232–239
c/c Vestnik Moskovskogo Universiteta Seriya 3 Fizika Astronomiya. 4. (2015), 18–25, doi:10.3103/S0027134915040177.
[21] K. Zhukovsky, High harmonic generation in undulators for FEL, Nucl. Instrum. Methods B (2015). http://dx.doi.org/10.1016/j.nimb.2015.10.041.
[22] K. Zhukovsky, Inhomogeneous and homogeneous losses and magnetic field effect in planar undulator radiation, Prog. Electromagnet. Res. B 59 (2014)
245–256.
[23] A. Erdélyi, W. Magnus, F. Oberhettinger, F.G. Tricomi, Higher Transcendental Functions, vol. II, McGraw-Hill Book Company, New York, Toronto and
London, 1953.
[24] G. Dattoli, H.M. Srivastava, K. Zhukovsky, A new family of integral transforms and their applications, Integral Transform. Spec. Funct. 17 N1 (2006) 31.
[25] K.V. Zhukovsky, G. Dattoli, Evolution of non-spreading Airy wavepackets in time dependent linear potentials, Appl. Math. Comput. 217 (2011)
7966–7974.
[26] D.S. Chandrasekharaiah, Thermoelasticity with second sound: a review, Appl. Mech. Rev. 39 N3 (1986) 355–376.
[27] D.S. Chandrasekharaiah, Hyperbolic thermoelasticity: a review of recent literature, Appl. Mech. Rev. 510 N12 (1998) 705–729.
[28] D.Y. Tzou, Macro-to Microscale Heat Transfer, Taylor & Francis, Washington, DC, 1997.
[29] D.D. Joseph, L. Preziosi, Heat waves, Rev. Mod. Phys. 61 (1989) 41.
[30] M. Massoudi, M.M. Mehrabadi, Implicit constitutive relations in thermoelectricity, Int. J. Nonlinear Mech. 46 N1 (2011) 286–290.
[31] M. Massoudi, On the heat flux vector for flowing granular materials—part 1: effective vector conductivity and background, Math. Methods Appl. Sci.
29 N13 (2006) 1585–1598.
[32] M. Massoudi, On the heat flux vector for flowing granular materials—part 2: derivation and special cases, Math. Methods Appl. Sci. 29 N13 (2006)
1599–1613.
[33] J.P.J. Fourier, The Analytical Theory of Heat, Cambridge University Press, London, 1878.
[34] L. Onsager, Reciprocal relations in irreversible processes, Phys. Rev. 37 (1931) 119.
[35] C. Cattaneo, Sur une forme de l’equation de la chaleur eliminant le paradoxe d’une propagation instantanee, C. R. Acad. Sci. Paris 247 (1958) 431–433.
[36] V. Peshkov, Second sound in Helium II, J. Phys. 8 (1944) 381 (Moscow).
[37] C.C. Ackerman, W.C. Overton, Second sound in solid helium-3, Phys. Rev. Lett. 22 N15 (1969) 764.
[38] C.C. Ackerman, R.A. Guyer, Temperature pulses in dielectric solids, Ann. Phys. 50 N1 (1968) 128–185.
[39] T.F. McNelly, S.J. Rogers, D.J. Channin, R. Rollefson, W.M. Goubau, G.E. Schmidt, J.A. Krumhansl, R.O. Pohl, Heat pulses in NaF: onset of second sound,
Phys. Rev. Lett. 24 N3 (1970) 100.
[40] V. Narayanamurti, R.D. Dynes, Observation of second sound in bismuth, Phys. Rev. Lett. 26 (1972) 1461–1465.
[41] J. Baringhaus, M. Ruan, F. Edler, et al., Exceptional ballistic transport in epitaxial graphene nanoribbons, Nature 506 (2014) 349–354, doi:10.1038/
nature12952.
[42] A.I. Hochbaum, R. Chen, R.D. Delgado, W. Liang, E.C. Garnett, M. Najarian, A. Majumdar, P. Yang, Enhanced thermoelectric performance of rough silicon
nanowires, Nature (London) 451 (2008) 163.
[43] A.I. Boukai, Y. Bunimovich, J. Tahir-Kheli, J.-K. Yu, W.A. Goddard, J.R. Heath, Silicon nanowires as efficient thermoelectric materials, Nature (London)
451 (2008) 168.
[44] C. Chiritescu, D.G. Cahill, N. Nguyen, D. Johnson, A. Bodapati, P. Keblinski, P. Zschack, Ultralow thermal conductivity in disordered, layered WSe2
crystals, Science 315 (2007) 351.
[45] M. Maldovan, Transition between ballistic and diffusive heat transport regimes in silicon materials, Appl. Phys. Lett. 101 (2012) 113110, doi:10.1063/1.
4752234.
[46] H.B.G. Casimir, Note on the conduction of heat in crystals, Physica 5 (1938) 495.
[47] R.A. Guyer, J.A. Krumhansl, Thermal conductivity, second sound and phonon hydrodynamic phenomena in non-metallic crystals, Phys. Rev. 148 (1966)
778–788.
[48] Tzu-Kan Hsiao, Hsu-Kai Chang, Sz-Chian Liou, Ming-Wen Chu, Si-Chen Lee, Chih-Wei Chang, Observation of room-temperature ballistic thermal con-
duction persisting over 8.3 mm in SiGe nanowires, Nat. Nanotechnol. 8 (2013) 534–538.
[49] W. Kaminski, Hyperbolic heat conduction equations for materials with a nonhomogeneous inner structure, J. Heat Transfer 112 (1990) 555–560.
[50] K. Mitra, S. Kumar, A. Vedavarz, M.K. Moallemi, Experimental evidence of hyperbolic heat conduction in processed meat, J. Heat Transfer 117 (1995)
568–573.
[51] H. Herwig, K. Beckert, Fourier versus non-Fourier heat conduction in materials with a nonhomogeneous inner structure, J. Heat Transfer 122 (2) (20 0 0)
363–365.
[52] W. Roetzel, N. Putra, S.K. Das, Experiment and analysis for non-Fourier conduction in materials with non-homogeneous inner structure, Int. J. Thermal
Sci. 42 (6) (2003) 541–552.
[53] E.P. Scott, M. Tilahun, B. Vick, The question of thermal waves in heterogeneous and biological materials, J. Biomech. Eng. 131 (2009) 074518.
[54] S. Both et al., Deviation from the Fourier law in room-temperature heat pulse experiments. arXiv:1506.05764v1.
[55] A.A. Sokolov, I.M. Ternov, V.Ch. Zhukovskij, A.V. Borisov, Gauge Fields (Kalibrovochnye Polya), Moscow State University, Moscow, 1986, p. 264. (in Rus-
sian).
[56] W.J. Parker, R.J. Jenkins, C.P. Butler, G.L. Abbott, Flash method of determining thermal diffusivity, heat capacity, and thermal conductivity, J. Appl. Phys.
32 (9) (1961) 1679.
[57] K.V. Zhukovsky, Operational solution of differential equations with derivatives of non-integer order, Black–Scholes type and heat conduction, Moscow
Univ. Phys. Bull. 71 N3 (2016) 237–244.
[58] K.B. Wolf, Integral Transforms in Science and Engineering, Plenum Press, New York, 1979.
[59] C. Bai, A.S. Lavine, On hyperbolic heat conduction and the second law of thermodynamics, J. Heat Transfer 117 (1995) 256–263.
[60] J.M. Porrà, J. Masoliver, G.H. Weiss, When the telegrapher’s equation furnishes a better approximation to the transport equation than the diffusion
approximation, Phys. Rev. E 55 (1997) 7771–7774.
[61] C. Körner, H.W. Bergmann, The physical defects of the hyperbolic heat conduction equation, Appl. Phys. A 67 (1998) 397–401.
[62] T.J. Bright, Z.M. Zhang, Common misperceptions of the hyperbolic heat equation, J. Thermophys. Heat Transfer 23 (2009) 601–607.
[63] J. Shiomi, S. Maruyama, Non-Fourier heat conduction in a single-walled carbon nanotube: Classical molecular dynamics simulations, Phys. Rev. B 73
(2006) 205420 [7 pp.].
[64] Z.M. Zhang, Nano/Microscale Heat Transfer, McGraw-Hill, New York, 2007.
[65] G. Lebon, H. Machrafi, M. Gremela, Ch. Dubois, An extended thermodynamic model of transient heat conduction at sub-continuum scales, Proc. R. Soc.
A 467 (2011) 3241–3256.
[66] K. Zhukovsky, A method of inverse differential operators using ortogonal polynomials and special functions for solving some types of differential
equations and physical problems, Moscow Univ. Phys. Bull. 70 (2) (2015) 93–100.
 K.V. Zhukovsky, H.M. Srivastava / Applied Mathematics and Computation 293 (2017) 423–437 437

[67] K.V. Zhukovsky, Exact solution of Guyer-Krumhansl type heat equation by operational method, Int. J. Heat Mass Transfer. 96 (2016) 132–144, doi:10.
1016/j.ijheatmasstransfer.2016.01.005.
[68] K.V. Zhukovsky, Operational method of solution of linear non-integer ordinary and partial differential equations, Springer Plus 5 (2016) 119 Volume 5,
Issue 1, pp. 1–25, doi:10.1186/s40064- 016- 1734- 3.
[69] K.V. Zhukovsky, Violation of the maximum principle and negative solutions with pulse propagation in Guyer–Krumhansl model, Int. J. Heat Mass
Transfer 98 (2016) 523–529, doi:10.1016/j.ijheatmasstransfer.2016.03.021.
[70] H.M. Srivastava, K.V. Zhukovsky, Solutions of some types of differential equations and of their associated physical problems by means of inverse
differential operators, in: Th.M. Rassias, V. Gupta (Eds.), Mathematical Analysis, Approximation Theory and Their Applications, Springer Optimization
and Its Applications Series, vol. 111, 2016, pp. 573–629.