Geometric and Physical Interpretation

of Fractional Integration
and Fractional Differentiation
arXiv:math/0110241v1 [math.CA] 22 Oct 2001

Igor Podlubny
Department of Informatics and Control Engineering
BERG Faculty, Technical University of Kosice
B. Nemcovej 3, 04200 Kosice, Slovak Republic
e-mail: [email protected]

February 1, 2008

Abstract
A solution to the more than 300-years old problem of geometric and
physical interpretation of fractional integration and differentiation (i.e.,
integration and differentiation of an arbitrary real order) is suggested
for the Riemann-Liouville fractional integration and differentiation, the
Caputo fractional differentiation, the Riesz potential, and the Feller
potential. It is also generalized for giving a new geometric and physical
interpretation of more general convolution integrals of the Volterra
type. Besides this, a new physical interpretation is suggested for the
Stieltjes integral.

Keywords: fractional derivative, fractional integral, fractional calculus,

geometric interpretation, physical interpretation.

MSC: 26A33 (main), 26A42, 83C99, 44A35, 45D05

1 Introduction
It is generally known that integer-order derivatives and integrals have clear
physical and geometric interpretations, which significantly simplify their use
for solving applied problems in various fields of science.
However, in case of fractional-order integration and differentiation, which
represent a rapidly growing field both in theory and in applications to real-
world problems, it is not so. Since the appearance of the idea of differentia-
tion and integration of arbitrary (not necessary integer) order there was not
any acceptable geometric and physical interpretation of these operations
for more than 300 years. The lack of these interpretations has been ac-
knowledged at the first international conference on the fractional calculus in

1
2 Report TUKE–10–2001 I. Podlubny

New Haven (USA) in 1974 by including it in the list of open problems [21].
The question was unanswered, and therefore repeated at the subsequent
conferences at the University of Strathclyde (UK) in 1984 [15] and at the
Nihon University (Tokyo, Japan) in 1989 [19]. The round-table discussion
[13, 10, 14] at the conference on transform methods and special functions
in Varna (1996) showed that the problem was still unsolved, and since that
time the situation, in fact, still did not change.
Fractional integration and fractional differentiation are generalisations
of notions of integer-order integration and differentiation, and include n-th
derivatives and n-folded integrals (n denotes an integer number) as partic-
ular cases. Because of this, it would be ideal to have such physical and
geometric interpretations of fractional-order operators, which will provide
also a link to known classical interpretations of integer-order differentiation
and integration.
Since the need for the aforementioned geometric and physical interpre-
tations is generally recognised, several authors attempted to provide them.
Probably due mostly to linguistical reasons, much effort have been devoted
to trying to relate fractional integrals and derivatives, on one side, and frac-
tal geometry, on the other [18, 27, 9, 16, and others]. However, it has been
clearly shown by R. Rutman [22, 23] that this approach is inconsistent.
Besides those “fractal-oriented” attempts, some considerations regarding
interpretation of fractional integration and fractional differentiation were
presented in [16]. However, those considerations are, in fact, only a small
collection of selected examples of applications of fractional calculus, in which
hereditary effects and self-similarity are typical for the objects modelled with
the help of fractional calculus. Although each particular problem, to which
fractional derivatives or/and fractional integrals have been applied, can be
considered as a certain illustration of their meaning, the paper [16] cannot
be considered as a definite answer to the posed question.
A different approach to geometric interpretation of fractional integration
and fractional differentiation, based on the idea of the contact of α-th order,
has been suggested by F. Ben Adda [1, 2]. However, it is difficult to speak
about an acceptable geometric interpretation if one cannot see any picture
there.
Obviously, there is still a lack of geometric and physical interpretation
of fractional integration and differentiation, which is comparable with the
simple interpretations of their integer-order counterparts.
In this paper we present a new approach to solution of this challenging
old problem.
We start with introducing a simple and really geometric interpretation
of several types of fractional-order integration: the left-sided and the right-
sided Riemann–Liouville fractional integration, the Riesz potential, and the
Feller potential.
Based on this, a physical interpretation of the Riemann–Liouville frac-
Geometric and Physical Interpretation of Fractional Integration and Differentiation 3

tional integration is proposed in terms of inhomogeneous and changing (non-

static, dynamic) time scale. Moreover, on this way we give a new physical
interpretation of the Stieltjes integral. We also try to persuade the readers
that the suggested physical interpretation of fractional integration is in line
with the current views on space–time in physics. We also suggest physical
interpretation for the Riemann-Liouville fractional differentiation and for
the Caputo fractional differentiation. Finally, we show that the suggested
approach to geometric interpretation of fractional integration can be used
for providing a new geometric and physical interpretation for convolution
integrals of the Volterra type.

2 Geometric interpretation of fractional

integration: Shadows on the walls
In this section we first give a geometric interpretation of left-sided and right-
sided Riemann–Liouville fractional integrals, and then consider the Riesz
potential.

2.1 Left-sided Riemann–Liouville fractional integral

Let us consider the left-sided Riemann–Liouville fractional integral [20, 24]
of order α,
Zt
α 1
0 It f (t) = f (τ )(t − τ )α−1 dτ, (1)
Γ(α)
0

and write it in the form

Zt
α
0 It f (t) = f (τ )dgt (τ ), (2)
0

1 n o
gt (τ ) = tα − (t − τ )α . (3)
Γ(α + 1)
The function gt (τ ) has an interesting scaling property. Indeed, if we take
t1 = kt and τ1 = kτ , then

gt1 (τ1 ) = gkt (kτ ) = kα gt (τ ). (4)

Now let us consider the integral (2) for a fixed t. Then it becomes simply
a Stieltjes integral, and we can utilize G. L. Bullock’s idea [3].
Let us take the axes τ , g, and f . In the plane (τ, g) we plot the function
gt (τ ) for 0 ≤ τ ≤ t. Along the obtained curve we “build a fence” of the
varying height f (τ ), so the top edge of the “fence” is a three-dimensional
line (τ, gt (τ ), f (τ )), 0 ≤ τ ≤ t.
4 Report TUKE–10–2001 I. Podlubny

This “fence” can be projected onto two surfaces (see Fig. 1):

• the area of the projection of this “fence” onto the plane (τ, f ) corre-
sponds to the value of the integral
Zt
1
0 It (t) = f (τ )dτ ; (5)
0

• the area of the projection of the same “fence” onto the plane (g, f )
corresponds to the value of the integral (2), or, what is the same, to
the value of the fractional integral (1).

In other words, our “fence” throws two shadows on two walls. The first
of them, that on the wall (τ, f ), is the well-known “area under the curve
f (τ )”, which is a standard geometric interpretation of the integral (5). The
“shadow” on the wall (g, f ) is a geometric interpretation of the fractional
integral (1) for a fixed t.
Obviously, for gt (τ ) = τ both “shadows” are equal. This shows that
classical definite integration is a particular case of the left-sided Riemann–
Liouville fractional integration even from the geometric point of view.
What happens when t is changing (namely growing)? As t changes, the
“fence” changes simultaneously. Its length and, in a certain sense, its shape
changes. For illustration, see Fig. 2. If we follow the change of the “shadow”
on the wall (g, f ), which is changing simultaneously with the “fence” (see
Fig.3), then we have a dynamical geometric interpretation of the fractional
integral (1) as a function of t.

2.2 Right-sided Riemann–Liouville fractional integral

Let us consider the right-sided Riemann–Liouville fractional integral [20, 24],

Zb
α 1
t I0 f (t) = f (τ )(τ − t)α−1 dτ, (6)
Γ(α)
t

and write it in the form

Zb
α
t I0 f (t) = f (τ )dht (τ ), (7)
t

1 n o
ht (τ ) = tα + (τ − t)α . (8)
Γ(α + 1)
Then we can provide a geometric interpretation similar to the geometric
interpretation of the left-sided Riemann–Liouville fractional integral. How-
ever, in this case there is no any fixed point in the “fence” base – the end,
Geometric and Physical Interpretation of Fractional Integration and Differentiation 5

10

6
F(t)

0
0

2
0
4
2
6 4
6
8
8
10 10
Gt(τ) t, τ

Figure 1: The “fence” and its shadows: 0 It1 f (t) and 0 Itα f (t),
for α = 0.75, f (t) = t + 0.5 sin(t)

3 0
4 1
2
5 3
6 4
5
7
6
8 7
8
9
G (τ)
t 9 t, τ
10 10

Figure 2: The process of change of the fence basis shape

for 0 Itα f (t), α = 0.75.
6 Report TUKE–10–2001 I. Podlubny

10

F(t)
5

0
0 1 2 3 4 5 6 7 8 9 10
G (τ)
t

Figure 3: Snapshots of the changing “shadow” of changing

“fence” for 0 Itα f (t), α = 0.75, f (t) = t + 0.5 sin(t),
with the time interval ∆t = 0.5 between the snap-
shops.

corresponding to τ = b, moves along the line τ = b in the plane (τ, g) when

the “fence” changes its shape. This movement can be observed in Fig. 4.
(In the case of the left-sided integral, the left end, corresponding to τ = 0,
is fixed and does not move.)
All other parts of the geometric interpretation remain the same: the
“fence” changes its shape as t changes from 0 to b, and the changing shad-
ows of this “fence” on the walls (g, f ) and (τ, f ) represent correspondingly
the right-sided Riemann–Liouville fractional integral (6) and the classical
integral with the moving lower limit:

Zb
1
t Ib (t) = f (τ )dτ ; (9)
t

Obviously, for gt (τ ) = τ both “shadows” are equal. Therefore, we see

that not only the left-sided, but also the right-sided Riemann-Liouville frac-
tional integration includes the classical definite integration as a particular
case even from the geometrical point of view.
Geometric and Physical Interpretation of Fractional Integration and Differentiation 7

3 0
4 1
2
5 3
6 4
5
7
6
8 7
8
9
Gt(τ) 9 t, τ
10 10

Figure 4: The process of change of the fence basis shape

α f (t), α = 0.75.
for t I10

2.3 Riesz potential

The Riesz potential [20, 24]

Zb
α 1
0 Rb f (t) = f (τ )|τ − t|α−1 dτ (10)
Γ(α)
0

is the sum of the left-sided and the right-sided Riemann–Liouville fractional

integrals:

Zt Zb
α 1 α−1 1
0 Rb f (t) = f (τ )(t − τ ) dτ + f (τ )(τ − t)α−1 dτ. (11)
Γ(α) Γ(α)
0 t

The Riesz potential (10) can be written in the form

Zb
α
0 Rb f (t) = f (τ )drt (τ ), (12)
0

1 n o
rt (τ ) = tα + sign(τ − t) |τ − t|α . (13)
Γ(α + 1)
The shape of the “fence”, corresponding to the Riesz potential, is de-
scribed by the function rt (τ ). In this case the “fence” consists of the two
parts: one of them (for 0 < τ < t) is the same as in the case of the left-sided
Riemann–Liouville fractional integral, and the second (for t < τ < b) is the
8 Report TUKE–10–2001 I. Podlubny

3 0
4 1
2
5 3
6 4
5
7
6
8 7
8
9
Gt(τ) 9 t, τ
10 10

Figure 5: The process of change of the fence basis shape

α f (t), α = 0.75.
for the Riesz potential 0 R10

same as for the right-sided Riemann–Liouville integral, as shown in Fig. 5.

Both parts are joined smothly at the inflection point τ = t.
The shape of the “fence”, corresponding to the Riesz potential, is shown
in some of its intermediate position by the bold line in Fig. 5. Obviously,
Fig. 5 can be obtained by laying Fig. 4 over Fig. 2, which is a geometric
interpretation of the relationship (11).
The shadow of this “fence” on the wall (g, f ) represents the Riesz po-
tential (10), while the shadow on the wall (τ, f ) corresponds to the classical
integral
Zb
I(t) = f (τ )dτ. (14)
0

For α = 1 both “shadows” are equal. This shows that the classical definite
integral (14) is a particular case of the Riesz fractional potential (10) even
from the geometric point of view. We have already seen this inclusion in the
case of the left-sided and the right-sided Riemann–Liouville fractional inte-
gration. This demonstrates the strength of the suggested geometric inter-
pretation of these three types of generalization of the notion of integration.

2.4 Feller potential

The Feller potential operator Φα f (t) is, similarly to the Riesz potential, also
a linear combination of the left- and right-sided Riemann–Liouville fractional
Geometric and Physical Interpretation of Fractional Integration and Differentiation 9

integrals, but with general constant coefficients c, d [24, Chap. 3]:

Φα f (t) = c a Itα f (t) + d t Ibα f (t). (15)

The geometric interpretation of the Feller potential can be easily ob-

tained by properly scaling and then superimposing Fig. 4 and Fig. 2. The
“fence” obtained in this way is, in general, discontinuous at τ = t. Its
shadow on the wall (τ, f ) is equal to the classical definite integral (14). The
shadow on the wall (g, f ) consists, in general, of the two areas, which may
overlap depending on the values of the coefficients c and d.

3 Two kinds of time – I

The geometric interpretation of fractional integration, given in the previous
sections, is substantially based on adding the third dimension (for gt (τ ))
to the classical pair τ , f (τ ). If we consider τ as time, then g(τ ) can be
interpreted just as a “deformed” time scale. What could be the meaning of
having – and using – two time axes? To answer this question, let us recall
some facts of the history of the development of the notion of time.
That were contributions of Barrows and Newton to the development of
mathematics and physics in the XVII century which led to the appearance of
the “mathematical time”, which is postulated to “flow equably” and which
is usually depicted as a semi-infinite straight line [26].
Newton himself postulated [17]:

“Absolute, true and mathematical time of itself, and from its own
nature, flows equably without relation to anything external.”

Such a postulate was absolutely necessary for developing Newton’s dif-

ferential calculus and applying it to problems of mechanics [26]:

“The outstanding mathematical achievement associated with the

geometrization of time was, of course, the invention of the cal-
culus of fluxions by Newton.”
“Mathematically, Newton seems to have found support for
his belief in absolute time by the need, in principle, for an ideal
rate-measurer.”

The invention of differential and integral calculus and today’s use of them
is the strongest reason for continuing using homogeneous equably flowing
time.
Time is often depicted using the time axis, and the geometrically equal
intervals of the time axis are considered as corresponding to equal time
intervals (Fig. 6).
10 Report TUKE–10–2001 I. Podlubny

t
0 1 2 3 4 5 6 7

Figure 6: Homogeneous time axes.

t
0 3 4 5 6 7

Figure 7: Time slowing down.

This assumption, however, cannot be neither proved nor rejected by

experiment. Two lengths of geometric intervals can be measured and com-
pared, since they are available for measurement simultaneously, at the same
time (or, more precisely, at the same time and at the same place). Two
time intervals can never be compared, because they are available to us for
measurement (or for observation) only sequentially.
Indeed, how do we measure time intervals? Only by observing some
processes, which we consider as regularly repeated. G. Clemence wrote [7]:

“The measurement of time is essentially a process of counting.

Any recurring phenomenon whatever, the occurences of which
can be counted, is in fact a measure of time.”

Clocks, including atomic clocks, repeat their “ticks”, and we simply

count those ticks, calling them hours, minutes, seconds, milliseconds, etc.
But we are not able to verify if the absolute time which elapsed between,
say, the fifth and the sixth tick (the sixth “second”) is exactly the same as
the time, which elapsed between the sixth and the seventh tick (the seventh
“second”). This possible inhomogeneity of the time scale is illustrated in
Fig. 7.
The fact that time measurement as a process of counting of repeating
discrete events does not really exclude inhomogeneity of time, has been
nicely mentioned by L. Carroll in Alice’s Adventures in Wonderland [6,
Chap. 7]:

“. . . I know I have to beat time when I learn music.”

“Ah! That accounts for it,” said the Hatter. “He [Time]
won’t stand beating. Now, if you only kept on good terms with
him, he’d do almost anything you liked to do with the clock. . . ”

Figures 6 and 7 show those “clock ticks”, which we can register, only
symbolically. One can interpret them as if there exists some absolute, or
cosmic, inhomogeneous time axis, to which we can compare individual ho-
mogeneous time, represented by some “clock ticks”. Our picture of the
Geometric and Physical Interpretation of Fractional Integration and Differentiation 11

Table 1: Recording speed using slowing-down clocks

Person N Recorded values Observer O
individual of velocity [m/s] absolute (cosmic)
“seconds” “seconds”
0 10 0
1 11 1
2 12 3
3 13 7
4 12 15
5 11 31
6 10 63
7 9 127

individual homogeneous time has the form shown in Fig. 6. The cosmic
time may be not necessarily flowing equably, like that shown in Fig. 7.
To illustrate the idea, let us consider the following situation. Suppose
person N has two devices: one is a speedometer, and another one is the
clock, which is slowing down, so the interval between the two subsequent
ticks is double comparing to the interval between the previous ticks (see
Fig. 7). Person N reads the velocity values indicated by the speedometer
at each encountered “second”, without knowing that the clock is, in fact,
slowing down.
Using these two series of data, namely the recorded sequence of values
of speed, and the sequence of the counted “seconds”, person N can estimate
the distance which he has passed.
For simplicity, let us suppose that the first “second” of the time shown by
the clocks is equal to the absolute time “second”. The results of observations
in this hypothetical experiment are given in Table 1.
Person N will compute the distance he has passed as

SN = 10 · 1 + 11 · 1 + 12 · 1 + 13 · 1 + 12 · 1 + 11 · 1 + 10 · 1 = 79.

However, if there would be an independent observer O, knowing about

the slowing-down clock, then such an observer would obtain a notably dif-
ferent result for the distance passed by person N:

SO = 10 · 1 + 11 · 2 + 12 · 4 + 13 · 8 + 12 · 16 + 11 · 32 + 10 · 64 = 1368.

Below we use this idea for giving a new mechanical interpretation of the
Stieltjes integral.
12 Report TUKE–10–2001 I. Podlubny

4 Physical interpretation of
the Stieltjes integral
Imagine a car equipped with two devices for measurements: the speedometer
recording the velocity v(τ ), and the clock which should show the time τ .
The clock, however, shows the time incorrectly; let us suppose that the
relationship between the wrong time τ , which is shown by the clock and
which the driver considers as the correct time, on one hand, and the true
time T , on the other, is described by the function T = g(τ ). This means
that where the driver “measures” the time interval dτ , the real time interval
is given by dT = dg(τ ).
The driver A, who do not know about wrong operation of the clock, will
compute the passed distance as the classical integral:

Zt
SA (t) = v(τ )dτ . (16)
0

However, the observer O knowing about the wrong clock and having the
function g(τ ), which restores the correct values of time from the driver’s
wrong time τ , will compute the really passed distance as

Zt
SO (t) = v(τ )dg(τ ). (17)
0

This example shows that the Stieltjes integral (17) can be interpreted
as the real distance passed by a moving object, for which we have recorded
correct values of speed and incorrect values of time; the relationship between
the wrongly recorded time τ and the correct time T is given by a known
function T = g(τ ).

5 Physical interpretation of fractional integration:

Shadows of the past
Now let us consider the left-sided Riemann–Liouville fractional integral

Zt
α
SO (t) = v(τ )dgt (τ ) = 0 It v(t), (18)
0

where gt (τ ) is given by (3).

The fractional integral SO (t) of the function v(τ ) can be interpreted as
the real distance passed by a moving object, for which we have recorded the
local values of its speed v(τ ) (individual speed) and the local values of its
Geometric and Physical Interpretation of Fractional Integration and Differentiation 13

time τ (individual time); the relationship between the locally recorded time
τ (which is considered as flowing equably) and the cosmic time (which flows
non-equably) is given by a known function gt (τ ).
The function gt (τ ) describes the inhomogeneous time scale, which de-
pends not only on τ , but also on the parameter t representing the last mea-
sured value of the individual time of the moving object. When t changes, the
entire preceding cosmic time interval changes as well. This is in agreement
with the current views in physics. Indeed, B. N. Ivanov [12, p. 33] mentioned
that time intervals depend on gravitational fields. Similarly, S. Hawking [11,
p. 32–33] wrote that:

“. . . time should appear to run slower near a massive body like

the earth.” [. . . ]
“. . . there is no unique absolute time, but instead each in-
dividual has his own personal measure of time that depends on
where he is and how he is moving.”

When a moving body changes its position in space–time, the gravita-

tional field in the entire space–time also changes due to this movement. As
a consequence, the cosmic time interval, which corresponds to the history
of the movement of the moving object, changes. This affects the calculation
(using formula (18)) of the real distance SO (t) passed by such a moving
object.
In other words, the left-sided Riemann–Liouville fractional integral of the
individual speed v(τ ) of a moving object, for which the relationship between
its individual time τ and the cosmic time T at each individual time instance
t is given by the known function T = gt (τ ) described by the equation (3),
represents the real distance SO (t) passed by that object.

6 Physical interpretation of the Riemann-Liouville

fractional derivative
On the other hand, we can use the properties of fractional differentiation and
integration [20, 24] and express v(t) from the equation (18) as a left-sided
Riemann–Liouville fractional derivative of SO (t):
α
v(t) = 0 Dt SO (t) (19)

where 0 Dtα denotes the Riemann–Liouville fractional derivative [20, 24],

which is for 0 < α < 1 defined by

Zt
α 1 d f (τ )dτ
0 Dt f (t) = . (20)
Γ(1 − α) dt (t − τ )α
0
14 Report TUKE–10–2001 I. Podlubny

This shows that the left-sided Riemann–Liouville fractional derivative of

the real distance SO (t) passed by a moving object, for which the relationship
between its individual time τ and the cosmic time T at each individual time
instance t is given by the known function T = gt (τ ) described by equation
(3), is equal to the individual speed v(τ ) of that object.
On the other hand, we can differentiate the relationship (18) with respect
to the cosmic time variable t, which gives the relationship between the ve-
locity vO (t) = SO′ (t) of the movement from the viewpoint of the independent

observer O and the individual velocity v(t):

d α 1−α
vO (t) = 0 I v(t) = 0 Dt v(t), (21)
dt t
Therefore, the (1 − α)-th–order Riemann–Liouville derivative of the in-
dividual velocity v(t) is equal to the velocity vO (t) from the viewpoint of the
independent observer, if the individual time τ and the cosmic time T are re-
lated by the function T = gt (τ ) described by equation (3). For α = 1, when
there is no dynamic deformation of the time scale, both velocities coincide:
vO (t) = v(t).

7 Physical interpretation of the Caputo

fractional derivative
Applying fractional integration of order β = 1 − α to both parts of the
relationship (21) gives:
C α
v(t) = 0 It1−α vO (t) = 0 It1−α SO
′
(t) = 0 Dt SO (t), (22)
where C α
0 Dt denotes the Caputo fractional derivative [4, 5, 20], which is for
0 < α < 1 defined by
Zt
C α 1 f ′ (τ )dτ
0 Dt f (t) = . (23)
Γ(1 − α) (t − τ )α
0

The relationship (22) is similar to (19). Therefore, the Caputo fractional

derivative has the same physical interpretation as the Riemann–Liouville
fractional derivative (see Section 6). This coincidence becomes more obvious,
if we recall [20] that if f (0) = 0, then the Riemann–Liouville derivative
and the Caputo derivative of order α (0 < α < 1), coincide: C α
0 Dt f (t) =
α
0 Dt f (t).

8 Two kinds of time – II

The suggested physical interpretation of fractional integration and fractional
differentiation is based on using two kinds of time: the cosmic time and the
individual time.
Geometric and Physical Interpretation of Fractional Integration and Differentiation 15

As mentioned above, due to the history of the development of mathemat-

ics and physics, we are taught to think about the time, in fact, geometrically.
The real roots of this go even far back to Euclid [26]:

“Euclid considered space as the primary concept of science and

relegated time to poor second.”

The entire integral and differential calculus is based on using mathemat-

ical (homogeneous, equably flowing) time. There is no chance to change
this state, and there is nothing to suggest instead of the classical calculus.
Moreover, there is probably even no need for this. We can just realize that
the classical calculus provides tools for describing the dynamic properties of
the cosmic time, which – according to physicists – is inhomogeneous (flowing
non-equably). Indeed [11, p. 33–34],

“The old idea of an essentually unchanging universe that could

have existed, and could continue to exist, forever was replaced
by the notion of a dynamic, expanding universe that seemed to
have begun a finite time ago. . . ”

Clearly, the expansion of the universe implies that neither spatial scale
nor time scale remains homogeneous; they both are dynamic. For describ-
ing the inhomogeneous time, the ideal homogeneous time scale can be used.
This approach is not new; it has already been used in the theory of relativ-
ity for describing shortening of time intervals. This means that in fact two
time scales are considered simultaneously: the ideal, equably flowing homo-
geneous time, and the cosmic (inhomogeneous) time. The change of scale of
the cosmic time is described using the homogeneous time scale as a reference
scale. In other words, the homogeneous time scale is just an ideal notion,
which is necessary for developing mathematical models describing inhomoge-
neous cosmic time and its change. In this respect we can, without discussing
other views on this subject, recall the remark made by A. Daigneault and
A. Sangalli in their essay [8] about I. E. Segal and his two-time cosmology
(“chronometric cosmology”, or CC) [25] – note that “perhaps! ”:

“According to CC, Einstein’s model is the correct one to under-

stand the universe as a whole (i.e., global space–time), except
that there are two kinds of time: a cosmic or Einstein’s time t,
and a local or Minkowski’s time x0 , which is (perhaps!) the time
measured by existing techniques. [. . . ] Simply put, Einstein’s
cosmic time is the “real” one, whereas Minkowski’s time is only
an approximation of t.”

So, the ideal model of equably flowing homogeneous time can be considered
as a rough approximation of the cosmic time.
16 Report TUKE–10–2001 I. Podlubny

9 Geometric and physical interpretation

of the Volterra convolution integral
It should be mentioned that we can also provide a geometric and physical
interpretation for more general integrals.
The Riemann–Liouville fractional integral is a particular case of convo-
lution integrals of the Volterra type:

Zt
K ∗ f (t) = f (τ )k(t − τ )dτ (24)
0

Assuming that k(t) = K ′ (t), we can write this integral in the form

Zt
K ∗ f (t) = f (τ )dqt (τ ), (25)
0

qt (τ ) = K(t) − K(t − τ ). (26)

The geometric and physical interpretation of the Volterra convolution in-
tegral is then similar to the suggested interpretations for fractional integrals.
The function qt (τ ) determines the changing shape of the “live fence” (in the
case of the geometric interpretation, see Figs. 1 and 2) and the relationship
between the individual time and the cosmic time of a moving object (in the
case of the physical interpretation).

Acknowledgment
This work has been done during the author’s visit to Department of Applied
Mathematics and Theoretical Physics (DAMTP) of the University of Cam-
bridge, UK, in July 2000, supported by the Cambridge Colleges Hospitality
Scheme.
The author will always remember late Professor David G. Crighton, the
former Head of DAMTP and the Master of the Jesus College, who invited
the author and arranged all details of the author’s stay and research at
DAMTP. Without this the present paper could not appear.
The author also expresses his deep gratitude to Professor Stephen Heath,
the President of the Jesus College, and to all members of the College, for
their cordial approach during the whole period of the author’s visit.

References
[1] F. Ben Adda. Geometric interpretation of the fractional derivative.
Journal of Fractional Calculus, vol. 11, May 1997, pp. 21–52.
Geometric and Physical Interpretation of Fractional Integration and Differentiation 17

[2] F. Ben Adda. Interprétation géometrique de la différentiabilité et du

gradient d’ordre réel. C. R. Acad. Sci. Paris, t. 326, Série I, 1998,
pp. 931–934.

[3] G. L. Bullock. A geometric interpretation of the Riemann–Stieltjes

integral. American Mathematical Monthly, vol. 95, no. 5, May 1988,
pp. 448–455.

[4] M. Caputo. Linear model of dissipation whose Q is almost frequency

independent – II. Geophys. J. R. Astr. Soc., vol. 13, 1967, pp. 529–539.

[5] M. Caputo. Elasticità e Dissipazione. Zanichelli, Bologna, 1969.

[6] L. Carroll. Alice’s Adventures in Wonderland. Progress, Moscow, 1979

(or any other edition).

[7] G. Clemence. Time and its measurement. The American Scientist,

vol. 40, 1952, pp. 260–269.

[8] A. Daigneault and A. Sangalli. Einstein’s static universe: An idea whose

time has come back? Notices of the AMS, vol. 48, no. 1, 2001, pp. 9–16.

[9] Fu-Yao Ren, Zu-Guo Yu, Feng Su. Fractional integral associated to
the self-similar set of the generalised self-similar set and its physical
interpretation. Phys. Lett. A, vol. 219, 1996, pp. 59–68.

[10] R. Gorenflo. Afterthoughts on interpretation of fractional derivatives

and integrals. In: P. Rusev, I. Dimovski, V. Kiryakova (eds.), Transform
Methods and Special Functions, Varna’96, Institute of Mathematics
and Informatics, Bulgarian Academy of Sciences, Sofia, 1998, ISBN
954-8986-05-1, pp. 589–591.

[11] S. W. Hawking. A Brief History of Time. Bantam Press, London, 1988.

[12] B. N. Ivanov. Printsipy sovremennoi fiziki (Principles of Modern

Physics). Nauka, Moscow, 1973 (in Russian).

[13] V. Kiryakova. A long standing conjecture failed? In: P. Rusev, I. Di-

movski, V. Kiryakova (eds.), Transform Methods and Special Func-
tions, Varna’96, Institute of Mathematics and Informatics, Bulgarian
Academy of Sciences, Sofia, 1998, ISBN 954-8986-05-1, pp. 579–588.

[14] F. Mainardi. Considerations on fractional calculus: interpretations and

applications. In: P. Rusev, I. Dimovski, V. Kiryakova (eds.), Transform
Methods and Special Functions, Varna’96, Institute of Mathematics
and Informatics, Bulgarian Academy of Sciences, Sofia, 1998, ISBN
954-8986-05-1, pp. 594–597.
18 Report TUKE–10–2001 I. Podlubny

[15] A. McBride, G. Roach (eds). Fractional Calculus. Research Notes in

Mathematics, vol. 138, Pitman, Boston–London–Melbourne, 1985.

[16] M. Monsrefi-Torbati and J.K. Hammond. Physical and geometrical in-

terpretation of fractional operators. J. Franklin Inst., vol. 335B, no. 6,
1998, pp. 1077–1086.

[17] I. Newton. Mathematical Principles (trans. A. Motte, ed. F. Cajori),

Berkeley, 1934.

[18] R. R. Nigmatullin. A fractional integral and its physical interpretation.

Theoret. and Math. Phys., vol. 90, no. 3, 1992, pp. 242–251.

[19] K. Nishimoto (ed.). Fractional Calculus and Its Applications. Nihon

University, Koriyama, 1990.

[20] I. Podlubny. Fractional Differential Equations. Academic Press, San

Diego, 1999.

[21] B. Ross (ed.). Fractional Calculus and Its Applications, Lecture Notes
in Mathematics, vol. 457, Springer-Verlag, New York, 1975.

[22] R. S. Rutman. On the paper by R.R. Nigmatullin “A fractional integral

and its physical interpretation”. Theoret. and Math. Phys., vol. 100,
no. 3, 1994, pp. 1154–1156.

[23] R. S. Rutman. On physical interpretations of fractional integration

and differentiation. Theoret. and Math. Phys., vol. 105, no. 3, 1995,
pp. 1509–1519.

[24] S. G. Samko, A. A. Kilbas, O. I. Marichev. Integraly i proizvodnye

drobnogo poryadka i nekotorye ich prilozheniya (Integrals and Deriva-
tives of the Fractional Order and Some of Their Applications). Nauka i
Tekhnika, Minsk, 1987 (English translation: S.G. Samko, A.A. Kilbas,
and O.I. Marichev, Fractional Integrals and Derivatives, Gordon and
Breach, Amsterdam, 1993).

[25] I. E. Segal. Mathematical Cosmology and Extragalactic Astronomy,

Academic Press, New York, 1976.

[26] G. J. Whitrow. The Natural Philosophy of Time. Thomas Nelson and

Sons Ltd., London and Edinburgh, 1961.

[27] Zu-Guo Yu, Fu-Yao Ren and Ji Zhou. Fractional integral associated to
generalized cookie-cutter set and its physical interpretation. J. Phys.A:
Math. Gen., vol. 30, 1997, pp. 5569–5577.