From "Fractional Calculus and its Applications", Springer Lecture Notes in Mathematics, volume 57, 1975, pp.1-36.

A BRIEF HISTORY AND EXPOSITION OF THE FUNDAMENTAL THEORY OF FRACTIONAL CALCULUS

BERTRAM ROSS Abstract: audience This opening lecture is intended to serve as a propaedeutic engineers and educators. idea

for the papers to be presented at this conference whose nonhomogeneous includes scientists, mathematicians, lecture, This expository and developmental from its birth in intellectual mental structure of fractional for the use of fractional cated. ference. a case study of mathematiThe fundais indi-

cal growth, surveys the origin and development of a mathematical curiosity to applications. calculus is outlined.

The possibilities

calculus

in applicab]e mathematics

The lecture closes with a statement of the purpose of the con-

Fractional tension of meaning. of real numbers

calculus has its origin in the question of the exA well known example is the extension of meaning and another is the extension of of the to factorials of complex numbers.

to complex numbers,

meaning of factorials of integers extension of meaning is: order dny/dx n irrational, fractional

In generalized integration and differentiation the question be extended to have meaning where or complex? Perhaps, n n

Can the meaning of derivatives of integral is any number---

Leibnitz play with symbols possibility that L'Hospital. n

invented the above notation. that prompted L'Hospital be a fraction. [i] in 1695 replied, "What if

it was naive

to ask Leibnitz about the be ?", asked one day useand stated "It will lead to a paradox."

Leibnitz

But he added prophetically, Wallis's result.

"From this apparent paradox, ~/2, used the notation

ful consequences will be drawn." infinite product for that differential

In 1697, Leibnitz, referring to d2y

calculus might have been used to achieve the same

In 1819 the first mention of a derivative of arbitrary order appears in a text. The French mathematician, S. F. Lacroix [2],

published

a 700 page text on differential than two pages

and integral

calculus

in

which he devoted less

to this topic.

Starting with n a positive integer,

y = xn~ he found the dmy _ n! dx m (n-m) ! mth n -m

x

derivative

to be

Using Legendre's and by replacing Lacroix obtained

symbol m by

F 1/2

which denotes and n

the generalized

factorial, a,

by any positive formalists

real number

in the manner typical

of the classical

of this period,

the formula d2y = F(a+l) dx r (a+) of arbitrary for y = x 2~ /-~ F(2) = i. This result is the same first order 1/2 of the funcxa-

which expresses tion x a.

the derivative

He gives the example d dx

and gets

(x)

because tional raised

F(3/2)

= F()

= /-# and

yielded by the present derivative. the question

day Riemann-Liouville 279 years

definition devoted

of a fracto this topic~

It has taken

since L'Hospital

for a text to appear solely

[3].
Euler and Fourier made mention order but making they gave no applications belongs calculus the fractional arises sometimes the first application of derivatives of arbitrary of [4] in 1823. problem. such the bead
time

or examples.

So the honor

to Niels Henrik Abel

Abel applied This problem,

in the solution of an integral of the tautochrone is that of findplane to the

equation which

in the formulation wire

called the isochrone problem,

ing the s~hape of a frictionless lowest point is placed. of slide. Abel's attracted attempt of the wire

lying in a vertical regardless

that the time of slide of a bead placed on the wire slides in the s a m e
time

of where

The brachistochrone

problem

deals with the shortest

solution was so elegant of Liouville definition

that it is my guess

it He

the attention

[S] who made the first major of a fractional derivative.

to give a logical

published

three long memoirs Liouville's

in 1832 and several more through is the known result

1855.

starting point

for deriva-

tives of integral

order Dme ax = ame ax

which he extended

in a natural way to derivatives DYe ax = aVe ax

of arbitrary

order

He expanded

the function

f(x)

in the series

(1)

f(x)

9o

cn e

anX ,
order f(x) to be

n=O

and assumed (2)

the derivative DVf(x)

of arbitrary
co

=
n=O

Cn a v e anx

This formula obvious the series

is known

as Liouville's that v

[6] first definition to values

and has the such that

disadvantage converges. Liouville's

must be r e s t r i c t e d

second m e t h o d was applied He considered

to explicit

functions

of the form
(3)

x "a, a > O.
I =

the integral

f
1
r(a)

ua-le-XUdu.

The transformation (4)

xu = t x-a _

gives the result I. after operating on both sides

Then, with the use of (I) he obtained, of (4) with (5) D v, the result DVx -a = (-l)Vr(a+v)
r(a)

x -a-v

[7]

Liouville problems

was successful theory. [8].

in applying these definitions were too narrow is restricted to to

to

in potential of v

"These concepts

last," said Emil Post certain values class of functions.

The first definition

and the second method

is not suitable

to a wide

Between 1835 and 1850 there was a c o n t r o v e r s y w h i c h c e n t e r e d on two definitions favored Lacroix's mathematicians [I0] of a fractional derivative. George Peacock [9] Other

g e n e r a l i z a t i o n of a case of integral order. definition. Augustus

favored Liouville's

De M o r g a n ' s

judgement p r o v e d to be accurate when he stated that the two may very p o s s i b l y be parts of a more general system. In 1850

versions

W i l l i a m Center versions

[ii] observed that the d i s c r e p a n c y b e t w e e n the two derivative focused on the fractional deriva-

of a fractional

tive of a constant. fractional while derivative

A c c o r d i n g to the P e a c o c k - L a c r o i x v e r s i o n the of a constant yields formula a result other than zero

according to Liouville's

(5) the fractional derivative

of a constant equals

zero because

r(o) = ~. century is now at

The state of affairs cleared up.

in the m i d - n i n e t e e n t h [12] states,

Harold Thayer Davis

"The m a t h e m a t i c i a n s

that time were aiming for a p l a u s i b l e definition of g e n e r a l i z e d differentiation but, in fairness to them, one should note they lacked of their definition in the com-

the tools to examine the consequences plex plane." Riemann posthumously [13]

in 1847 while a student wrote a paper p u b l i s h e d a definition of a fractional operation.

in which he gives

It is my guess that Riemann was i n f l u e n c e d by one of Liouville's memoirs in which Liouville wrote, "The ordinary differential equation

dny = O dx n has the complementary solution + "'" + Cn-I x n - 1

Yc = Co + ClX + c 2 x 2 Thus

du

f(x)

dx u should clined have to a corresponding Riemann complementary saw fit to add solution." So, I am i n function to

believe

a complementary

his definition of a fractional (6) D-v f(x) = 1

integration: (x-t)v-lf(t)dt + ,(x).

r (v)

;c

Cayley

[13] remarked in 1880 that Riemann's

complementary

function

is of indeterminate nature.

The development Peacock made several he m i s a p p l i e d which

of mathematical

ideas is not without calculus

error. when

errors

in the topic of fractional did not always

the Principle of the Permanence Liouville made

of Equivalent Forms
apply to the

is stated for algebra and which of a complementary

theory of operators. in his discussion

an error when he failed to note function that the specialization Riemann became hopefunction. different Two reyielded complementary I suggest

of one of the parameters lessly entangled with different sults when Heaviside century, versions

led to an absurdity. derivative Thus, silence

an indeterminate

of a fractional

applied to a constant. p u b l i s h e d his work

that when Oliver

in the last decade

of the nineteenth but also because of

he was met with haughty

and disdain not only because concept of fractional

of the hilarious operators.

jibes he made at mathematicians had in the general

the distrust mathematicians

The subject ness of notation papers that notation follow

of notation

cannot be minimized. various notations

The succinctIn the The

of fractional

calculus

adds to its elegance. are used.

in this text,

I prefer was

invented by Harold T. Davis.

All the informa-

tion can be conveyed by the symbols

cD~ v f(x),
denoting scripts integration c and x integral of arbitrary denote which defines becomes order

O, The subof The adjoinsymbol

along the x-axis. integration.

the limits

(terminals)

of integration

a definite

fractional

ing of these subscripts to avoid ambiguities

a vital part

of the operator

in applications. the mathematical problem of defining formalizing frac-

We now consider tional integration mentioned maticians

and differentiation.

It is clear that the mathebut were try-

so far were not merely

ing to solve a p r o b l e m which itly formulate. f(z), irrational

'

they well understood but did not explicfor every function and every number D v f(z)
c z

Briefly what is w a n t e d is this: wide real class, or complex, a function

z = x + {y, of a sufficiently fractional

v, or

= g(z)
'

D v f(x) = g(x) when z is purely cx to the following criteria: i. able If f(z)

should be assigned subject

is an analytic v f(z) cDz

function of the complex varifunction of v and z.

z, the derivative

is an analytic

2. as o r d i n a r y
is a negative result with as its same along

The operation
differentiation integer, ordinary n-1 say

C X

D v f(x)
v is

must produce
a positive
C

the same result

integer. must If produce must v the vanish

when

v = -n,

then

D- n
X

f(x)
C

n-fold

integration at x = c.

and

D- n
X

f(x)

derivatives

3. changed:

The operation
C X

of order
= f(x)

zero leaves

the function

un-

DO f ( x )

4.

The fractional

operators

must be linear: + b cD-Vx g(x) of arbitrary

cDx v [af(x) + bg(x)]

5.

= a cD-Vx f(x)

The law of exponents cDx u- cD-Vx f(x)

for integration

order holds: = cD-U-Vx f(x) these

fX

A definition of Riemann (7)

which is

fulfills

criteria

named

in honor

and Liouville

cD-Vx f(x)

= irtvjlr ~ jc

(x-t)v-lf(t)dt"

This definition Riemann's we have shown

for integration definition (see the above

of arbitrary

order

is the same as When c = O to is equivalent

definition

but has no complementary and when [6], pp. 176-178). criteria, that will

function. Although

Riemann's

c = -= , (7)

Liouville's to establish

definitions

(7) can be

to fulfill

stated

it might be of interest (7) uniquely. 379. four differ-

a set of criteria is discussed

characterize

This question

later in this text p. (7) can be obtained

The definition ent ways.

in at least

Euler had shown that ~0 x (x-t)bt d dt = F(b+l) F(d+1)_ xb+d+l F(b+d+2) b and d > -I

(s)
For b = 3 and

d = 4, (8) gives
r

the result
x s

(4)

8.7-6 -5 If one were constant to integrate the function x4 four times and take the

of integration

each time to be zero,

the result will be

8-7"6"5
Inquisitive experimentation

x8
lead one to guess that

of this type might

the above two results may be connected by the expression: x4 = i F (4)

oDx 4 or in general (9)

fox
~OX

(x-t) 3 t 4 dt,

oD; n f(x)

?(n)

( x - t ) n-1 f ( t )

dt.

The above is generalized by letting

n = v.

iterated

The same r e s u l t integral

can be o b t a i n e d by c o n s i d e r i n g

the n - f o l d

- 2

Xn - 1

F(x) = ~cx dx 1 ~cxl d x 2 - ' This iterated integral

~c xn

f dXn-1 Jc
as a single is

f(Xn) dXn '

integral by the over an appro-

can be written [14]. 1 ~

method devised by Dirichlet, priate triangular region F(x) =

that is, by integrating The result

rX
r(n) Jc
(x x

- n)

n-I

f(Xn) dxn "

and of inte-

If we denote gration as Dx

the operators

of differentiation ~c x . . . .

D; 1 and

dx,

We may write

F(x) n

= cD; n f(x). with v

Then letting

xn = t at (7).

and generaliz-

ing by replacing

we again arrive

A third approach linear differential

to (7) may be deduced using the theory of Let dn - 1 dxn-I + ... + Pn(X) coefficients Green's Pk' 0 ~ k ~ n and for L. function

equations. dn

.. L = Po(X) dx n be a linear differential are continuous Po(X) > O on I. Let

+ Pl(X)

operator whose H

on some closed finite

interval

I = [a,b]

be the one-sided

Then if in

is any function x C I,

continuous

on

I, and

xo

is any point

I, then for all

g(x)

H(x,~) f(~)d~

o equation Ly = f ( x ) which satis-

is fies

the

solution

of the

nonhomogeneous

the boundary

conditions g ( k ) ( x o) = 0 , 0 ~ k ~ n-1 .

[For further details York (19631; Chapter

see, 3.]

for example,

K. S. Miller,

Linear DifferInc., New

ential Equations in the ReaZ Domain, W. W. Norton and Co.,

The Green's

function

is given explicitly

by

el(X)

qb2 (x) 2(~)

... ..-

~bn(X) (~n(~)

qbl (~)
(-i) n-I

H(x,{) =

Po(~)W(~)

where and W

{k[l ~ k # n}

is a fundamental

set of solutions

of

Ly = O,

is their Wronskian.

Now s u p p o s e L = Dn
Then

dn dx n
set of solutions of

{l

x, x 2 , and

..., x n-l)

is a fundamental

Dny = O

i(~)
w(~) :

2()

--.

Cn(~)

(~)

"'"

. .

cn -i

(n-l) n-2

(n-l) (n-Z) n-3

o
= (n-l) !!

( n - I) !

where n-I (n-l) !! = II k! k=O Thus in this special case

2 X n-i

1 (-i) n-I

. .

cn-i

H(X,)

( n - l ) !!

(n-l) n-2

o
is a p o l y n o m i a l of degree n-i in

o
x

o
with leading

(n-l) !
coefficient

(-1) n-1 _ _ [(_l)n+l(n_2) ( n - l ) !!

But

! !] =

1 (n-l) !

H(x,)

= O , 0 < k < n-2

3xk
Thus (I0) Hence

x=

is a zero of m u l t i p l i c i t y

n-I

and

H(x,)
if xo

1 -I - (n-1) l(x-)n
=

a,

10

rx
(11)
is the unique

g(x)

~ (n-l')!

Ja (x-~)n-I

f(~)d~

solution of the differential

equation

dny - f(x) dx n which write assumes (II) as the initial values g~k~(a)r ~ = O, 0 < k < n-l. We may

~aX
aDxnf(x) = 1 r (n)
n by v

(x_~)n-lf(~)d~

Now,

of course, we replace

(with

Re ~ > O)

in the above

formula~ [15]. The fourth method of arriving tour integration these generalized tegral references Nekrassov in 1890. in the complex plane. operators at definition (7) is by confact that inonly passing P.A.

It is a curious for themselves theory.

and their connection with the Cauchy in securing starting in complex variable

formula have succeeded in standard works Laurent

in 1888 used a contour loop.

at the origin as did A. Krug

in 1884 used a contour that started and ended at -~,

now called a Laurent Cauchy's

integral

formula

is

f(n) (z) = D z f(z)

= 2--~-~-~J ic

n!

f(t) dt. (t-z) n+l

In the previous n to fractional But here To keep

three methods values

of obtaining

(7) the generalization because

of

creates

no difficulties

v! = F(v+l).

i/(t-z) v+l the function

no longer contains single valued, at t = x > O c,

a pole but a branch point. cut be the infinity on to negative Let C

we let the branch

semi-infinite

line starting starts

the real t axis as in the figure below. (or loop) which It-x I < ~ a the point the cut, then goes along

be the open contour

c < x, on the lower edge of to c along

the real axis to A, around the circle sense to B, and then back

in the positive

the upper edge of the cut.

ii

Im(t)

t-plsne

cut

B/~ex

~.C

,Re(t)

of

Generalizing gives

the Cauchy integral formula to arbitrary values

F(v+I) D v f(z) = " 2 ~ where we define

(t-z)

-v-lf(

t) dt

(t_x) -v-I = e(-V-l)in(t-x), and where In(t-x) is real when t-x is a positive real number.

By standard methods of contour to (7) (see [6] pp. 198-202).

integration we are a~ain led

The general validity of definition positive we are concerned with criterion

(7) for

v = n, n induction.

a Here

integer,can be established by mathematical 2 which stipulates

that the definition There is no

must produce the same result as ordinary zero. We have 1 ODxnf ~ fx) = F(n)

integration.

loss of generality by taking the lower limit of integration to be

(x_t)n-lf(t) dt.

The above is obviously oD if(x)~

true for =

n = i, for

fo

f(t) dt.

Now assume the formula true for

n = k:

12

- k f<x) o r(k) 1 oOx Replace k with k+l : =

~0X( x _ t ) k - l f ( t )

dt.

oD; (k+l) f(x)

kr (k)

1 ...........

SO X ( x _ t ] k f l t ]

dt.

Operate on both sides of the above with d = 77 1 (x

oDx = d/dx

and we have

(x-t)kf(t)dt.

(12) Applying

ODx kLeibnitz's

f(x)

~}(k)flo
of an integral dt dx gives

rule for the derivative 1 - kr(k)

-k f(x) 0Dx

fox
of

~ (x_t)kf(t)

g(O,x) - 7 T where the function g is the integrand

d(O)

+ g(x,x) (x-t)kf(t)

--~

in (12).

The last two terms d(O)/dx have -k f(x) = oDx and since k = n = O, and g(x,x) = 0

on the right above vanish because because the factor (x-x). T h e n we

rx kr('~'i
k Io (x-t)k-lf(t)dt' induction we have by mathematical
fX

oDxn f ( ~ ) = r(~"i
This result

(x_t)n-lf(t)dt.

is the same as (9) obtained heuristically. for differentiation criterion of arbitrary order will differentia-

The definition tion. It follows that

be shown later to be an integration (7) fulfills and integration. Criterion 3 states

followed by ordinary

2 for differentiation

that the operation of order zero leaves of of

the function unchanged, whether D v f f as

C x

0 f(x) = O. The investigation that is, cDx v 0 shows a concern with the continuity We have

the
C

D -v operator
x

at

v = O.

13
-X

(13)

c DO x f(x) of letting

- F(O) 1 v = O

~ ( x - t ) O-1 f ( t ) d t , in (7). The f a c t o r 1/r(O) can

as a consequence be taken general, O-~ .

equal to zero because F(O) = ~ . The i n t e g r a l would, in be divergent a n d we h a v e t o d e a l w i t h t h e i n d e t e r m i n a t e form several

f(t)

There are

ways of handling
is e x p a n s i b l e

this

situation.
series w i t h

We assume remainder: f(t) and t a k i n g

in a T a y l o r ' s

= f(x)

+ (t-x)f'(x) sides

+ (t-x)2f"(x)/2i vO

+ ---

limits

of b o t h

of (7) as

we have

vO

lim

oDxVf(x)

vO

lim

[Jo

(x-t)v-1
F (v)

f(x) dt

~Ox ( x - t ~ , , v f ' ( x ) d t + - - "

r(v)

+ (-1) n

_t) vn-1 ;ox ( x n !r(v)

f(n) (x)dt

+ (_l)n+ 1 f(n+l)(@) (n+l) I Because except

~OX

(x_t)v+n F (v) ....... d t right has above vanish the value

l i r a r ( v ) = ~ , a l l t h e t e r m s on t h e v+O the first because the first integral

xVf(x) r (v+l)
due to the g a m m a no of Thus lira vO oDx v f(x) = lira xVf(x) vO F (v+l) = f(x) function relation vF(v) = r(v+l). We n o t e O there is

loss of g e n e r a l i t y c.

u s i n g the

l o w e r limit of i n t e g r a t i o n

instead

D O f(x) Ox
We can also arrive

at the same

result

in the f o l l o w i n g

manner.
Let the f u n c t i o n The integral f be c o n t i n u o u s on the interval (c,x). (7) can then be w r i t t e n as the sum of two i n t e g r a l s :

14

(14)

c D-v x f(x) - F1 (v) ~cX If (t)

-X

f(x)] (x-t) v- ldt

r (v)

f(x)

(x_t) V-ldt.
in (14) will be shown to into two sub-

The first tend to zero as intervals v

integral

on the right

tends to zero.

It can be divided

of integration:

(15)

F(v) 1

f(t)
r(v)

f(x)]

(x-t) V-ldt

~xX6 If(t)- f(x)] (x-t)V-ldt

number. + of B. If(t) f(x) I by s. Thus, Let us designate these in-

where tegrals

6 as

is any small positive A In B denote

the maximum

If(t)
in integral f B, where s is continuous.

f(x) l _-< ~(s)

on 6, and lim s(6) 6+0 = 0 because

depends

We then have

1 IBI =< r--'~) ~(6)

1

~xX
6
~v
(x_t)V-ld t

~ (~)~v

--< ~
Then, for all

s(~)''T-:

r(v+l)

v ~ O, we have
lim

6+0

IBI

: o.

After evaluating Eq. (14) is written as

the second

integral

on the right

in (14),

cDVf(x)-~_
where IBI 0 with 6.

= A B + (x - c ) v

{(x)

r (v+l)

15

We can now consider maximum of If(t) f(x) I

the integral Then

in (15).

Denote

the

by M.

M ~c x-6 (x-t)V-ldt IAI ~ ~7-~)

r(v+l)

(~v _ (x_c)V).
Now choose 6 ~, IAI + 0 as so that v + O. To

Let e be any arbitrary positive number. IBI < e for all v ~ O. For this fixed both sides of (16) add -f(x). Ken cDxVf(x) Because f(x)

$ [A I + IBI + If(x) l LF(v+l)

-i

IBI < e, we have lim sup I cDx -v f(x) vO - f(x) I ~ 0 + ~ + 0

Since

can be chosen as small as we wish, lim I cDx vf(x) vO f(x) I = O,

it follows

that

or

lim
vO

D-Vf(x)
C x

= f(x).

Another

approach

to the above result using the theory of f(x) f(x) only be such

Laplace transforms might be of interest. If we define in [O,L] then f can be taken as zero in x > L. Let that

L
exists for some real ~(s) a.

~ e_aXf (x) dx

Then it follows e-Stf(t)dt

that

f
x
J

is an analytic

function of

in

Re(s)

> a, that, with

v > -i,

-sx

r (v)

(x_t)v-lf(t) d t

16 exists in Re(s) > a, and in fact, that

oo

7(s,v) where g(x,v)

-'o

e-SXg(x,v)

dx ~ s - v ~ ( s ) ,

denotes the right side of (7).

It is also true that 1 ~ f l (x) ="~T~

eS x ~(s) ds
Re (s) > 0 and where fl(x)

where

is any vertical path lying in f(x)

differs from

on, at most, a countable number of points.

Furthermore, g(x,v) - 2 1 ~

~G

e sx g(s,v) ds.

But, for such a path

G, everything is uniformly bounded and ; e st f(s)ds

z

lim g(x,v) - 2 1 ~ vO

/c

fl(x)

which is the result wantedj[16]. We now consider criterion 5: By definition (7) we have 1 ~c X (x-s) U-ld s r(u)
s

D-u CX D-v f(x) = cDxu-v f ( x ) . CX

(17)

F(vi

(S

_t)v- 1

f(t)dt.

The repeated integral above corresponds to a double integral to which Dirichlet's formula, mentioned earlier, may be applied. We have

(18)

-v f(x) - r(u)r(v) cDx u c D x

/cx

f(t) dt

~-x I (x-s)U-I (s-t) v-I ds. t

17 When either to (18) u or v is on the interval (0,I), the passage from (17)

can be justified by a minor m o d i f i c a t i o n

of the Dirichlet

proof over a smaller triangle. Make the transformation gral on the right in (18) is then y = (s-t)/(x-t). The second inte-

(x-t) u+v-I

fo 1

(I -y)U-lyV-ldy

which

is a beta integral

that has the value

r (u) r (v) (x_t)u+v-1 r (u+v)

When this is substituted into 1 (18), we obtain

~D~ u cDx ~ f(x)

The integral on the right

= r(u+v)

X(x_t)u+v_if (t)dt.

above is definition We then have problem

(7) with

u+v

playing

the role of arbitrary

order.

the required result.

A subtle mathematical the law of indices atives of arbitrary get the divergent order.

arises when one seeks to extend of arbitrary order to derivwe will

stated for integration integral

If we follow the preceding method,

Du Vf = 1 c x cDx (x) F(-u-v)

(x-t)

-(u+v)

-if(t)dt.

To establish

the relation

(19)
it will be required which vanishes This proof

f(x) c Du x cD~f(x ) = c _u+v Ux to impose the restriction that f be a function f(c) = O.

at the lower limit of integration,

namely

is omitted here but details The restriction that f

can be found in [6]. at x = c and at its n-i

vanishes

derivatives, interchange ing (19).

as stated in criterion

2, is necessary

to justify the

of the order of operations For example, the relation

used in the proof of establish-

DD -I f(x) = DOf(x)

= f(x)

18 always holds. But the relation

(20)
is not

D-1D f ( x )
always valid.

DOf(x) = f ( x )
(7)

For, by definition = cDx - 1 f'(x)

-I cDx f(x) C DX

= r(1)
= f(x) -

(x-t)Of'(t)dt
f(c),

and (20) holds only when The definition For differentiation However, sion. by means Let v = m-p v, and

f(c)

= O. of arbitrary order. expres-

(7) is for integration

of arbitrary where

order it cannot be used directly. m is the least integer of arbi-

of a simple

trick, we can find a convergent Then for differentiation

for convenience

greater than

O < p _ _ < i.

trary order we have D v fix)

C X

=
C

Dm
X C

D -p f(x]
X

(21)

_ dm 1 dx m r(p)

~c x (x-t)P-lf(t) dr'

where we take a~Ivantage of the knowledge mth derivative definition that operator dm/dx m. Dm-p = DmD -p. referred Dm-p

that

Dm
X

is an ordinary of this

We have assumed for purposes

The simple trick sults from the fact that operator D -v. fractional

to above, namely

D v = Dm-p

re-

is the analytic

continuation

of the

It is obvious

that criterion the definition than zero.

1 which re(7) for inteorder is We have

quired analyticity, by hindsight.

and also the other four criteria,

were established

The question of extending v

gration of arbitrary order to differentiation answered by letting be real and greater

of arbitrary

(22)

(v,x)

= oDxVf(x)

= F(v) 1

~O

~x

(x-t)v-lf(t)dt

which

is in general

convergent

for

v > O.

For any

we can write

19

(v,x)

= oD-Vx f(x) = oDmx oDx p- f(x) dm 1 dx m F(p) if (x-t) p(t) dt,

~0 X

where

-v = m-p, When

m = O, I, 2, v > O choose

.... m = O. Thus v = p and ~ -- ,. Now~

(22) can be written ~(v,x) = d~oX [~(v)~oX(x-t)v-lf(t)dtl dx.

By Dirichlet's

formula, we have d 1 ~0 x (x-t)vf(t)dt = ~-~ r(v+l)

~(v,x)

which

is convergent ~(v,x)

for

v > -i. for for

We then have m = I. v ~ -n, n and q ~ in a positive is analytic RI~R 2 Dm-p for integer. in R2 Now for

= ~(V,X)

This process v > -n. point of ~.

can be repeated in R1 where ~ = ,

is analytic Since

v > O then

on a set of points

with a limit continuation Dv. will be

in the right half plane, This justifies Some explicit

is the analytic

the trick of writing examples

of fractional

derivatives

useful. letting formula

For the fractional v = m-p, m

derivative

of a constant

k, we have by

the least integer k

> v, and the use of (21), the -v

(23)

oD~, k - r ( I - v )
Another example

is the integration logarithm.

and differentiation

of

arbitrary order of the natural By definition

(7) we have

(24)

D -v In x = Ox

F(v)

1/oX

(x-t)V-lln

t dt,

V > O.

20 Let t : x+ t - x, x>O

: x(1
Then

+ t~x).

in t = In x + in(l + t;x)

with the restriction

-I <

t-x < i .
X =

Using the Taylor's

series

expansion
co

for

In(l+@),

we get

In t = in x +

(-l)n-l(t-x)n
nx n

L_.
n=l where the interval of convergence is

O < t ~ 2x.

Substituting

the

right side of the above

into the right

side of (24) gives

in x

~0 X (x_ t) v- Idt co
1 r(v)

r (v)

f
xv r(v)

v-1 (x-t)

n=l

(x-t) n -- nx n

dt

Term by term integration, gives the result oD;V

permissible

because
co

of uniform convergence,

xVln x

in x = r(v+l)

1 ~<~+k~

In terms written

of the psi function,

the above result can be

C ~(v+l) ]

oDx v in x =
where C is Euler's In x

XV
r (v+l)

[ In

constant.

For differentiation

of arbitrary

order of

we have D m-p in x Ox

Dv I n x = Ox

am
dx m (p+l)

r(p)

xpZ
k=l

k(

where the usual

criteria

for termwise

differentiation

is to be applied.

21
Although we now know how to interpolate orders where of the derivative such procedures of functions such as might be applicable. between integral is known this

in x, little

In this connection

writer

s u b m i t t e d a p r o b l e m t o t h e American Mathematical Monthly t o vin r(x). This will permit appear in winter 1974-75, concerning 0 Dx interpolation b e t w e e n i n t e g r a l o r d e r s o f t h e p s i f u n c t i o n and m i g h t h a v e u s e in t h e s u m m a t i o n o f s e r i e s o f t h e f o r m ~ 1 / ( 1 + x ) u.
Eric Russell Love [17] has defined order integration n where of pure of integra> O to

imaginary order in such a way as to extend the properties

tion and differentiation the case where cosine Re(n) the Riemann-Liouville imaginary of arbitrary = O. Francis H. Northover makes (7) can be connected

Re(n)

the claim that to the Fourier of pure

definition

and Fourier sine transforms order as follows. 1

l~ X

by means of derivatives

cDx v- F(x)

- r(v)

I (x-t)V-IF(t)dt, ;cl

Re(v)

> O.

Make the transformation t = x - (x-c)e -@. The limits have cDx v- F(x) = (terminals) of integration then become O and ~, and we

(X-c),V"
r(v)

-V~ F(#)

d~),

v F(x

-c)V .... ( X r(v$---S{

Now let

v = -in, and assume

cD1x n F(x)
oo

exists.

Then

(26)

cDlxn F(x)

- (x-c)-{n r(-in)

L
(2)

e ~n~ F(~) {c(n)

d~

(x-c) -in r(-in)

where

+ { s(n)}

C(n) = (2)

50

F(~)

cos n~ d~

and

S(n)

= (2)

V(~) s i n

n~ d~.

22 Love has shown that suitably derivatives of any order function which preceding exist. Consider now
.X

restricted = O

functions

have of a in the to

of all orders v for Re(v)

where > O.

Re(v)

but have no derivative a derivative is assumed

He has also cited an example but does not possess that cD~ ~ F(x) caution was exercised

is locally integrable

of any imaginary order. paragraph where

For this reason, it was stated

I = Assume f(t)

-v aDx

f(x) = F~v)

/a

(x-t)v-lf(t)dt" series

is expansible

in a Taylor's
co

f(t) = 1 n=O
The substitution of the series for

(-1)n f(n)(X)n! (x-t)n

f(t)
co

in the integrand

above gives

(26a)
Now if f(x) = (x-a) p,
_

1
I =

~ n=O

V (V)

(-1) n f(n)(x ) (x-a) v+n (v+n) n!

p > -i, then

r(p+l)

(x-a)V+P,

aDx v(x-a)p

= F(p+v+l) the identity

+

where we have noted without proof

r (v) ~ = 'r ( p + v + l )
If

1 vr(p+l)

1 ( v + l ) r (p)

(v+2) 2!r

(p-l)

f(x) = (x-b) p ,

p > -i, then from

(26a)

(x_a)V(x_b)p
F(v)

y(_l)ncx_a~
~x-b j n=O

r(p+l)

......

aDxV(x-b)P forO<b~8. We recall D -u-v f(x)

a x

(v+n) n! r(p-n+l)

the laws of exponents With the results

or indices

-v a D x -u f(x) = aDx of integra-

is written

for the case when both terminals of the index rule,

tion are the s ~ e . a measure

just given one can investigate say for example,

of deviation

23 f(x) = x: aDx v bDx u x. Some special arbitrary nection with the Bessel oD;(P+) (See Open Questions, functions function: # 3, p. 376this text). of

can be represented function.

as an integral

order of an elementary

We wish to show the con-

cos~~ = 2p /7 up-Zjo(/~).

For

Re(p)

> - , we have

Jp(X) =

(x/2) p 7 P (P+)

(l-t2) p acos xt dt. [18]

Make the t r a n s f o r m a t i o n =

xt = w, the above becomes 2 SO x (x2-w2)p-cos w dw.

Jp(X)
Let

(2x) P/7 r (p+)

x 2 = u, w 2 = v, and the above becomes

~U

2pfg uP/2jp(~) = F(p+)l *)Oi (u-v) p- cscrv~ dv.

These transformations forms
COS ~/~

have

given us an integral which order p+, and

con -

to our definition

(7), of arbitrary

f(u) =

/g

So, the above may be written 2p~~ uP/2jp(~-~) = oD~(P+)

in the form cos/~

g
which is the result we sought to verify. Here we show how a hypergeometric sented by the fractional tions. (27) ab b+l x 2. + 1 + l ~g x + a(a+l) .b ( . .) . 2!g(g+l) series because it is a generalization of operation function can be reprefunc-

of a product

of elementary

is called a h y p e r g e o m e t r i c the geometric in common use: series

1 + x + x 2 -'-

The following notations

are

(r)n = (r+l) (r+2)'''(r+n-l), 2Fl(a,b ;g;x)

24 The subscript 2 preceding F denotes two parameters in the numerator. The subscript 1 denotes one parameter in the denominator. Using this notation, (27) can conveniently be written in summation form: (28) 2Fl(a,b;g;x) = n=O Some properties of the gamma and beta functions which will be needed later are briefly outlined. (b)n = b(b+l)'''(b+n-l) r(b+n) r(~) (a)n(b)n n n!(g)n x .

(29)

(g)n

g(g+l)...(g+n-1)

=~ F(b)

"r(g+n)
(29) becomes

Using the gamma-beta relation

B(p,q) = F(p)F(q)/F(p+q),

(30)
Thus,

~=

(b)n

B(b+n. g-b]

B(b, g-b)

(28) becomes
co

(31)

1 2FI ta'b;g;xJ'' - B(b, g-b)

~ n-O

(a)/~B (b+n,g-b) n!

.x n

where the factor I/[B(b,g-b)] is placed before the summation sign because it is independent of n. Writing symbol 2FI B(b+n, g-b) as a beta integral, we then have and using the

instead of

2Fl(a,b;g;x),
co

(32)

2F1 = B ( b , g - b )

/__z n! n=O

xn

/o I(l_t) g_b_itb+n_id t "

The interchange of the summation sign and the integral sign is permissible because of the uniform convergence of the series:
oo

(33)

2FI = B(b~g-b)

;o I(l-t)

g_b_itb- 1
=

(a)n(Xt) n

ni

dr.

Using the fact that

25

n=O
we find that 2FI = ~ 1 valid if Ixl < i, and All the right To do this let

(n n! (xt)n = (1-xt)-a'

(33) becomes (l-t)g -b -Itb -l(l-xt) adt,

g,b > O. the integral on

that is required now is to transform xt = s, and we have x_g+l ~x 10 (x-s) g-b- isb-i (l-s) -ads .

above to an integral

of the form of the definition

(7).

2FI = B(b,g-b) Using the relation

B(b,g-b)

= r(b)r(g-b)/r(g), we obtain

and writing

the inte-

gral above

in operator

notation,

the result

xg-lr (b) -(g-b)x(b-l(l_x )-a F(g) 2Fl(a'b ;g ;x) = oDx

Before tional There One is calculus, appears fractional

turning

our attention

to some applications another

of fracof

it will be useful and another

to mention

definition

integration

access to a fractional Weyl's

derivative. definition.

to be two representations of Hermann

x W-v~

f(x) = F(v) 1 r j ~ Jx

(t-x) v-I f(t)

dt, Re(v)

> O.

The significant Liouville function W ~+B

differences

between

this

definition

and the Riemannand the kernel exists, the Weyl W~W B = inte-

definition here being ~ and

are the terminals (t-x) v-l. B way. Kenneth

of integration integral derives

When the Weyl S. Miller

for all

gral in the following Let L dn L = Po(X) dx n

be the linear d n-I + Pl (x) dx n-

differential

operator

1 + "'" + Pn (x)~

26

whose finite

coefficients interval of L and x ~ I, g(x)

Pk' 0 ~ k ~ n, are of class and Po(X) on > 0 I, and on I. H (x,~) its one-sided

C~

on some closed Let L be the Then in I, function.

I = [a,b]

adjoint if f

Green's xo

is any function continuous

is any point

then for all (36)

=
Xo

H (x,~)f(~)d~

is the solution satisfies

of the nonhomogeneous conditions

equation

L y = f(x)

which

the boundary g(k)(

X O

) = O,

0 = < k = < n-l. that for H (x,~) L. xo = b =-H(~,x) where H(x,~) is

Now let

xo = b .)

and recall

the one-sided Green's cited on p.90

function

(See p. 37 of Miller's

text

Then if we let

g(x)

t t ( ~ , x ) f ( ~ ) d~

is the solution of 0 < k < n-l. Now if

L y = f(x)

with initial

conditions

g(k)(b)

= O,

d n -

dx n then L is formally self-adjoint L, since L = (-l)nL. We recall that

for this p a r t i c u l a r

(as in (i0)), (x_~)n-I

H(x,~) Thus

i - (n-l)!

g(x)

r(n)

~xb (~_x)n-lf(~) d ~

is the unique solution

(-i) n

of the adjoint

equation

dnX,' = f(x)
dx n

(with the initial may call

conditions

g(k) (b) = O,

0 =< k < n - 1 . )

So we

27

xWbv f ( x ) - F(w)

if
fixed, lim

(g-x)W-lf(g)dg

Re

>

the adjoint fractional integral

it)
.

(unless someone else has already named

Now for

sufficient

condition

that

xWb~f(x)

b~

exists

is f(x) = 0

x < 0

and
foo

j
0

ix,21zfZ(x)d x

<

(Apply the

Cauchy-Schwarz Formally

inequality.)

dx x w~'~f(x) = _x%(,~- 1 ) f(x)

and, for e x a m p l e ,

xW~ (,J
Make

- 1)

e x

-~
x = y

r-~)

( x)'~ le-~d~'

X > O.

the t r a n s f o r m a t i o n

and we have

xW-(V~ -I) e -x

d e -x ~-~ ~,(,;)

yU_le -Ydy

e -x

r(,~)

-X
= e

One n o t e s

that -d m w -v f(x) dxmX~ = ( i] m w m - v

---x~

f(x)

so that

W- e-X ~ e-X, x ~

28 and
wm- x co e-X = e-X

for any nonnegative

integer

m. Re ~ > O and Re v > O.

The laws of exponents hold for The argument

_ _

is similar to (17) and (18):

,

xW'~[x W~vf(x)]

- F(~)F(~)

(t-x)~-idt

(~-t)~-if(~)d~

= r (~)r(~)

f(<)d<

fx

(t-x)~-l(~-t)W-ldt

oo

B(~ ,v)

~x f(~) (~-x) ~+V-ld~

X w-(~+v)f(x) co

,
See also the paper by Kenneth S. Miller,

which is a law of exponents. this text, pp. 80 - 90.

Another gorm of Weyl's series Mikol&s involving periodic

definition

is that of an infinite have been used

functions which will be mentioned by Mikl6s These definitions of certain Fourier series. of defining derivatives unable ~ of

in his paper this afternoon.

by A. Zygmund in the treatment

One of the most recent methods fractional Paul L. Butzer and Ursula Westphal tend but their paper appears of Butzer and Westphal, nonperiodic function

order is by the limit of a fractional later in this text. is obtained

different

qnotient. to atof the

are, unfortunately, of arbitrary

Using the notation order We define

the derivative

xY/F(y+l)

as follows.

~f(t)

= ~(-l)J(~)

f(x-tj)

j=O

fy

(x) =

XY _

(x > O) (x < O)

r(O+l)

29

[t-~&~f (x)] t y

(s)

= t -a 1--~ sY+l _ 1 sy-a+l

(1-e-St)

~l-e-St) ~ st

=~[(f~-a(u) ,1g pa(~))(x)](s)

where the function pa(x) is defined by

pa(x)

- r(~)
O__<j<x

(-1)J

(]) (x-j)a-1

X > O.

pa(x)

belongs

to

LI(o,~)

and has the + fy-a for

Laplace

transform

s-a(i-e-S)

Thus we h a v e

t -aAa t f y(x)

t O+ a bit contrived, of the too far operations.

f(y) of a weir h of

The f o l l o w i n g the purposes frequency a backdrop The i n t e g r a l a n d an i n t e g r a l equation

problem, is

admittedly kernel for it

serves (x-t) v Its to use flow.

o f s h o w i n g how t h e consideration is solved

function is not

form fetched of fluid

formulated

in the physical

sciences.

deserves

of economics,demand equation

a n d money f l o w i n s t e a d by fractional

The problem is to determine notch, an opening Q, through the notch is expressed the notch~ [19]. We first establish

the shape

in a dam, in which the volume flow rate of fluid, as a function the equation of the height

Q(h)

= c

(h-y)f(y)dy.

Consider

front and side views

y

of the notch below:

Y

JlJiiiiJlJlJiiiJiiJJ !
,

T h

Side view Fig. 1

Front view Fig.

30 Assuming the points (37) a and that Bernoulli's y in Fig. equation can be applied between

I, [20],we

obtain

Pa V2 V2 - - + gh + a = Py + gY + _y_ p 2 p 2 ' Pa' Py' Va' Vy y; g are the pressures and velocities and p at points

where and

is the gravitational

acceleration

is the fluid

density. The pressures atmospheric ing. (38) so that

(59) V Y = (2g) (h-y)

at

and

are both

taken to be nearly at a is assumed to is slow mov-

so that

Pa = Py'

and the velocity the fluid behind

be negligible Thus,

(V a = O) since (37) becomes

the notch

gh = gy + V~/2,

gives

the velocity

of the fluid at distance

above

the notch

floor

(x-axis). The elemental (40) So, by definition,

(41)

area

(shaded

region

in Fig.

2) is given by

dA = 2f(y) dy. the elemental volume flow rate through dA is

dQ = V dA = 2 ( 2 g ) ( h - y ) 2 f ( y ) d y . Y

Denoting

2(2g)

by volume

and integrating flow rate through

(41)

from

y = O

to

y = h

gives the total

the notch:

(42)

Q(h)

= c

(h-y) f(y)dy.

We find fractional (43) Operating

f(y)

by finding (7)~ Eq. = r(~) (42)

f(h).

By the definition

of

integration Q(h)
C

can be written

in the form

oDh 3/2 f(h). oD~/2 gives the result

on both sides of (43) with

(44)

f(h)

i F(3/2)

oD~/2qc(h)

31
But oDh
312

= O D2 oDh (44)

where

oDh

is

d21dh 2 -h

Denoting

Q(h)/c

by

g(h), we can write

as follows:

(45) Since
second

1 M 2 1 fO (h-Y) -g(y) dY" f ( h ) - F(3/2) dh 2 F(%) Q(h)/c i s known, then

the beta we obtain

g(h)
f(h). let

and
Thus, g(h)

g(y)

are known.
above, f(y). Then

Then,
its

after evaluating derivative

integral

on the right

and taking

we have = ha .

As a specific f(h) - 2
7[ 2

example

(45) yields

r(a+l__.___~) ha_3/2

r (a-h)

valid and if

for

a > -.

If

a = 7/2,

the weir

is shaped

like a parabola,

a = 3/2, In his

the weir

is a rectangle. to this writer, problem. h H. and The minimum are at H Robert M. Hashway, of circuof The

communication

Warwick,

R.I.

suggests above through t(H)

a similar

A fluid reservoir respectively.

lar symmetry the reservoir fluid exits reach by t(z)

is to be designed. the ground an orifice above = O. height

and maximum heights

The time for the fluid to say height z, is given the shape of the reservoir. of Leibnitz's rule

a particular where

the ground,

Determine

We will now for the nth derivative

consider

a generalization

of a product:
oo

(46)

oDVf(x) g(x)

~,

~(n) (v) OVx

f(x)

oD(V-n) g (x) .

n=O
D (n) is ordinary differentiation the identity x a+b = xax b Operate on both sides with D v. Treat (46). series the left hand side as a frac(21) and treat the of By equating coefficients and D (v-n) is fractional differ-

entiation.

Consider

tional derivative right hand side like powers, F(a+b+l) r(a+b+v+l)

in accord with the definition in accord with

we get an infinite I I F(b+v+l)

of gamma functions: v(v+l)a(a -I) + 2!F(b+v+3) I "'" "

= r(b+l)

va F(b+v+2)

32
The case for fractional plicity it offers type. the Volterra calculus might well lie in the simintegral equations f(x) of explicitly

in the solution of certain

Consider the problem of finding

given the equation xf(x) =

2x
(x-t) - f(t)dt. F() we have D - f(x). O x subscripts for convenience

By definition

(7) the right hand side above is

Omitting

(483 Operating

xf(x) on b o t h

= ~ D " f ( x ) . of the above with D yields

sides

(49)
Apply formula

D2xf(x)

= ~ f ( x ) .

(46) to get

(50)
Substituting

xDf(x)

+ 1D-f(x)

= ~ f ( x ) .

(48) into (50) gives

xDf(x) + 2vr~
for

(51)

x f ( x ) ....

= ~ f ( x ) .
by operating on both sides of

We can get an expression (48) with D:

Df(x)

(52) or
(53) Our objective

D[xf(x)]
xf'(x)

= ~ D f ( x ) ,
+ f(x) = ~ D f(x). (53) is substituted equation into (51).

has been reached when

We arrive at the ordinary

differential

x2f,(x)
which has the solution

3y x _ 7 ) f(x) = 0 + (-

f(x) = ke-~/Xx -3/2, author of Laplace Transforms and other series, suggests the following solution

Murray R. Spiegel, texts in the Schaum's outline to the previous problem.

33

xF(x)

~0 X( x - u ) - F ( u ) d u
f(s) = (---~) f ( s )

= x - * F ( x ) .

_ d_

= ~

f(s)

ds
f'(s)/f(s)

s
= / ' ~ s -

s~

'

In f(s) = 2/-~ s + c I

f(s) = ce -2/-~

F(x) = c X -I

- -

-4~/4x

2 ~JTTT~
= cx-3/2e-~/x No claim can be made that the fractional is better than some other approach. Parker Higgins who confided there is a succinctness the fractional functional calculus equation of notation However, calculus approach Theodore in

to paraphrase

in me tha~ he paraphrased and simplicity

A. Erd&lyi, to a complicated

of formulation

that might suggest a solution

that is not readily

obtained by other means. properties The

In 1940 and 1941 Erd&lyi of a generalization tions. Professor topic of fractional early nineteen others formal

and Kober investigated

of the Riemann-Liouville calculus lay relatively Mikol~s,

and of the Weyl definidormant from 1941 to the More papers Osler and to the of some

Sneddon will survey some of these results. when a modest Higgins, resurgence began.

sixties,

were published by Erd~lyi, applied mathematician techniques

AI-Bassam,

in the 1960's and early 1970's. for the solution,

Of particular

interest

in the last decade was the development by means of fractional equations

operations,

of dual and triple integral value problems

that stem from mixed boundary

of mathematical

physics.

The pair of equations

f O ~ K(x,t)G(t)f(t)dt

= g(x)

O<x<l,

34

O~ K ( x , t ) f ( t ) d t where the f(t) idea gral tial is is to kernel reduce K(x,t), G(t), are

= h(x) g(x), known as physical in finding

x > 1 h(x) dual are to known f u n c t i o n s equations. of dual for a pair and The inte-

t o be d e t e r m i n e d for

integral

a specific example,

problem

equations,

an e x p r e s s i o n

the poten-

in the field of an electrified hold over two different When the problem

disc where different

boundary

conditions

parts of the same boundary. conditions hold intein the

is such that different

over three different ient to determine gral equations.

parts of the same boundary,

it is often convenfurther trends

the solution by constructing Professor Mikol~s will discuss of fractional calculus

a set of triple in his lectures.

theory and applications Fractional maticians variety of papers

calculus

is old but studied little. at this conference has several

Many matheThe wide

and scientists

are unfamiliar with this topic.

to be presented This conference

will help to One obvious scienwhich and cur-

fill this void. purpose ricula.

singular purposes,

singular being taken in the sense of Sherlock Holmes. is to popularize Another purpose tists and mathematicians to include

the topic in the hope it will induce it in their research is to exchange and impart information

may serve to suggest new areas of research. Fractional matics. proper objects mathematicians, useful chemistry, calculus can be categorized as applicable matheare

The properties

and theory of these fractional Scientists

operators calculus

of study in their own right. in the last decade, fields: rheology, theory,

and applied electro-

found the fractional quantitative However, biology, transport

in various

scattering

diffusion,

theory, probabilthey So,

ity, potential and scientists of fractional

theory and elasticity. are unfamiliar

many mathematicians

with this topic possibly because

have not been exposed to its applications. calculus has developed, of this conference formal methods another objective cover additional with mathematical fractional

Thus, while the theory attempts to disof

its use has lagged behind. is to encourage physical phenomena

of representing

models that can be treated with the elegance

calculus.

35

REFERENCES

[I] Leibnitz, G.W., Leibnitzen's Mathematische Schriften, Germany: Georg Olm, 1962, v. 2, pp. 301-302.

Hildesheim,

[2] Lacroix, S.F., Trait~ du Calcul DiffJrentiel et du Calcul Integral, Paris: Mme. vecourcier, 1819, Tome Troisi&me, seconde &dition, pp. 409-410. [3] Spanier, Jerome and Oldham, Keith B., The Fractional Calculus, New York: Academic Press, 1974. [4] Abel, Niels Henrik, "Solution de quelques probl~mes a'l'aide d'int&grales d~finies," Oeuvres Completes, Christiania, 1881, tome premiere, 16-18. [5] Liouville, Joseph, "M&moire sur quelques Qu&stions de G&ometrie et de M&canique, et sur un nouveau genre de Calcul pour r&soudre ces Qu&stions," Journal de l'Ecole Polytechnique, 1832, tome XIII, XXI e cahier, pp. 1-69. [6] A more detailed discussion of Liouville's first and second definitions and also of their connection with the Riemann definition can be found in The Development of the Gamma Function and A Profile of Fractional Calculus, by Bertram Ross, New York University dissertation, 1974, Chapter V, pp. 142-210. University Microfilms, Ann Arbor, Mich., #74-17154, PO #45122. [7] Debnath, Lokenath and Speight, T.B., "On Generalized Derivatives," Pi Mu Epsilon Journal, v. 5, 1971, ND 5, pp. 217-220, East Carolina University. [8~through[ll] Details will be found in "A Chronological Bibliography of Fractional Calculus with Commentary," by Bertram Ross in The Fractional Calculus [3], pp. 3-15, and in [6]. [12] Davis, Harold Thayer, The Theory of Linear Operators, ton, Indiana: The Principia Press, 1936; p. 20. [13] See [6], pp. 158-162. Blooming-

[14] The first one to apply Dirichlet's method to kernels of the form (x-t) v is Wallie Abraham Hurwitz in 1908. Cited by Whittaker and Watson, A Course in Modern Analysis, 4th edition, 1963, p. 76. [15] I am indebted to Dr. Kenneth S. Miller, Riverside Research Institute, New York City, for this contribution. [16] This approach was recommended by George F. Carrier, Harvard University. [17] Love, Eric Russell, "Fractional Derivatives of Imaginary Order," The Journal of the London Mathematical Society, Volume III (Second Series), 1971, pp. 241-259. [18] Farrell, Orin J. and Ross, Bertram, Solved Problems in Analysis, New York: Dover Publications, 1971, 279. First published in 1963, New York: The Macmillan Co.

36

[19] Brenke, W.C., "An Application of Abel's Integral Equation," American Mathematical Monthly, 1922, v. 29, 58-60. [20] Bernoulli's equation is strictly valid for steady, frictionless flow in a stream tube. It is used, however, in engineering for flows with friction by modification of solutions with a suitable friction factor.