Applicable Analysis and Discrete Mathematics

available online at http://pefmath.etf.rs

Appl. Anal. Discrete Math. x (xxxx),xxx–xxx.

doi:10.2298/AADMxxxxxxxx

LAPLACE TRANSFORM OF ANALYTIC COMPOSITE

FUNCTIONS VIA BELL’S POLYNOMIALS

Paolo Emilio Ricci, Diego Caratelli, Pierpaolo Natalini

Bell’s polynomials have been already used in many different fields, ranging
from number to operator theory. We show in this article another application
in the theory of Laplace Transform (LT), describing a method to compute the
LT of nested functions. To this aim, a table containing the first few values
of the complete Bell polynomials is shown, and a program for computing the
subsequent ones is reported. Furthermore a program for approximating the
Laplace Transform of analytic composite functions is derived by using the
general method presented in the article.

1. INTRODUCTION
In this short note we want to disprove the common view that there is no formula
for the definite integration of composite functions. In what follows, we show that,
as in the case of the derivative of nested functions, it is possible to use the Bell’s
polynomials [1, 5, 6, 8, 11] in order to obtain the desired result.
The Bell’s polynomials are also exploited in very different frameworks, which range
from number theory [9, 13, 14] to operators theory [12], and from differential
equations [8] to integral transforms [3, 10]. It is almost useless to recall how much
is used the Laplace Transform (LT) in the theory of ordinary and partial differential
equations and their applications to Analysis and Mathematical Physics problems,
ranging form signals analysis to imagine processing. Applications of the LT, and
of the even more useful Fourier Transform are recalled e.g. in [2].
We use the classical definition
∫ ∞
L(f ) := exp(−s t)f (t) dt = F (s) .
0

2020 Mathematics Subject Classification. 44A10, 05A10, 11B65, 11B83.

Keywords and Phrases. Laplace transform, Bell polynomials, Composite functions.

1
2 Paolo Emilio Ricci, Diego Caratelli, Pierpaolo Natalini

The LT converts a function of a real variable t (representing the time) to a function

of a complex variable s (representing the complex frequency).
The LT can be applied to local integrable functions on [0, +∞). It converges in
each half plane Re(s) > a, where a is a constant (called the convergece abscissa),
depending on the growth behavior of f (t).
Owing the importance of this transformation in the solution of the most diverse
differential problems, a large number of LT, together with the respective antitrans-
forms, are reported in literature (see e.g. [7]).
However, excluding some particular cases, we have not found in literature a gen-
eral method to compute the LT of composite analytic functions, moreover we have
found on the Internet the common opinion that such a formula has not been found.
This seems to be strange since, as in the case of the derivative of nested functions
problem, the Bell’s polynomials could be used.
To this aim, considering the Taylor’s expansion of a fixed analytic function, as the
coefficients of this expansion are expressed in terms of Bell’s polynomials computed
at the initial point, we approximate the integration problem by a series expansion.
Obviously, if the considered integral is convergent, the obtained series is convergent
too.
In the last sections the same methodology is applied to the case of the LT of nested
functions, starting from the easier case of the LT of a nested exponential function.
The result is a Laurent expansion approximating the considered LT. It is worth to
note that only in very few cases our results can be compared with the LT of special
nested functions present in literature, as it is shown in equations (22)-(23) and
(24)-(25). The main problem is to construct a table of Bell’s polynomials, which
exhibit a highly increasing number of addends, but their evaluation at a fixed point
does not present any difficult problem, using the modern computer programs.
To this aim, we show in the Appendix I the first few values of the complete Bell
polynomials, obtained by the program reported in Appendix II. These polynomials
are applied in the LT of nested exponential functions.
In Appendix III the computer program used for the LT of general nested functions
is shown. All these results were obtained by the second and third authors by using
Mathematica⃝ c
.

2. RECALLING THE BELL POLYNOMIALS

Consider the composite function Φ(t) := f (g(t)), where x = g(t) and y =

f (x), are differentiable functions (up to the considerad order), defined in suitable
intervals of the real axis, so that Φ(t) can be differentiated n times with respect to
t, by using the chain rule.
Here, and in what follows, we use the notations

Φm := Dtm Φ(t), fh := Dxh f (x)|x=g(t) , gk := Dtk g(t).

 Laplace Transform of analytic composite functions via Bell’s polynomials 3

Then the n-th derivative of Φ(t) is represented by

∑
n
(1) Φn = Yn (f1 , g1 ; f2 , g2 ; . . . ; fn , gn ) = An,k (g1 , g2 , . . . , gn ) fk ,
k=1

where the Yn denote the Bell polynomials, and the coefficients An,k , for all k =
1, . . . , n, are polynomials of the variables g1 , g2 , . . . , gn , are homogeneous poly-
nomials of degree k and isobaric of weight n (i.e. they are a linear combination of
monomials g1k1 g2k2 · · · gnkn whose weight is constantly given by k1 + 2k2 + . . . + nkn =
n).
The Bell polynomials satisfy the recurrence relation


 Y0 := f1 ;

 Yn+1 (f1 , g1 ; . . . ; fn , gn ; fn+1 , gn+1 ) =
(2) ∑ n ( )

 =
n

 Yn−k (f2 , g1 ; f3 , g2 ; . . . ; fn−k+1 , gn−k )gk+1 .
k
k=0

and they are given explicitly by Faà di Bruno’s formula (although this formula has
a very high computational complexity)
Yn (f1 , g1 ; f2 , g2 ; . . . ; fn , gn ) =
∑ n! [ g ] r1 [ g ] r2 [ g ]rn
1 2 n
(3) fr ··· ,
r1 !r2 ! . . . rn ! 1! 2! n!
π(n)

where the sum runs over all partitions π(n) of the integer n, ri denotes the number
of parts of size i, and r = r1 + r2 + · · · + rn denotes the number of parts of the
considered partition [11].

The An,k coefficients satisfy the recursion ∀ n

(4) An+1,1 = gn+1 , An+1,n+1 = g1n+1 ,
and, ∀ k = 1, 2, . . . , n − 1
∑
n−k ( )
n
(5) An+1,k+1 (g1 , g2 , . . . , gn+1 ) = An−h,k (g1 , g2 , . . . , , gn−h ) gh+1 .
h
h=0

Remark 1. Actually the coefficients An,k depend only by the values g1 , g2 , . . . , gn−k+1 ,
so that the above equation, shifting the indexes, writes
Bn,k (g1 , g2 , . . . , gn−k+1 ) =
(
∑ n−1
n−k )
(6) Bn−h−1,k−1 (g1 , g2 , . . . , gn−k−h+1 ) gh+1 ,
h
h=0

where we have used the traditional notation for the partial Bell polynomials
(7) An,k (g1 , g2 , . . . , gn ) = Bn,k (g1 , g2 , . . . , gn−k+1 ) ,
4 Paolo Emilio Ricci, Diego Caratelli, Pierpaolo Natalini

so that equation (1) writes

∑
n
(8) Yn (f1 , g1 ; f2 , g2 ; . . . ; fn , gn ) = Bn,k (g1 , g2 , . . . , gn−k+1 ) fk .
k=1

3. DEFINITE INTEGRALS OF COMPOSITE FUNCTIONS

Under the conditions of Sect. 2, let f (g(t)) be a composite function analytic

in a neighborhood of the origin, so that it is expressed by the Taylor’s expansion
∞
∑ tn
(9) f (g(t)) = an , an = Dtn [f (g(t))]t=0 .
n=0
n!

According to the preceding equations, it results

◦
a 0 = f (g 0 ) ,
(10) ∑
n
◦ ◦ ◦ ◦
an = Dtn [f (g(t))]t=0 = Bn,k (g 1 , g 2 , . . . , g n−k+1 ) f k , (n ≥ 1) ,
k=1

where
◦ ◦
(11) f k := Dxk f (x)|x=g(0) , g h := Dth g(t)|t=0 .

We can state the following

Theorem 2. Consider now the definite integral
∫ b
f (g(t)) dt ,
0

with 0 < b ≤ +∞ , then, assuming the limit value, for a convergent integral, when
b = +∞, it results
∫ b ∞
( n )
◦ ∑ ∑ ◦ ◦ ◦ ◦ bn+1
(12) f (g(t))dt = b f (g 0 ) + Bn,k (g 1 , g 2 , . . . , g n−k+1 ) f k .
0 n=1
(n + 1)!
k=1

Proof. Using equation (10) for the coefficient of Taylor’s expansion (9) we find
∫ b ∞ ∫
∑ b∑
n
◦
◦ ◦ ◦ ◦ tn
f (g(t))dt = b f (g 0 ) + Bn,k (g 1 , g 2 , . . . , g n−k+1 ) f k dt =
0 n=1 0 k=1 n!
∞
( n )∫
◦ ∑ ∑ ◦ ◦ ◦ ◦ b n
t
= b f (g 0 ) + Bn,k (g 1 , g 2 , . . . , g n−k+1 ) f k dt ,
n=1 0 n!
k=1
 Laplace Transform of analytic composite functions via Bell’s polynomials 5

where we have used the uniform convergence of Taylor’s series. Furthermore, when
b = +∞, convergence of the series in equation (12) - depending on the numerical
values of the coefficients appearing into brackets - is certainly ensured in case of
a convergent integral since the second member in equation (12) is just a different
expression of the first member.

4. LAPLACE TRANSFORMS OF COMPOSITE FUNCTIONS

The above results, applied in the case of a LT, give the result
Theorem 3. Considering a composite function f (g(t)) analytic in a neighborhood
of the origin, for its LT the following equation holds
∫ +∞
f (g(t))e−ts dt =
0
(13) ◦ ∞
( )
f (g 0 ) ∑ ∑
n
◦ ◦ ◦ ◦ 1
= + Bn,k (g 1 , g 2 , . . . , g n−k+1 ) f k .
s n=1
sn+1
k=1

Proof. Representing the coefficients of Taylor’s expansion in (9) in terms of Bell

polynomials, and using the uniform convergence of Taylor’s series, we find
∫ +∞
f (g(t))e−ts dt =
0
◦ ∞ ∫
∑ +∞ ∑
n
f (g 0 ) ◦ ◦ ◦ ◦ tn −ts
= + Bn,k (g 1 , g 2 , . . . , g n−k+1 ) f k e dt =
s n=1 0 n!
k=1
◦ ∞
( )∫
f (g 0 ) ∑ ∑
n
◦ ◦ ◦ ◦ +∞ n
t −ts
= + Bn,k (g 1 , g 2 , . . . , g n−k+1 ) f k e dt .
s n=1 0 n!
k=1

Then the result follows using the LT of powers.

4.1. THE PARTICULAR CASE OF THE EXPONENTIAL

FUNCTION

In the particular case when f (x) = ex , that is considering the function eg(t) ,
and assuming g(0) = 0, the we have the more simple form
∑
n
◦ ◦ ◦ ◦
Bn,k (g 1 , g 2 , . . . , g n−k+1 ) f k =
k=1
(14) ∑
n
◦ ◦ ◦ ◦ ◦ ◦
Bn,k (g 1 , g 2 , . . . , g n−k+1 ) = Bn (g 1 , g 2 , . . . , g n ) ,
k=1
6 Paolo Emilio Ricci, Diego Caratelli, Pierpaolo Natalini

where the Bn are the complete Bell polynomials. It results B0 (g0 ) := f (g0 ), and
the first few values of Bn , for n = 1, 2, . . . , 5, are given by
B1 (g1 ) = g1 ,
B2 (g1 , g2 ) = g12 + g2 ,
(15) B3 (g1 , g2 , g3 ) = g13 + 3g1 g2 + g3 ,
B4 (g1 , g2 , g3 , g4 ) = g14 + 6g12 g2 + 4g1 g3 + 3g22 + g4 ,
B5 (g1 , g2 , g3 , g4 , g5 ) = g15 + 10g13 g2 + 15g1 g22 + 10g12 g3 + 10g2 g3 + 5g1 g4 + g5 .
Further values are reported in the Appendix I, at the end of this article.

The values of the complete Bell polynomials for particular parameter choices
can be found in [9].
The complete Bell polynomials satisfy the identity (see e.g. [8])
n ( )
∑ n
(16) Bn+1 (g1 , . . . , gn+1 ) = Bn−k (g1 , . . . , gn−k ) gk+1 .
k
k=0

In this case equation (14) reduces to

∫ +∞ ◦ ∞
−ts exp(g 0 ) ∑ ◦ ◦ ◦ 1
(17) exp(g(t)) e dt = + Bn (g 1 , g 2 , . . . , g n ) n+1 .
0 s n=1
s

In what follows we show the approximation of the LT of nested functions.

It were obtained by the second author by using the computer algebra program
Mathematica⃝ c
.

4.2. EXAMPLES
We start considering the case of the LT of nested exponential functions
• Let f (x) = ex and g(t) = sin t. Then g1 = 1, g2 = 0, g3 = −1, g4 = 0, and in
general g2h = 0, g2h+1 = (−1)h , h = 1, 2, 3, . . . .
According to equation (15) it results
B1 (1) = 1, B2 (1, 0) = 1, B3 (1, 0, −1) = 0, B4 (1, 0, −1, 0) = −3, B5 (1, 0, −1, 0, 1) = −8 .
Then
∫ +∞ ( )
−ts 1 1 1 3 8 1
(18) exp(sin t) e dt = + 2 + 3 − 5 − 6 + O .
0 s s s s s s7
• Consider the Bessel function g(t) := J1 (t) and the LT of the corresponding
exponential function. We find
∫ +∞
1 1 1 3 11 19
exp(J1 (t)) e−ts dt = + 2 + 3 − 4 − 5
− +
0 s 2s 4s 4s 16s 32s6
(19) ( )
91 701 953 15245 1
+ + − +O .
64s7 128s8 256s9 512s10 s11
 Laplace Transform of analytic composite functions via Bell’s polynomials 7

• Let g(t) = − arctan(t), we find

∫ +∞
1 1 1 1 7 5 145
exp(− arctan(t)) e−ts dt = − 2 + 3 + 4 − 5 − 6 + 7 +
0 s s s s s s s
(20) ( )
5 6095 5815 1
− − + O .
s8 s9 s10 s11
• Consider the complete elliptic integral of the second kind g(t) := E(t) and the
LT of the corresponding exponential function. We find
∫ +∞
eπ/2 π π 2 − 3π π 3 − 9π 2 + 30π
exp(E(t)) e−ts dt = − 2+ − +
0 s 8s 64s3 512s4
(21) ( )
π 4 − 18π 3 + 147π 2 − 525π 1
+O .
4096s5 s6

4.3. THE GENERAL CASE

Examples of the above method for approximating the LT of general nested

functions are reported in what follows.

• Assuming f (x) = log(x), g(t) = cosh(t), it results

∫ +∞ [ ( ) ( s )]
1 1 s 1
(22) log[cosh(t)] e−ts dt = ψ + −ψ − 2 ,
0 2s 2 4 4 s
and using our approximation, we find
∫ +∞
log[cosh(t)] e−ts dt =
0
(23) ( )
1 2 16 272 7936 1
= 3
− 5 + 7 − 9 + 11 + O .
s s s s s s12

• Considering the Bessel function f (x) = J0 (x), a > 0 and g(t) = at2 , it results
∫ +∞
πs {[ ]2 }
(24) J0 (at2 ) e−ts dt = J1/4 (s2 /8a) + Y1/4 (s2 /8a) , ℜs > 0 ,
0 16a
and using our approximation, we find
∫ +∞
J0 (at2 )e−ts dt =
0
(25) ( )
1 6a2 630a4 207900a6 141891750a8 1
= − 5 + − + +O .
s s s9 s13 s17 s21
8 Paolo Emilio Ricci, Diego Caratelli, Pierpaolo Natalini

• Assuming f (x) = cos(x), g(t) = ae−t , it results

∫ +∞
cos(a) a sin(a) a[a cos(a) + sin(a)]
cos[ae−t ] e−ts dt = + − +
0 s s2 s3

a[3a cos(a) + (1 − a2 ) sin(a)] a[(a3 − 7a) cos(a) + (6a2 − 1) sin(a)]

(26) + +
s4 s5
( )
a[(15a − 10a3 ) cos(a) + (1 − 25a2 + a4 ) sin(a)] 1
6
+ O .
s s7

• Assuming f (x) = arcsinh(x), g(t) = at3 , it results

∫ +∞
arcsinh(at3 ) e−ts dt =
0
(27) ( )
6a 60480a3 98075577600a5 1
= 4
− 10
+ +O .
s s s16 s17

• Assuming f (x) = arctan(x), g(t) = log(1 + t), it results

∫ +∞
1 1 6 22 30
arctan[log(1 + t)] e−ts dt =
2
− 3+ 5− 6 − 7+
0 s s s s s
(28) ( )
952 5656 9952 508320 1
+ 8 − 9 + 10 − 11
+O .
s s s s s12

• Assuming f (x) = arcsinh(x), g(t) = t + 1, it results

∫ +∞
arcsinh(1) 1 3
arcsinh(t + 1) e−ts dt = + √ − √ +
0 s 2 2s 2 8 2s3
(29) ( )
19 189 2601 1
+ √ − √ + √ +O .
32 2s4 128 2s5 512 2s 6 s7

A lot of further examples can be constructed using the above method and
the most of them have not a close expression in terms of special functions.
 Laplace Transform of analytic composite functions via Bell’s polynomials 9

5. GRAPHICAL DISPLAY IN TWO KNOWN CASES

5.1. TEST CASE #1

Let us consider the function l(t) = log[cosh(t)]. As per the expression (22),
the Laplace transform of l(t) is given by:
[ ( ) ( s )]
1 1 s 1
(30) L(s) = ψ + −ψ − 2 ,
2s 2 4 4 s

for ℜs > 0.
By applying the approximation methodology presented in this study, we have
demonstrated in (23) that:
1 2 16 272 7936
(31) L(s) ≃ L̃(s) = − 5 + 7 − 9 + 11 ,
s3 s s s s
so that, by inverse Laplace transformation, one can readily conclude that:
( 2 4 6 8 10
)
(32) ˜l(t) ≃ t − t + t − 17t + 31t H(t) ,
2 12 45 2520 14175

with H(·) denoting the classical Heaviside distribution.

(a) (b)

Figure 1: Magnitude (a) and argument (b) of the Laplace transform of l(t) =
log[cosh(t)] as evaluated through the approximant L̃(s) and the rigorous analytical
expression L(s) for s = σ + i ω with ω = 1.

The distributions of L(s) and L̃(s) along the cut sections ω = ℑs = 1 and
σ = ℜs = 5 are reported in Figures 1 and 2, respectively. As it can be noticed, the
agreement between the exact transform (30) and the relevant approximation (31)
is very good especially as s → +∞. Conversely, the functions l(t) and ˜l(t) tend to
match for t → 0+ as one would expect from theory (see Figure 3).
10 Paolo Emilio Ricci, Diego Caratelli, Pierpaolo Natalini

(a) (b)

Figure 2: Magnitude (a) and argument (b) of the Laplace transform of l(t) =
log[cosh(t)] as evaluated through the approximant L̃(s) and the rigorous analytical
expression L(s) for s = σ + i ω with σ = 5.

Figure 3: Distribution of l(t) = log[cosh(t)] and the relevant approximant ˜l(t).

5.2. TEST CASE #2

Let us consider the function l(t) = J0 (t2 ). As per the expression (24), the
Laplace transform of l(t) is given by:
πs {[ ]2 [ ]2 }
(33) L(s) = J1/4 (s2 /8) + Y1/4 (s2 /8) ,
16
for ℜs > 0.
By applying the approximation methodology presented in this study, we can
demonstrate, on the basis of (25), that:

1 6 630 207900 141891750 164991726900

(34) L(s) ≃ L̃(s) = − + 9 − + − ,
s s5 s s13 s17 s21
so that, by inverse Laplace transformation, one can readily conclude that:
( )
˜ t4 t8 t12 t16 t20
(35) l(t) ≃ 1 − + − + − H(t) ,
4 64 2304 147456 14745600

with H(·) denoting the classical Heaviside distribution.

 Laplace Transform of analytic composite functions via Bell’s polynomials 11

(a) (b)

Figure 4: Magnitude (a) and argument (b) of the Laplace transform of l(t) = J0 (t2 )
as evaluated through the approximant L̃(s) and the rigorous analytical expression
L(s) for s = σ + i ω with ω = 1.

(a) (b)

Figure 5: Magnitude (a) and argument (b) of the Laplace transform of l(t) = J0 (t2 )
as evaluated through the approximant L̃(s) and the rigorous analytical expression
L(s) for s = σ + i ω with σ = 5.

Figure 6: Distribution of l(t) = J0 (t2 ) and the relevant approximant ˜l(t).

The distributions of L(s) and L̃(s) along the cut sections ω = ℑs = 1 and
σ = ℜs = 5 are reported in Figures 4 and 5, respectively. As it can be noticed, the
agreement between the exact transform (33) and the relevant approximation (34)
is very good especially as s → +∞. Conversely, the functions l(t) and ˜l(t) tend to
match for t → 0+ as one would expect from theory (see Figure 6).
12 Paolo Emilio Ricci, Diego Caratelli, Pierpaolo Natalini

B1 = g1 , Appendix I

B2 = g12 + g2 ,

B3 = g13 + 3g1 g2 + g3 ,

B4 = g14 + 6g12 g2 + 4g1 g3 + 3g22 + g4 ,

B5 = g15 + 10g13 g2 + 15g1 g22 + 10g12 g3 + 10g2 g3 + 5g1 g4 + g5 ,

B6 = g16 + 15g14 g2 + 45g12 g22 + 15g23 + 20g13 g3 + 60g1 g2 g3 + 10g32 + 15g12 g4 + 15g2 g4 + 6g1 g5 + g6 ,

B7 = g17 + 21g15 g2 + 105g13 g22 + 105g1 g23 + 35g14 g3 + 210g12 g2 g3 + 105g22 g3 + 70g1 g32 + 35g13 g4 +
105g1 g2 g4 + 35g3 g4 + 21g12 g5 + 21g2 g5 + 7g1 g6 + g7 ,

B8 = g18 + 28g16 g2 + 210g14 g22 + 420g12 g23 + 105g24 + 56g15 g3 + 560g13 g2 g3 + 840g1 g22 g3 + 280g12 g32 +
280g2 g32 + 70g14 g4 + 420g12 g2 g4 + 210g22 g4 + 280g1 g3 g4 + 35g42 + 56g13 g5 + 168g1 g2 g5 +
56g3 g5 + 28g12 g6 + 28g2 g6 + 8g1 g7 + g8 ,

B9 = g19 + 36g17 g2 + 378g15 g22 + 1260g13 g23 + 945g1 g24 + 84g16 g3 + 1260g14 g2 g3 + 3780g12 g22 g3 +
1260g23 g3 + 840g13 g32 + 2520g1 g2 g32 + 280g33 + 126g15 g4 + 1260g13 g2 g4 + 1890g1 g22 g4 +
1260g12 g3 g4 + 1260g2 g3 g4 + 315g1 g42 + 126g14 g5 + 756g12 g2 g5 + 378g22 g5 + 504g1 g3 g5 +
126g4 g5 + 84g13 g6 + 252g1 g2 g6 + 84g3 g6 + 36g12 g7 + 36g2 g7 + 9g1 g8 + g9 ,

B10 = g110 + 45g18 g2 + 630g16 g22 + 3150g14 g23 + 4725g12 g24 + 945g25 + 120g17 g3 + 2520g15 g2 g3 +
12600g13 g22 g3 + 12600g1 g23 g3 + 2100g14 g32 + 12600g12 g2 g32 + 6300g22 g32 + 2800g1 g33 +
210g16 g4 + 3150g14 g2 g4 + 9450g12 g22 g4 + 3150g23 g4 + 4200g13 g3 g4 + 12600g1 g2 g3 g4 +
2100g32 g4 + 1575g12 g42 + 1575g2 g42 + 252g15 g5 + 2520g13 g2 g5 + 3780g1 g22 g5 + 2520g12 g3 g5 +
2520g2 g3 g5 + 1260g1 g4 g5 + 126g52 + 210g14 g6 + 1260g12 g2 g6 + 630g22 g6 + 840g1 g3 g6 +
210g4 g6 + 120g13 g7 + 360g1 g2 g7 + 120g3 g7 + 45g12 g8 + 45g2 g8 + 10g1 g9 + g10 .

Appendix II

ntot = 10;
B0 = 1;
∑n
Do [ Bn+1 = k=0 Binomial [n, k] ∗ Bn−k ∗ gk+1 , {n, 0, ntot, 1}];
Do [Bn = Expand[Bn , {n, 0, ntot, 1}];
Do [Print [“B”n , “ = ”, Bn ], {n, 0, ntot, 1}];
 Laplace Transform of analytic composite functions via Bell’s polynomials 13

Appendix III

In this appendix we report the Mathematica⃝ c code used to evaluate the

approximation of the LT of general nested functions through Bell’s polynomials.

6. CONCLUSION

We have shown a method for approximating the integral of analytic composite

functions. We started from the Taylor expansion of the considered function in a
neighborhood of the origin. Since the coefficients con be expressed in terms of Bell’s
polynomials, the integral is reduced to the computation of an approximanting series,
which obviously converges if the integral is convergent.
Then this methodology has been applied to the case of the LT of an analytic
composite function, starting from the case of analytic nested exponential functions.
To this aim, a table of the used complete Bell polynomials is reported in Appendix
14 Paolo Emilio Ricci, Diego Caratelli, Pierpaolo Natalini

I, and the relevant computer program is shown in Appendix II. In the last section
the LT of a general analytic nested funtion is considered, and the used program
is described in Appendix III. We want to stess that even we dealt with a basic
subject, we have not found in literature any general method for approximating this
type of LT’s, a gap which, in our opinion, deserves to be overcome.

REFERENCES

1. E.T. Bell: Exponential polynomials. Annals of Mathematics, 35 (1934), 258–277.

2. R.J. Beerends, H.G. Ter Morsche, J.C. Van Den Berg, E.M. Van De Vrie:
Fourier and Laplace Transforms. Cambridge Univ. Press, Cambridge, 2003.
3. D. Caratelli, C. Cesarano, P.E. Ricci: Computation of the Bell-Laplace trans-
forms. Dolomites Res. Notes Approx., 14 (2021), 74–91. 14, 74–91.
4. C. Cassisa, P.E. Ricci: Orthogonal invariants and the Bell polynomials. Rend. Mat.
Appl., (Ser. 7) 20 (2000), 293–303.
5. L. Comtet: Advanced Combinatorics: The Art of Finite and Infinite Expansions. D.
Reidel Publishing Co., Dordrecht, Holland, 1974.
6. F. Faà di Bruno: Théorie des formes binaires. Brero, Turin, 1876.
7. A. Ghizzetti, A. Ossicini: Trasformate di Laplace e calcolo simbolico. (Italian),
UTET, Torino, 1971.
8. R. Orozco López: Solution of the Differential Equation y (k) = eay , Special Values
of Bell Polynomials, and (k, a)-Autonomous Coefficients. J. Integer Seq., 24 (2021),
Article 21.8.6.
9. F. Qi, D-W. Niu, D. Lim, Y-H. Yao: Special values of the Bell polynomials of the
second kind for some sequences and functions. J. Math. Anal. Appl., 491 (2) (2020),
124382.
10. P.E. Ricci: Bell polynomials and generalized Laplace transforms. arXiv preprint:
2103.07267.
11. J. Riordan: An Introduction to Combinatorial Analysis. J. Wiley & Sons, Chichester,
U.K., 1958.
12. D. Robert: Invariants orthogonaux pour certaines classes d’operateurs. Annales
Mathém. pures appl. 52 (1973), 81–114.
13. S.M. Roman: The Faà di Bruno Formula. Amer. Math. Monthly, 87 (1980), 805–809.
14. S.M. Roman, G.C. Rota: The umbral calculus. Advanced in Math., 27 (1978),
95–188.

Paolo Emilio Ricci

Sezione di Matematica
International Telematic University UniNettuno
E-mail: p.ricci@uninettunouniversity.net
 Laplace Transform of analytic composite functions via Bell’s polynomials 15

Diego Caratelli
Department of Electrical Engineering
Eindhoven University of Technology
E-mail: d.caratelli@tue.nl

Pierpaolo Natalini
Dipartimento di Matematica e Fisica
Università degli Studi Roma Tre
E-mail: natalini@mat.uniroma3.it