Integral Talk Sonoma

Download as pdf or txt
Download as pdf or txt
You are on page 1of 21
At a glance
Powered by AI
The document discusses Liouville's theorem and its application to proving certain integrals cannot be expressed with elementary antiderivatives.

Liouville's theorem characterizes functions that can be primitively integrated in terms of elementary functions. It states that an integral can be expressed in terms of elementary functions only if the integrand is part of a tower of simple extensions built from rational functions.

The two integrals discussed are the integrals of e^{-x^2} and e^{x/x} with respect to x.

Introduction Elementary Functions and fields Liouville’s Theorem An example

Why certain integrals are “impossible”.

Pete Goetz

Department of Mathematics
Sonoma State University

March 11, 2009


Introduction Elementary Functions and fields Liouville’s Theorem An example

Outline

1 Introduction.

2 Elementary fields and functions.

3 Liouville’s Theorem.

4 An example.
Introduction Elementary Functions and fields Liouville’s Theorem An example

Probability

Central Limit Theorem

! x
1 2 /2
Φ(x) = √ e −u du
2π −∞

For probability applications, we need Φ(∞) = 1.

This is not proved by finding a formula for Φ(x) (by finding


2
an explicit antiderivative of e −u /2 ) and taking the limit as
x → ∞.
Introduction Elementary Functions and fields Liouville’s Theorem An example

Number Theory

Prime Number Theorem

π(x) = #{n ≤ x | n is prime }


! x
1
Li(x) = dt
2 ln(t)

π(x) ∼ Li(x) as x → ∞

1
This is not proved by finding an explicit antiderivative of ln(t) .

" 1
" eu
If u = ln(t), then ln(t) dt = u du.
Introduction Elementary Functions and fields Liouville’s Theorem An example

Elementary formulas

" 2 " eu
The indefinite integrals e −u du and u du do not have
elementary formulas.

How does one prove such claims?

First have to give a precise definition of “elementary formula”.

" 2 "u 2
After all e −u du = a e −x dx + C for any constants a and
C by FTC.
Introduction Elementary Functions and fields Liouville’s Theorem An example

History

Newton was perfectly happy to solve an integral by a power


series.

Leibniz preferred integration in ”finite terms” and allowed


transcendental functions like logarithms.
Introduction Elementary Functions and fields Liouville’s Theorem An example

Elementary function

An elementary function (roughly) should be a function of


one variable built out of polynomials, exponentials,
logarithms, trigonometric functions, and inverse trigonometric
functions, by using the operations of addition, multiplication,
division, root extraction, and composition.

sin−1 (x 3 − 1)
Example: #
ln x + cos(x/x 2 + 1)
Introduction Elementary Functions and fields Liouville’s Theorem An example

A simplification

We will use C-valued functions of the real variable x, i.e., our


constants will be complex numbers.

All trigonometric functions and inverse-trigonometric


functions can be written in terms of complex exponentials and
logarithms.

e ix − e −ix e ix + e −ix
sin(x) = , cos(x) =
2i 2
1 x −i
tan−1 (x) = (ln( ) − iπ)
2i x +i

−1 −1 x −1 −1 1 − x2
sin (x) = tan (√ ), cos (x) = tan ( )
1 − x2 x
Introduction Elementary Functions and fields Liouville’s Theorem An example

Meromorphic functions

A meromorphic function is a function defined on an open


interval I of the real numbers whose values are complex
numbers or ∞ with the property that sufficiently close to any
x0 in I the function is given by a convergent Laurent series in
x − x0 .

Rational functions are meromorphic on R.

Given a meromorphic function f , both e f and ln f are


meromorphic (one may have to restrict the domain of f ).
Introduction Elementary Functions and fields Liouville’s Theorem An example

Fields of meromorphic functions

Let C(x) denote the field of rational functions. Notice that


this field is closed under differentiation.

Any elementary function (under our rough definition) should


be in some “extension” of C(x).
Introduction Elementary Functions and fields Liouville’s Theorem An example

Fields of meromorphic functions

If f1 , . . . , fn are meromorphic functions, let C(f1 , . . . , fn )


denote the set of all meromorphic functions h of the form

p(f1 , . . . , fn )
h=
q(f1 , . . . , fn )

for some n-variable polynomials p, q '= 0 and q(f1 , . . . , fn ) is


not identically zero.
This definition captures the operations of addition,
multiplication, and division.
It is not hard to show that the set C(f1 , . . . , fn ) is a field and
that this field is closed under differentiation.
Example: K = C(x, sin x, cos x) = C(x, e ix ).
Introduction Elementary Functions and fields Liouville’s Theorem An example

Elementary fields

A field K is an elementary field if K = C(x, f1 , . . . , fn ) and


each fj is
an exponential or logarithm of an element of
Kj−1 = C(x, f1 , . . . , fj−1 )
or fj is algebraic over Kj−1 , that is fj is a solution to an
equation gl t l + · · · + g1 t + g0 = 0 where g0 , g1 , . . . , gl ∈ Kj−1

An elementary field is built from the the field of rational


functions in finitely many steps by adjoining an exponential, a
logarithm, or a solution to a polynomial.
Composition is captured by adjoining exponentials or
logarithms. Root extraction is captured by the adjunction of
algebraic solutions.
Elementary fields are closed under differentiation.
Introduction Elementary Functions and fields Liouville’s Theorem An example

Elementary functions

A meromorphic function f is an elementary function if it lies


in some elementary field.

$
Example: f (x) = 3 ln x + cos( x ) is an elementary
x 2 +i
function
x x
i( ) i( )
C(x) ⊂ C(x, ln x) ⊂ C(x, ln x, e x 2 +i ) ⊂ C(x, ln x, e x 2 +i ,f )
Introduction Elementary Functions and fields Liouville’s Theorem An example

Elementary integration

A meromorphic function f can be integrated in elementary


terms if f = g # for some elementary function g .

Recall an elementary field is closed under differentiation so if


f can be integrated in elementary terms, then necessarily f is
also elementary.
Introduction Elementary Functions and fields Liouville’s Theorem An example

Differential Galois theory

We can rephrase our problem: Given an elementary function


f , when does the differential equation dy
dx − f = 0 have an
elementary solution?

The answer is in the affirmative precisely when we can find a


tower of fields with special properties.

Consider the analogy with ordinary Galois theory.


Introduction Elementary Functions and fields Liouville’s Theorem An example

Liouville’s Thereom

Theorem (Liouville, 1835): Let f be an elementary function


and let K be any elementary field containing f . If f can be
integrated in elementary terms then there exist nonzero
c1 , . . . , cn ∈ C, nonzero g1 , . . . , gn ∈ K , and an element
h ∈ K such that
% gj#
f = cj + h # .
gj

& g! &
If f = cj gjj + h# , then g = cj ln(gj ) + h is an elementary
antiderivative of f .

The theorem is proved by induction on the length of a tower


of fields constructing K (g ) where g is an antiderivative of f .
Introduction Elementary Functions and fields Liouville’s Theorem An example

An important corollary

Corollary: Let f and g be in C(x) with f '= 0 and g


nonconstant. If f (x)e g (x) can be integrated in elementary
terms then there is a function R(x) in C(x) such that
R # (x) + g # (x)R(x) = f (x).

If R(x) ∈ C(x) satisfies R # (x) + g # (x)R(x) = f (x), then


R(x)e g (x) is an antiderivative of f (x)e g (x) .

2
We can apply this corollary to show that e −x and e x /x have
no elementary antiderivatives.
Introduction Elementary Functions and fields Liouville’s Theorem An example

2
Proof for e −x

Taking f = 1 and g = −x 2 in the Corollary, we must show the


differential equation

R # (x) − 2xR(x) = 1 (∗)

has no solution for R(x) ∈ C(x).

ODE’s shows" the general solution of (∗) is


2 2
R(x) = e x ( e −x dx + c) for any c ∈ C ... but this doesn’t
help!
Introduction Elementary Functions and fields Liouville’s Theorem An example

2
Proof for e −x

Suppose that R(x) ∈ C(x) is a solution to (∗).

R cannot be a constant or a polynomial in x (by degree


considerations).

Write R(x) = p(x)


q(x) for some nonzero relatively prime
polynomials p(x), q(x) with q(x) nonconstant.

Let z0 ∈ C be a root of q(x) of multiplicity µ ≥ 1. Then


p(z0 ) '= 0 and p(x)/q(x) = h(x)/(x − z0 )µ with h(x) ∈ C(x)
having numerator and denominator that are non-vanishing at
z0 .
Introduction Elementary Functions and fields Liouville’s Theorem An example

2
Proof for e −x

The quotient rule yields

p(x) # −h(x) h# (x)


( ) = +
q(x) µ(x − z0 )µ+1 (x − z0 )µ

As z → z0 in C the absolute value of (p(x))/q(x))# |x=z blows


up like A/|z − z0 |µ+1 with A = |h(z0 )/µ| =
' 0.
| − 2z · (p(z)/q(z))| has growth bounded by a constant
multiple of 1/|z − z0 |µ as z → z0 .
Therefore
p(x) # p(x) A
|(( ) − 2x · ( ))|x=z | ∼
q(x) q(x) |z − z0 |µ+1
as z → z0 .
This contradicts the identity R # (x) − 2xR(x) = 1.
Introduction Elementary Functions and fields Liouville’s Theorem An example

References

1 B. Conrad, Impossibility theorems for elementary integration,


http://math.stanford.edu/ conrad/papers/finalint.pdf.

2 T. Kasper, Integration in Finite Terms: The Liouville Theory,


Mathematics Magazine, Vol. 53, No. 4 (Sep., 1980), pp.
195-201.

3 E. A. Marchisotto and G. Zakeri, An Invitation to Integration


in Finite Terms, The College Mathematics Journal, Vol. 25,
No. 4 (Sep., 1994), pp. 295-308.

4 M. Rosenlicht, Integration in Finite Terms, The American


Mathematical Monthly, Vol. 79, No. 9 (Nov., 1972), pp.
963-972.

You might also like