Solving Polynomial

Equations
Part III
 Ferrari and the Biquadratic
Ferrari's solution of the quartic (biquadratic) equation involved the
introduction of a new variable and then specializing this variable to
put the equation into a form that could easily be solved. Finding the
right specialization involved solving a cubic equation (called the
resolvent of the original quartic). Here are the details, again using
modern techniques.
Consider the general quartic equation
x4 + ax3 + bx2 + cx + d = 0,
and rewrite it as
x4 + ax3 = -bx2 -cx -d.
Now add ¼ a2x2 to both sides to make the LHS a perfect square:
(x2 + ½ ax)2 = (¼ a2 – b)x2 – cx -d.
 Ferrari and the Biquadratic
We introduce a new variable by adding y(x2 + ½ ax) + ¼ y2 to both
sides of the equation (this keeps the LHS a perfect square):
(x2 + ½ ax + ½ y)2 = (¼ a2 – b + y)x2 + (-c + ½ ay)x + (-d + ¼ y2).
If we can chose a y so that the RHS is a perfect square, the resulting
quartic equation would be very easy to solve. Now, a quadratic
Ax2 + Bx + C
is a perfect square (has two equal roots) if and only if B2 – 4AC = 0.
So we consider the equation (resolvent):
(-c + ½ ay)2 = 4(¼ a2 – b + y)(-d + ¼ y2) or
y3 – by2 + (ac – 4d)y + 4bd – a2d – c2 = 0.
With y being any solution of this cubic, we obtain
(x2 + ½ ax + ½ y)2 = (ex + f)2,
where (in general) ay
−c


2
a 2
e= −b y and f = .


4 a
2
2 −b y
4
 Ferrari and the Biquadratic
The solutions of the quartic can now be obtained by solving the two
quadratic equations:
x2 + ½ ax + ½ y = ex + f and x2 + ½ ax + ½ y = -ex – f.

We illustrate this procedure with a simple example

x4 + 3 = 4x.
Here a = b = 0, c = -4 and d = 3.
The resolvent is
y3 -12y - 16 = 0.
We could use the Cardano formula to obtain a root, but inspection
yields y = 4 in this case. This gives e = 2 and f = 1. Therefore, the
quadratics we solve are:
x2 + 2 = 2x + 1 and x2 + 2 = -2x -1
or
x2 – 2x + 1 = 0 and x2 + 2x + 3 = 0.
 Complex Roots
From the first we obtain x = 1 (repeated) and the second gives
x = -1 ± i√2.

Cardano would have rejected the complex roots of the second

equation here, but he did, in the Ars Magna, at least consider them
in regards to a quadratic problem. He stated, after verifying that the
complex roots satisfied the original quadratic:
So progresses arithmetic subtlety the end of which, as is said, is
as refined as it is useless.

Full recognition of complex roots of equations came about 15

years later in the Algebra of Rafael Bombelli (1526 - 1572).
 The Fifth Degree Equation
During the next two centuries several alternate solutions for the
cubic and quartic equations were found. In all of these solutions an
auxiliary equation (the resolvent) was used. In some cases, the
resolvent equation had a degree larger than the equation that was
under consideration, but because of its special form a solution was
obtainable.

The success with the cubic and quartic equations naturally led to a
search for a similar solution of the general quintic (5th degree)
equation. But the methods that had worked with the lower degree
equations generally produced resolvent equations which could not
be solved. New ideas were needed for any progress to be made,
and these surfaced in the 18th Century.
 Lagrange
The two greatest mathematicians of the 18th Century were Euler
and Lagrange, and which of the two was the greater is a matter of
debate that often reflects the differing mathematical sensitivities of
the debaters. - Eves, An Introduction to the History of Mathematics

Joseph Louis, Comte Lagrange (1736 – 1813) was born in Turin.

He was Italian by birth, German by adoption, and Parisian by
choice.

He began his teaching as professor of mathematics in the artillery

school at Turin (1755) when he was only nineteen. He was not one
of the infant prodigies in mathematics, and showed no interest in
mathematics until he was seventeen; but within a few years he was
recognized as the greatest living scholar in his science.
 Lagrange
In 1766, Frederick the Great wrote that “the greatest king in Europe”
wanted “the greatest mathematician of Europe” at his court. As a
result of this letter, Lagrange went to Berlin, succeeding Euler as
mathematical director in the Berlin Academy and remained there for
more than 20 years.
With the death of Frederick in 1787 and the resulting turmoil in
Prussia, Lagrange accepted an invitation by Louis XVI to take up
residence in Paris, and became a member of the Académie des
Sciences.
Lagrange was made president of the commission that designed the
metric system. This work was valued at the time of the French
Revolution, and even though all foreigners were banished from
France, the Committee of Public Safety expressly excepted
Lagrange from this decree.
 Lagrange

Lagrange was revolted by the cruelties of the Terror that followed

the French Revolution. When the great chemist Lavoisier went to
the guillotine, Lagrange expressed his indignation at the stupidity
of the execution: “It took the mob only a moment to remove his
head; a century will not suffice to reproduce it.”
Lagrange had decided to leave Paris, but was invited to establish
the mathematics department at the newly created École Normale
(1795). This chance to establish a training program for teachers
based on the most thorough mathematical scholarship was
sufficient to change his mind. Two years later, he was asked to do
the same at the new École Polytechnique.
Under Napoleon he was made a senator and a count, and was
awarded other honors appropriate to his genius.
 Lagrange
Lagrange's mathematical work had a deep influence on later
mathematical research, for he was the earliest first-rank
mathematician to recognize the thoroughly unsatisfactory state of
the foundations of analysis and accordingly to attempt a
rigorization of the calculus.
He also gave a method for approximating the real roots of an
equation by means of continued fractions, obtained general
equations of motion of dynamical systems, worked on differential
and partial differential equations and contributed to the calculus of
variations. He also had a penchant for number theory and wrote
important papers in this field also, such as the first published proof
of the theorem that every positive integer can be expressed as the
sum of not more than four squares.
He also did fundamental work on the theory of equations ...
 Resolvents
Lagrange carried out a detailed investigation of all the various
solutions of cubic and quartic equations in order to discover
techniques that would work for quintic and higher degree equations.
In particular, he was interested in determining the relationship
between the original equations and the resolvents that were used in
their solutions.

Starting with the cubic equation x3 + nx + p = 0, the substitution

x = y – (n/3y) leads to the 6th degree resolvent y6 + py3 – (n3/27) = 0.
With r = y3, this is a quadratic equation in r which has two roots r1
and r2. The cube roots of r1 and r2 are solutions of the resolvent and
can then be used to obtain a value of x. Since r2 = -(n/3)3/r1, we get
x =  r 1  r 2 .
3 3
 Resolvents
This was Cardano's solution, but he only used the real cube roots of
r1 and r2. Lagrange knows that there are really three cube roots of a
number:
The three cube roots of a number r are
−1  −3 −1−  −3
3 r ,  3 r and 2 3 r where  = 2
and  = .
2 2

The numbers 1, ω and ω2 are the solutions of x3 – 1 = 0,

called the cube roots of unity.
Thus, the resolvent really has 6 roots (and not just the two real
roots). “Plugging” these values in, one obtains only three solutions
of the cubic since each of the cubic roots is obtained by two of the
resolvent roots. These solutions of the cubic are:
 Resolvents
x1 =  r 1  r 2
3 3

  r 1  r 2 and
3 23
x2 =
  r 1  r 2 .
23 3
x3 =
Cardano came close to the realization that other roots of the
resolvent would give other solutions to the cubic ... for he says:
I need not say whether having found another value for [a root of
the resolvent] ... we would come to two other solutions [for x]. If
this operation delights you, you may go ahead and inquire into this
for yourself.
 Resolvents
Having expressed all of the roots of the cubic equation in terms of
all of the roots of the resolvent, Lagrange proceeds to now write
the roots of the resolvent in terms of the roots of the cubic. He
finds that the six roots of the resolvent can all be obtained from
y = (1/3)(x' + ωx'' + ω2x''')
where x', x'' and x''' are the three roots of the cubic arranged in any
order (there are 3! = 6 such arrangements ... we call them
permutations today.)

Lagrange then makes three observations concerning the roots of

the resolvent equation.
1. Because there are 6 permutations of the roots of the cubic,
the degree of the resolvent equation is 6 = 3!.
 Resolvents
2. Even though there are 6 values of y, there are only 2 values
of y3. There are two sets of three permutations each, and each set
corresponds to one of the values of y3. Since there are only two
values of y3, y3 must satisfy a quadratic equation.
3. The coefficients of the resolvent equation are rational
functions of the roots of the cubic equation.
In the case we are looking at, y6 + py3 – (n3/27), the 0 coefficients
of y5, y4, y2 and y are all equal to x1 + x2 + x3, while
p = -x1x2x3 and
3
−n
3
− x 1 x 2 x 1 x 3 x 2 x 3 
= .
27 27
Lagrange looks at other solutions of the cubic and finds the same
relationships with the resolvents.
 Quartic Resolvents
Lagrange next looks at Ferrari's solution of the quartic. The roots of
the resolvent in this case look like,
y = ½ (x'x'' + x'''x'''')
where the x's are the four roots of the quartic in any of the 4!= 24
possible arrangements. This time, however, there are only three
distinct values of the y's. Thus, the resolvent has degree 3 and its
coefficients are again rational functions of the roots of the quartic.

Lagrange is now ready to tackle the general problem. He was going

to look for a resolvent of degree k (< n, the degree of the equation to
be solved) whose roots would be certain functions of the roots of the
original equation, functions that take on only k values when the
roots of the original equation are permuted in n! ways.
 Lagrange's Attempt
In his search for such functions he did prove that the degree of the
resolvent he sought would be a divisor of n! (but he could not show
that it was less than n!).

Furthermore, he proved that if all the permutations of the roots

which leave some function f fixed also leave another function g
fixed, then g can be expressed as a rational function of f together
with the coefficients of the original equation. Furthermore, if g is
not fixed by the permutations which fix f, but is changed to one of a
set of r different functions, then g will be a root of an equation of
degree r whose coefficients are rational functions of f and the
coefficients of the original equation.
 Lagrange's Attempt
Lagrange hoped to use these results to solve the general
polynomial equation of degree n. He would start with a symmetric
function of the roots (a function fixed by all the permutations of
the roots, such as x1 + x2 + .... + xn) and then find a function g that
takes r values under these permutations. This function would
satisfy an equation of degree r, with coefficients rational in the
original coefficients – the f here is one of the original coefficients.
If this equation could be solved then he would try to find a new
function h which would take on s values under the permutations
which leave g fixed. He would continue in this manner until the
solutions of the original were found. Unfortunately, Lagrange was
unable to find a general method of determining these intermediate
functions that could be solved by known methods. He had to quit.
 Ruffini

An Italian physician, Paola Ruffini (1765 – 1822)

proposed that the general quintic equation could not be
solved using radicals. His proof was privately published
and revised in 1803, 1805 and 1813. It is claimed that no
contemporary mathematicians could understand this
purported proof. It does have a gap. However, Ruffini
does develop all the essential ideas of group theory that
are used in this proof, and his approach is essentially
correct.
 Abel
Niels Henrik Abel (1802 – 1829) was born in Findö, Norway, the
son of a country minister. Plagued by poverty and suffering from a
pulmonary condition throughout his short life, this brilliant young
mathematician died at 26 from an attack of tuberculosis at Froland
in Norway.
His native abilities in mathematics were discovered by his
mathematics teacher in high school who encouraged Abel to read
advanced mathematics texts. After graduating from the university,
he obtained a small stipend which enabled him to tour Europe and
advance his knowledge. He wrote several papers during this year of
visiting France, Germany and Italy. These were highly regarded by
the European mathematicians. Five of them appeared in the first
issue of the new Journal für die reine und angewandte Mathematik
also known as Crelle's Journal.
 Abel
The editor of this journal, August Crelle, soon became one of
Abel's best friends and promoter. Abel returned to Norway in 1927
when his money was exhausted only to find that the only opening
at the university had just been filled by his former teacher. Unable
to obtain employment, he attempted to support himself by tutoring
and doing substitute teaching. Two days after his death, a delayed
letter arrived from Crelle informing him that a position had been
secured for him in Berlin.

Abel wrote papers in various areas of mathematics dealing with the

convergence of infinite series, the theory of doubly periodic
functions, elliptic functions and the theory of equations. His name
is associated with fundamental results in all these areas today.
 Abel
While he was a student at the university, he thought that he had a
solution of the quintic in terms of radicals. As his Norwegian
professors could not understand his proof, he sent it to Denmark.
Before it was published, he was asked to provide some numerical
examples, and while searching for these he realized that his method
was incorrect. He continued to work on the problem for the next
few years and eventually proved that no such solution is possible.
He privately published this result as a pamphlet in 1824. In order to
save printing costs, he had to give the paper in a very summary
form, which in a few places affects the lucidity of his reasoning. A
longer version appeared later in Crelle's Journal.
 Abel
After showing that the general quintic (and any higher degree)
equation could not be solved by radicals, Abel set himself the
problems of
1) Finding all equations of any degree which could be solved by
radicals, and
2) Finding a decision procedure which would determine whether
a given equation is algebraically solvable or not.

He did not live long enough to carry out this program, but he did
make some headway. In particular, in the case where all the roots
of an equation can be expressed as a rational function of one of
them, say x, then if for any two roots α(x) and β(x), we have
αβ(x) = βα(x), then the equation is algebraically solvable.
 Galois
Évariste Galois (1811 – 1832) had an even shorter and more tragic
life than Abel. Galois was born in Bourg-la-Reine, a town not far
from Paris in which his father was elected mayor in 1815. He had
mixed success in preparatory school (mostly due to trouble in the
humanities), but began to shine in mathematics to the exclusion of
all other subjects. He published a short paper before he was 18,
and submitted a memoir on the solvability of equations of prime
degree to the French Academy at the same time. However, he
twice failed the entrance exam to the École Polytechnique (the first
time because he wasn't prepared and the second attempt was a few
days after his father had committed suicide). Galois enrolled at the
École Normale to prepare himself for a teaching career.
 Galois
After the director locked the students into the building so that they
could not participate in the political activities leading to the July
revolution of 1830, Galois attacked the director in a letter for
favoring “legitimacy” over “liberty”. He was expelled.
Galois joined a heavily republican division of the National
Guard, a division that was soon dissolved because of its perceived
threat to the throne of King Louis-Phillipe. He was arrested twice
for his political activities and sent to prison for six months for the
second offense (wearing the uniform of the dissolved National
Guard division).
Although heavily involved in political activities, he continued to
study mathematics. He mastered the mathematical textbooks of his
time with the ease of reading novels and went on to the important
papers of Legendre, Jacobi and Abel, and then to creating his own.
 Galois
Two of his memoirs, sent to the French Academy were mislaid and
lost, and a revised version of his memoir on the solvability of
equations was rejected because the referee could not understand the
proofs.
While he was in prison his anger at the Academy for their failure
to appreciate his work grew to such a degree that he lashed out at
France's “official scientists” in a vicious diatribe intended to be a
preface to the private publication of his work.
At this time he also met Stéphanie-Felicie du Motel, daughter of
a physician who lived nearby. After he left prison she rebuffed his
advances and Galois was drawn into a duel of honor with a fellow
republician Pécheux D'Herbinville. On May 30, 1832, five months
before his 21st birthday he was shot in the stomach and died the
following day of peritonitis.
 Galois
The evening before the duel, expecting to die on the morrow, he
wrote a letter to his friend Auguste Chevalier in which he set forth
briefly his discovery of the connection of the theory of groups
with the solution of equations by radicals. At the end of the letter
he wrote:
I have often in my life ventured to advance propositions of
which I was uncertain; but all that I have written here has been in
my head nearly a year, and it is too much to my interest not to
deceive myself that I have been suspected of announcing theorems
of which I had not the complete demonstration.
Ask Jacobi or Gauss publicly to give their opinion, not as to the
truth, but as to the importance of the theorems.
Subsequently there will be, I hope, some people who will find it
to their profit to decipher all this mess.
 Galois
This letter was published, as requested, later that year in the Revue
encyclopédique, but his other manuscripts lay unread until they
were finally published in 1846 by Liouville in his Journal des
mathématiques. Within the next few years, several mathematicians
included Galois' material in university lectures or published
commentaries on the work. It was not until 1866, however, that
Galois theory was included in a text, the third edition of the Cours
d'algebre of Paul Serret (1827-1898). Four years later, Camille
Jordan (1838-1922) published his monumental Traité des
substitutions et des équations algébriques (Treatise on
Substitutions and Algebraic Equations), which contains a
somewhat revised version of Galois theory.
 Galois
Galois' life has been the subject of many biographies (one fictional).
Several of these have been criticized for being overly imaginative
and selective in what they report about Galois – trying to make a
case rather than reporting the facts. One of the worst offenders is
E.T. Bell in his very popular Men of Mathematics. Bell seems to
want to make the case for a gentile misunderstood genius who was
battered around by a cruel world. While Bell mentions his sources,
he seems to chose only certain items and distorts others in order to
support his point of view. A closer reading of the historical
documents do not paint Galois in as rosy a light as Bell would have
it. While his genius is beyond question, “brash”, “impatient”, and
“hothead” would be a bit more descriptive of his personality.