International Journal of Mathematical Education in

Science and Technology

ISSN: 0020-739X (Print) 1464-5211 (Online) Journal homepage: http://www.tandfonline.com/loi/tmes20

An efficient algorithm for computing the roots of

general quadratic, cubic and quartic equations
Munir Mahmood, Sali Hammad & Ibtihal Mahmood
To cite this article: Munir Mahmood, Sali Hammad & Ibtihal Mahmood (2014) An efficient
algorithm for computing the roots of general quadratic, cubic and quartic equations,
International Journal of Mathematical Education in Science and Technology, 45:7, 1095-1103,
DOI: 10.1080/0020739X.2014.902133
To link to this article: http://dx.doi.org/10.1080/0020739X.2014.902133

Published online: 08 Apr 2014.

Submit your article to this journal

Article views: 110

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tmes20
Download by: [Nat and Kapodistran University of Athens]

Date: 03 December 2016, At: 12:39

International Journal of Mathematical Education in Science and Technology 1095

It is worth noting that n = 8 is the largest even integer for which the taxi regular
n-gon is also a Euclidean regular n-gon. This is because adjacent sides in the octagon are
reflections across horizontal or vertical lines or lines that have slope 1 or 1 which preserve
Euclidean lengths and taxi lengths. For larger even integers there will be adjacent edges
of the n-gon that are not reflections over any of these special lines. Therefore, while the
Euclidean lengths of a side rotated about an end point would be preserved, the taxi lengths
will not be preserved. For these adjacent edges, the Euclidean lengths and the taxi lengths
cannot both remain unchanged.

Reference
[1] Krause E. Taxicab geometry. New York (NY): Dover Publications; 1986.

An efficient algorithm for computing the roots of general quadratic,

cubic and quartic equations
Munir Mahmooda , Sali Hammadb and Ibtihal Mahmoodc
a
Department of Mathematics & Natural Sciences, Gulf University for Science & Technology,
Hawally 32093, P.O. Box 7207, Kuwait; b Foundation Program Unit, Gulf University for Science &
Technology, Hawally 32093, P.O. Box 7207, Kuwait; c University of Melbourne, Parkville 3052,
Australia

(Received 26 October 2013)

While the solution to deriving the roots of the general quadratic equation is adequately
covered in a typical classroom environment, the same is not true for the general cubic and
quartic equations. To the best of our knowledge, we do not see the roots of the general
cubic or quartic equation discussed in any typical algebra textbook at the undergraduate
level. In this paper, we propose an efficient algorithm in order to calculate the roots of
the general quadratic, cubic and quartic equations. Examples are given to demonstrate
the usefulness of this proposed algorithm.
Keywords: roots; algorithm; quadratic; cubic; quartic

1. Introduction
An emphasis is given to the general quadratic equation, its solution and applications in most
of the algebra textbooks. On the other hand, the solutions to the general cubic and quartic
equations are not covered. As a consequence, the general cubic and quartic equations are
almost never taught in a classroom as a part of a lecture. The aim of our paper is to present
an efficient algorithm so that the students are able to study and compare the required steps
for the quadratic, cubic and quartic equations in order to arrive at their respective solutions.
One way to conduct this is to compute the roots of the general quadratic equation with
the algorithm, then to modify and extend it to calculate the roots of the general cubic and
quartic equations.

Corresponding author. Email: [email protected]

http://dx.doi.org/10.1080/0020739X.2014.902133

1096

Classroom Notes

The solution to the general cubic equation is due to Cardano (1539), who used complex
numbers to solve the roots.[1] Cardanos student Lodovico Ferrara discovered a method
of solving the general quartic equation. While Cardanos method is cited in the literature,
it is not suitable for teaching. The formula for the roots of the cubic equation is quite
complicated and the roots do not always lead to the desired results.[2] The solution of
the general quartic equation from Ferraras method has the same complexity in terms of
teaching.[2, 3] Recent articles have demonstrated alternative methods to determine the
roots of the cubic equation.[4, 5, 6, 7] To date, a systematic approach or algorithm is not
available in order to popularize the solution to the general cubic and quartic equations
among students. We propose a general algorithm in the next section, which can be easily
taught to solve for the roots of the general quadratic equation and then in subsequent
sections the algorithm is modified to calculate the roots of cubic and quartic equations.

2. The proposed algorithm: quadratic case

Consider the general quadratic equation given by
a2 x 2 + a1 x + a0 = 0 with a2 = 0.
Without loss of generality, we provide below the steps of the proposed algorithm for
x 2 + a1 x + a0 = 0

(1)

STEP 1: Substituting
x = u
in Equation (1) yields u2

a12
4

a1
2

(2)

+ a0 = 0 which may be written as

u =


1 2
a 4a0 .
2 1

(3)


STEP 2: If a12 4a0 > 0, then u = 12 a12 4a0 from Equation (3). Using Equation (2),
x =


1 2
a1

a 4a0
2
2 1

(4)

are the roots of Equation (1).

STEP 3: If a12 4a0 = 0, then u = 0 from Equation (3). Clearly,
x =

a1
2

(5)

is the root of Equation (1). We observe that the two roots are identical and that the left-hand
side of Equation (1) becomes a perfect square.

International Journal of Mathematical Education in Science and Technology 1097


STEP 4: If a12 4a0 < 0, then u = 2i 4a0 a12 from Equation (3). The two complex
roots, using Equation (2), are
x =


i
a1

4a0 a12 .
2
2

(6)

Remark: The proposed algorithm above is applicable in calculating the roots of the wellknown quadratic equation ax 2 + bx + c = 0 with a = 0, and is left as an exercise for
the readers.
3. The proposed algorithm: cubic case
Consider the general cubic equation given by a3 x 3 + a2 x 2 + a1 x + a0 = 0 with a3 = 0.
Without loss of generality, we provide below the steps of the proposed algorithm for
x 3 + a2 x 2 + a1 x + a0 = 0.

(7)

STEP 1: Substituting
x = u

a2
3

(8)

in Equation (7) yields



2 3
1 2
1
u + a1 a2 u +
a2 + a0 a1 a2 = 0
3
27
3
3

(9)

which may be written as

u3 + A1 u + A0 = 0,
where A0 =

2a23
27

a1 a2
3

+ a0 and A1 = a1

a22
3

(10)

are real.

STEP 2: If A0 = 0 and A1 = 0 in step 1 or equivalently a23 = 27a0 , then clearly u3 = 0

from Equation (10). Using Equation (8), x = a32 is the root of Equation (7). We observe
that all of the three roots are same and that the left-hand side of Equation (7) turns out to
be a perfect cube.
STEP 3: If A0 = 0 and A1 = 0in step 1, then Equation (10) becomes u3 + A1 u = 0,
which yields, u = 0 and u = i A1 . Therefore, using these values of u in Equation (8),
one of the roots is x = a32 . Using A1 and Equation (8) from step 1, as a consequence, the

a2
a2
remaining two roots are x = a32 i a1 32 . We note that a1 32 could be negative
and in that case, the roots become real.

STEP 4: If A0 = 0 and A1 = 0 in step 1, then u = 3 A0 using Equation (10).

From Equation (8), x = 3 A0 a32 is the root of Equation (7). This means x =

3
2a 3
272 + a13a2 a0 a32 using A0 of step 1. As one of the roots is known, we now

1098

Classroom Notes

apply synthetic or long division in order to obtain the quadratic equation. This allows us to
arrive at the remaining two roots by using the appropriate steps of Section 2.
13


 A0 
4A31
1
2
2 A0 + 27
and use
STEP 5: If A0 = 0 and A1 = 0, then calculate c0 =
2
any of the three values for c0 in steps 6 or 7. Proceed to step 6 if c0 is real. Otherwise, go to
step 7 if c0 is complex. We note that c0 becomes complex when the term under the square
root sign of c0 is negative.
A1
and then obtain the real root x using Equation (8). We
STEP 6: Calculate u = c0 3c
0
then apply synthetic or long division to arrive at the quadratic equation, from which we
calculate the other two roots by using the required steps of Section 2.
1
STEP 7: Since c0 is a complex number of the form c0 = [m i n] 3 from step 5, the
cube root of c0 requires the calculation of and r.

 
(1) Note m = r cos and n = r sin implies = tan1 mn and m = r cos yields
r = cosm .

1 
(2) c0 = r 3 cos 3 + i sin 3 using and r from (1). This implies



2 
1
+ r 3 sin2 3 = r 3 and then proceed to step 6. We observe that
 13

m2 + n2 .
an equivalent expression for c0 =
c0 =

2 
r 3 cos2

Remark: We note that steps 5 and 6 are due to [4].

4. The proposed algorithm: quartic case

Consider the general quartic equation given by a4 x 4 + a3 x 3 + a2 x 2 + a1 x + a0 = 0
with a4 = 0. Without loss of generality, we provide below the steps of the proposed
algorithm for
x 4 + a3 x 3 + a2 x 2 + a1 x + a0 = 0.

(11)

STEP 1: Substituting
x = u

a3
4

(12)

in Equation (11) yields





3
1
1
1
u4 + a2 a32 u2 + a1 a2 a3 + a33 u + a0 a1 a3
8
2
8
4
1
3
a2 a32
a4 = 0
+
16
256 3

(13)

which may be written as

u4 + A2 u2 + A1 u + A0 = 0,

(14)

International Journal of Mathematical Education in Science and Technology 1099

where A0 = a0 14 a1 a3 +
3 2
a are real.
8 3

1
a a2
16 2 3

3 4
a ,
256 3

A1 = a1 12 a2 a3 + 18 a33 and A2 = a2

STEP 2: If only A0 = 0, then Equation (14) becomes u4 + A2 u2 + A1 u = 0, which

yields u = 0 and then one of the roots from Equation (12) is x = a43 . As a consequence,
the remaining three roots are calculated from u3 + A2 u + A1 = 0 using step 5 and
onwards from the cubic algorithm of Section 3.
2
STEP 3: If only
0, then Equation (14) becomes u4 + A
A0 = 0, which
 A1 =
2 u +
2
A2
A2
1
yields u = 2 2 A2 4A0 and from Equation (12), x = 2 12 A22 4A0 a43 .

STEP 4: If A0 = 0 and A1 = 0, then Equation (14) is rewritten as

u4 + A2 u2 + A1 u + A0 = (u2 + u + s)(u2 u + t).

(15)

Solving Equation (14) would be equivalently solving for

(u2 + u + s)(u2 u + t) = 0

(16)

if s, t and can be obtained from Equation (15). Equating the like coefficients for the
variables of Equation (15) yields the following equations;
A2 + 2 A1
2
A2 + 2 + A1
t =
2


z3 + 2A2 z2 + A22 4A0 z A21 = 0,
s =

(17)
(18)
(19)

where z = 2 .
For further reading, the readers are directed to [8].
STEP 5: Substitute z = y

2A2
3

in Equation (19) to arrive at

y 3 + B1 y + B0 = 0,

where B0 =

72A0 A2 2A32 27A21

27

and B1 =

(20)

A22 12A0
.
3

STEP 6: Using step 5 of Section 3,

13

3
4B
B
1
0
1
c0 =
B02 +

2
2
27

(21)

using B0 and B1 of Equation (20).

B1
in the spirit of step 6 of Section 3 and then, z = y
STEP 7: Calculate y = c0 3c
0
using step 5, respectively.

STEP 8: Calculate = z of step 4 using the value of z from step 7.

2A2
3

1100

Classroom Notes

STEP 9: Calculate s =
STEP 10: Calculate t =

A2 + 2
2

A1

A2 + +
2
2

A1

of step 4 using steps 1 and 8.

of step 4 using steps 1 and 8.

STEP 11: Calculate the pair of u values from u =

factor of Equation (16) of step 4.

2 4s
2

STEP 12: Calculate the other pair of u values from u =

the second factor of Equation (16) of step 4.

say, u1 and u2 , from the first

2 4t
2

say, u3 and u4 , from

STEP 13:Finally, use x = u a43 of step 1 to obtain the solutions x1 , x2 , x3 and x4 from
the values of u from steps 11 and 12, respectively.
Remark: In general, the principal root value of c0 works towards the solutions of step 13
and it is recommended that the solutions are verified using Equation (11).

5. Illustrations
In this section, we provide various examples of the quadratic, cubic and quartic equations
in the Sections 5.15.3, respectively.

5.1. Illustrations (quadratic case)

We provide in this section various examples to demonstrate the steps of the proposed
algorithm for the general quadratic equation as demonstrated in Section 2.
Example 1: Consider the quadratic equation x 2 + 5x 6 = 0, where a1 = 5 and a0 =
6.

We note a12 4a0 > 0. Utilizing Equation (4) of step 2, x = 52 12 49 =

5
72 , from which the roots are 1 and 6.
2
Example 2: Let us solve another quadratic equation x 2 8x + 16 = 0 , where a1 =
= 4, so we have
8 and a0 = 16. We observe a12 4a0 = 0. From step 3, x = 0 8
2
two identical roots. Consequently, the left-hand side of x 2 8x + 16 = 0 is a perfect
square.
Example 3: We look at the solution of x 2 3x + 3 = 0, where a1 = 3 and a0 = 3.

We note a12 4a0 < 0. Using Equation (6) of step 4, x = 32 2i 4 (3) (3)2 =
3
2

i 3
,
2

which yields the two complex roots x =

3
2

i 3
2

and x =

3
2

i 3
.
2

5.2. Illustrations (cubic case)

In this section, we give several examples applying the steps of the proposed algorithm as
mentioned in Section 3, in order to compute the roots of the general cubic equation.
Example 1: Consider the cubic equation x 3 3x 2 + 3x 1 = 0.
In this example, a0 = 1, a1 = 3 and a2 = 3. Using step 1, A0 = 0 and A1 = 0. Uti= 1 . Consequently, the left-hand
lizing step 2, u3 = 0 and as a result, x = u a32 = 0 3
3
3
2
side of x 3x + 3x 1 = 0 is a perfect cube.

International Journal of Mathematical Education in Science and Technology 1101

Example 2: Consider another cubic equation x 3 6x 2 + 11x 6 = 0, where a0 =
2
= 1 = 0, using
6, a1 = 11 and a2 = 6. We obtain A0 = 0 and A1 = 11 (6)
3

a2
a2
a2
6
step 1. From step 3, x = u 3 = 0 3 = 2 and using x = 3 i a1 32 , the
other two roots are x = 1 or 3 .
Example 3: Let us solve the cubic equation x 3 + 3x 2 + 3x + 3 = 0, where a0 = 3, a1 = 3
1
and a2 = 3. We note A0 = 2 = 0 and A1 = 0 , using step 1. We obtain x = (2) 3 33 =
2.25992105 from step 4. Since we know one of the roots applying synthetic or long
division, we arrive at
x 2 + 0.74x + 1.3269999 = 0. By applying step 4 of Section 2, the two roots are
x = 0.37 + 1.090917i and x = 0.37 1.090917i.
Example 4: Suppose x 3 3x 2 = 0 , where a0 = 2, a1 = 3 and a2 = 0.
We get A0 = 2 = 0 and A1 = 3 = 0 using step 1. Applying step 5, c0 =
13

2 1
4(3)3
2
A1
3
(2)

+
= 1. We then calculate u = c0 3c
= 1 3(1)
= 2 using
2
2
27
0
step 6. From step 1, we arrive at the root x = u a32 = 2. By applying synthetic or long
division, we obtain x 2 + 2x + 1 = 0. From step 3 of Section 2, we arrive at the identical
roots of x = 1.
Example 5: Consider x 3 7x 6 = 0, where a0 = 6, a1 = 7 and a2 = 0.
We note that A0 = 6 = 0 and A1 = 7 = 0 using step 1. Substituting the values
1
of A0 and A1 in step 5, we arrive at c0 = [3 1.92449i] 3 . To calculate the cube root
of c0 , we follow step 7(1) to obtain = tan1 (0.641496666) = 32.68 and r = 3.5642.

1 


1 
Then, step 7(2) yields c0 = r 3 cos 3 + i sin 3 = (3.5642) 3 cos 32.68
=
+ i sin 32.68
3
3
1

1.49997 + 0.288667i. This step implies co = (3.5642) 3 = 1.527521583. Using step

 6,
A1
0
= 3.
=
3.05505

3.
Using
Equation
(8)
of
step
1,
we
obtain
x
=
3

u = c0 3c
3
0
2
Applying synthetic or long division yields x + 3 x + 2 = 0, from which we arrive at the
two roots x = 1 and x = 2 , using step 2 of Section 2.
5.3. Illustrations (quartic case)
In this section, we apply the steps of the proposed algorithm for quartic case of Section 4
and provide some examples in order to compute the roots of the general quartic equation.
Example 1: Consider the quartic equation x 4 + x 3 7x 2 x + 6 = 0, where a0 =
6, a1 = 1, a2 = 7 and a3 = 1. We note that A0 = 5.8008 = 0, A1 = 2.625 = 0 and
A2 = 7.375 = 0 using step 1. Using step 5, B0 = 91.2593 and B1 = 41.3333. From
step 6,

c0 =

13

91.2593 1
4 (41.3333)
(91.2593)2 +

= 3.6667 0.57735i.
2
2
27
3

Next in step 7, using any one of the values of c0 , we obtain y = 7.3333 and z = 12.25,
respectively. Taking the positive value of = 3.5 (the negative value of also works)
from step 8, we get s =

A2 + 2
2

A1

7.375+(3.5)2 2.625
3.5
2

= 2.0625 and t =

A2 + 2 +
2

A1

1102

Classroom Notes

7.375+(3.5)2 + 2.625
3.5
= 2.8125 from steps 9 and 10, respectively. Step 11 yields the two
2
of u1 = 0.75 and u2 = 2.25 while step 12 provides the remaining two values u3 =

values
2.75
and u4 = 1.25. Finally, we arrive at the solutions x1 = 1, x2 = 2, x3 = 3 and x4 = 1
given in step 13 from the values of u1 , u2 , u3 and u4 of step 12.

Example 2: Consider the quartic equation x 4 + 3x 2 + 3x + 2 = 0, where a0 = 2, a1 =

3, a2 = 3 and a3 = 0. We obtain A0 = 2 = 0, A1 = 3 = 0 and A2 = 3 = 0 using step 1.
Using step 5, B0 = 5 and B1 = 11. From step 6,

c0 =

91.2593
2


1
2

(91.2593)2 +

4(41.3333)3
27

 13

= 1.5302 1.1511i. Step 7

then provides y = 3.0604 and z = 1.0604, respectively, using any one of the values
of c0 . We can take the positive or the negative value of . We choose = 1.0298
from step 8 and obtain s =
3
3+(1.0298)2 + 1.0298
= 3.4868
2
ues of u1 = 0.51489 +

A2 + 2
2

A1

3
3+(1.0298)2 1.0298
2

= 0.57358 and t =

A2 + 2 +
2

A1

from steps 9 and 10, respectively. Step 11 provides the two val0.55541i and u2 = 0.51489 0.55541i, while step 12 yields
the remaining two values u3 = 0.51489 + 1.7949i and u4 = 0.51489 1.7949i. In
step 13, we obtain the solutions x1 = 0.51489 + 0.55541i, x2 = 0.51489 0.55541i,
x3 = 0.51489 + 1.7949i and x4 0.51489 1.7949i from the values of u1 , u2 , u3 and
u4 .

Example 3: Consider the quartic equation x 4 x 3 x 2 x 2 = 0, where a0 = 2,

a1 = 1, a2 = 1 and a3 = 1. From step 1, A0 = 2.3242 = 0, A1 = 1.625 = 0
and A2 = 1.375 = 0. Using step 5, we obtain B0 = 6.0741 and B1 = 8.6667. From
step 6,

c0 =

91.2593
2


1
2

(91.2593) +
2

4(41.3333)3
27

 13

= 1.0327 1.7887i.

Taking c0 = 1.0327 + 1.7887i in step 7 provides y = 0.33333 + 3i and z = 1.25 + 3i,

respectively. Using = 1.5 + i, step 8 yields s = 1.0625. Using steps 9 and 10, we get
t = 2.1875, where the value of from step 8 is utilized. Step 11 yields the two values
of u1 = 0.25 i and u2 = 0.25 + i, while step 12 provides the remaining two values
u3 = 1.75 and u4 = 1.25. In step 13, we utilize the values of u1 , u2 , u3 and u4 to obtain
the solutions x1 = i, x2 = i, x3 = 2 and x4 = 1.

6. Conclusions
This paper demonstrated an efficient algorithm to calculate the roots of the general quadratic,
cubic and quartic equations. More importantly, this algorithm gives the understanding of
how the roots are exposed systematically. Our proposed algorithm is simple and effective
for the purpose of classroom teaching. This algorithm would allow the students to work
comfortably with limited supervision or even independently to solve problems in a course
assignment or project.

Acknowledgements
The authors are grateful to the referee for the suggestions made on the earlier version of the paper.

International Journal of Mathematical Education in Science and Technology 1103

References
[1] Weisstein EW. Cubic equation. Available from: http://mathworld.wolfram.com/CubicEquation.
html
[2] Literka. Example of using Cardanos formula for cubic equations. Available from: http://www.
literka.addr.com/mathcountry/algebra/exmpcubic.htm
[3] Knaust H. The Quartic Formula. Available from: http://www.sosmath.com/algebra/
factor/fac12/fac12.html
[4] Kulkarni RG. Solving cubic equations by polynomial decomposition. Int J Math Edu Sci Technol.
2010;23(4):105108.
[5] Suen RY. Roots of cubics via determinants. College Math J. 1994;25(2):115117.
[6] Luthar RS. Luddhars method of solving a cubic equation with a rational root. College Math J.
1980;11(2):107110.
[7] Ward KJ. Solving the cubic equation (algebra). Available from: http://www.trans4mind.com/
personal_development/mathematics/polynomials/cubicAlgebra.htm
[8] Tanton J. Available from: http://www.jamestanton.com/wp-content/uploads/2012/03/CoolMath-Newsletter_April-20131.pdf

Some new mean value theorems of Flett type

Chengguan Tan and Songxiao Li
Department of Mathematics, Jiaying University, Meizhou, China
(Received 22 August 2013)
In this paper, we give some new mean value theorems, which are generalizations of
Flett, Myers and Tongs theorems.
Keywords: Rolles theorem; mean value theorem; Fletts theorem
2000 Mathematics Subject Classification: Primary 26A06; Secondary 26A24

1. Introduction
Various mean value theorems are important tools in mathematical analysis. It is worth
mentioning the pioneering contributions of Fermat, Rolle, Lagrange, Cauchy, Taylor and
others. The famous mean value theorem of Lagrange was stated as follows.
Theorem A (Lagranges Theorem): Let f(x) be a real continuous function on [a, b] and
differentiable on (a, b). Then there exists a point c (a, b) such that,
f  (c) =

f (b) f (a)
.
ba

When f (a) = f (b), then the mean value theorem of Lagrange reduces to Rolles theorem,
which is another important result in mathematical analysis. The mean value theorem of
Lagrange has been generalized by many authors. In 1958, a variation of the mean value theorem of Lagrange was given by Flett, [1] and it was later extended in [24] and generalized
in [510].

Email: [email protected], [email protected]

http://dx.doi.org/10.1080/0020739X.2014.904527