The Brachistochrone Problem

Subject: Mathematics SL
Examination Session: May 2020
Word Count:

1
1.0 INTRODUCTION

One of the most fascinating solved problems in mathematics is the Brachistochrone

Problem. It was one of the earliest problems posed in the calculus of variations, first
hypothesized by Galileo and later rediscovered in 1697 by Johann Bernoulli. The word
‘brachistochrone’ comes from the Greek words brachistos (the shortest) and chronos (time or
delay), meaning the curve of least time.  

In simple words, the brachistochrone puzzle asks the reader to find a line between two
points. According to Euclid’s first postulate, a straight line segment can be drawn by joining
any two points. The line segment drawn would naturally be the shortest distance path
between the two objects, but what if we wanted to find the shortest time between these two
points?

To illustrate this, take for instance a string with a bead threaded through it, in a
manner such that the bead can move freely from point A to B. For the sake of this example,
friction and drag forces are being disregarded. With a constant acceleration downwards, what
curve should the string be in to minimise the bead’s travel time?

                 

To carry out this investigation, I will first start with an exploration of the solution to
this problem, model the resulting brachistochrone curve and make a comparison of the travel
times of a straight line path and the brachistochrone curve by assigning numerical values. As
part of my investigation, I will construct a virtual model of the possible paths from A to B
and compare the travel times of an object across those paths by observation.

2.0 TRAVEL TIME BETWEEN TWO POINTS

2
 The bead starts at rest from A and is accelerated by gravity to B. For the sake of this
investigation, frictional forces have been disregarded. The first method of solving this
problem that comes to mind would be optimisation, after all, the travel time needs to be
minimised and this minimisation will take place over all possible paths from A to B. 
Let the point A be at the origin such that A(0,0) and the point B at B(x,y) where it is
assumed that a > 0. Therefore, y(0) = 0 and y(a) = b.

Lets define T to be the total travel time of the object,

T =∫ ds
For a curve y, the speed of the object at any point can be given by
ds
v=
dt
where ds denotes the change in distance of travel, and dt denotes the change in time.
Rearranging the term leaves
ds
dt=
v
ds
T =∫
v
to find a straight line (shortest distance) between A and B: for any curve, ds2 = dy2+ dx2,
(Pythagorean Theorem) such that
ds=√ d x2 + d y 2
rearrange the terms
2

√
ds= d x2∗ 1+( ( )) dy
dx
2
dy
ds=dx∗ 1+
√ ( )
dx

ds=√ 1+ y ' 2 dx

3
Assume ds=√ 1+ y ' 2 dx to be Equation A.
In order to derive a formula for the total travel time of an object from Point A to Point B, the
following will need to be considered. The velocity v(x) of the object at point (x, y) has already
been determined by y: the law of Conservation of energy dictates that sum of the kinetic and
potential energy of the object always remains constant. Since the object is at rest at (0,0) the
difference in the gravitational potential energy from (0,0) to (x, y) will be equal to mgy, m
being the mass of the object and g is the gravitational acceleration (U). This follows that
U =m∗gh
1
KE= m v 2
2
1
U =KE=m∗gh= mv 2
2
1
mv( x)2=mgy
2
giving
v 2=2 gy
v=√ 2 gy
Take v=√ 2 gy as Equation B.
Now, the length of the path from (0,0) to (x, y) is denoted as s(x). This is given by
Equation A as
x
s ( x )=∫ √ 1+ y '2 d x
0

Switch the space variable x from the velocity function to the time variable t. Velocity
is now denoted as
ds
v ( t )=
dt
ds 1
=
dt v( s)
if T is the total travel time,
T x
T =∫ dt =∫
0 0
( 2 1gy ) √1+ y '2
dx

(substitute Equation B above)

The Brachistochrone Problem can be formed as the minimization of the function
x
1+ y '2
T =F ( y)=∫ √ dx ,
0 √ 2 gy

4
in order to successfully minimise F(y), partial derivatives are considered:
√ 1+ y '2
i.e. f (x , y , y ')=
√ 2 gy
We use partial derivatives successfully differentiate a multi variable equation.
First, we differentiate with respect to y:
1+ y ' 2
∂
∂y
f ( x , y , y ' )=
−1
2
√ 2g
y 3 /2
∗1

similarly, differentiate with respect to y’:

1
∗1 −1
∂ √ 2 gy
f ( x , y , y ' )= ( y +1 ) 2 ∗2 y '
'2
∂ y' 2
∂ y'
f ( x , y , y ' )=
∂ y' √ 2 gy∗√ 1+ y ' 2
note that:
d ∂ ∂ d ∂ ∂
{
dx ∂ y
f ( x , y , y' ) =
∂y }
f ( x , y , y' ) ∵
dx
{ } ∂
dy
=
∂y
dx
∂ ∂
when we multiply both and with the constant √ 2 g , we get
∂ y' ∂y

−1 1+ y '2 d y'
2 y3
=
dx√ √ y (1+ y '2)
u ' v u' −v ' u
differentiate the RHS through the use of the quotient rule: f ( x ) = ; f ( x )=
v v2 ( )
' ' '2 2 ''
y '' √ y (1+ y ) y ∗y ( 1+ y ) +2 y y y
'2

−
d y' 1 2 √ y ( 1+ y ' 2)
=
dx √ y ( 1+ y '2
) y ( 1+ y '2 )

2 y' ' ( y ( 1+ y'2 ) ) − y '2 ( 1+ y '2 )−2 y y '2 y ' '

'2
d y' 2 √ y ( 1+ y )
=
dx √ y ( 1+ y '2
) y (1+ y ' 2)2

d y' 2 yy ' ' (1+ y '2 ) y '2 ( 1+ y '2 ) 2 y y '2 y ' '
= 3 /2
− 3/ 2
− 3 /2
dx √ y ( 1+ y '2
) 2( y ( 1+ y '2 )) 2 ( y ( 1+ y '2 ) ) 2( y ( 1+ y '2 ))

d y' y'' y '2

= −
dx √ y ( 1+ y '2
) √ y ( 1+ y '2
) 2 √ y3 ¿ ¿ ¿

5
 d y' y'' y '2
= −
dx √ y ( 1+ y '2
) √ y ( 1+ y '2
) 2 √ y3 ¿ ¿ ¿
Therefore, we get this equation:

−1 1+ y '2 y'' y'2

2 y3 √= −
√ y ( 1+ y '2) 2 √ y3 ¿ ¿ ¿
multiplying both sides with √ ( 1+ y '2) , we get

−1+ y '2 y ' ' y'2 y '2 y ' '

3/ 2
= − −
2y √ y 2 y 3/ 2 √ y (1+ y ' 2)
multiplying both sides by √ y , we get
−1+ y '2 y '2 y '2 y ' '
= y ' '− −
2y 2 y (1+ y ' 2)
rearrange the equation to have the terms with 2y as the denominator on the LHS
y '2 1+ y '2 y '2
2y
−
2y
= y ' ' 1−
(( 1+ y '2 ) )
y '2 −1− y ' 2 1+ y '2− y '2
2y
= y' '
((
1+ y ' 2 ) )
−1 y''
=
2 y 1+ y ' 2

−( 1+ y '2 )=2 yy ' '

y ' 2+ 2 y y '' +1=0
This quadratic equation denotes the vertical height with respect to the horizontal translation.
Multiplying both sides by y’ gives us
y ' 3+ 2 y y ' y '' + y '=0

Since the LHS of the equation above is the derivative of the function
y + yy ' 2
the function can be written as
d d
y + yy ' 2= C
dx dx
where C > 0. When integrating both sides we obtain
y + yy ' 2=C
yy ' 2=C− y
solving for y’ gives

6
 C− y
y '=
√ y
dy C− y
dx
=
y√
through a separation of variables, we obtain
y
dx=
√ C− y
dy

which can be integrated to give

y
x=∫
√ C− y
dy + D

for some constant D∈R. Substitute y=C sin2 t into the integral to solve for x using
substitution
note :dy=2 C sin t cos t dt

C sin2 t
x=∫
√ C−C sin2 t
∗2 C sin t cos t dt

sin t
x=∫ ∗2C sin t cos t dt
cos t
x=2 C ∫ sin2 t dt

x=2 C ∫( 1−cos2 2t ) dt =2 C [ 2t − sin24 t ]+ D

1
x=C (t− sin 2t )+ D
2
solve for y similarly to get

y=C ( 1−cos2 2t )
since the path has to pass through (0,0) D = 0. These are our final equations:
1
(
x (t)=C t− sin 2 t
2 )
1 cos 2 t
y (t)=C ( −
2 2 )

3.0 THE CYCLOID

7
 The equations derived above for the brachistochrone problem seem to be those of a
cycloid. A cycloid is the locus of a point on the rim of a circle rolling along a straight line,
studied and named by Galileo in 1599. It was heavily debated amongst 18th century
mathematicians, so much so that it was called the ‘Helen of Geometers’, and even alluded to
in Moby Dick. Despite its simple construction, this curve has complex and fascinating
properties.

The parametric equations of a cycloid can be constructed by studying the equations of a

circle. The equations for a circle are x=r cos θ and y=r sin θ, but in this case, we need to
allow for the rotation of circle clockwise with angle t, the following equation must be used:
x=−r sin t , y =r cos t
Since the cycloid moves in the positive x direction, this must be factored into its parametric
equation. Therefore, if t represents the number of radians the circle has moved, then
πr∗t
∆ x=2 =rt
2π
x=−r sin t+ rt
x=r ¿
With respect to y, the only change that needs to be made is that the centre of the circle
will no longer be (0,0) but rather at (r,r) so that the bottom of the cycloid is at the x-axis.
Therefore,
y=r−r cos t
y=r (1−cos t )
The parametric equations for a cycloid are x=r ¿ and y=r (1−cos t ), which are very similar
to the equations derived for the brachistochrone curve.

In order to explore this equation further, we have to find its differential equation. First, we
take the differentials of both these equations with respect to t

8
 dx d
= ¿
dt dt
dy d
= ¿
dt dt
Since
dy
∗dt
dy dt
=
dx dx
the derivative of a cycloid is
dy
=r ¿
dx
dy
=r ¿ ¿
dx
dy sin t
=
dx 1−cos t
when both sides are squared,
dy 2 sin 2 t
( )
dx
= ¿¿¿

1−cos 2 t
¿ ¿¿
¿(1+ cos t) ¿ ¿
1+ cos t
¿
1−cos t
as per y=r (1−cos t )
y
cos t=1−
r
y
dy2 ( r )
1+ 1−
( dx ) 1− 1− y
=
( r)
y
2−
r
¿
y
r
2− y
∗r
r
¿
y
dy 2 2 r− y
( )
dx
=
y

9
y ( y '2+1 ) =2 r
This was the equation that Johann Bernoulli used to solve this problem.
3.0 CALCULATION OF A PATH

Until now, the work done sought to explore the nature of the brachistochrone curve by
studying its derivation and parametric equations. Now, we can define a real curve using
points and investigate the time taken by an object when it follows this path as compared to a
simpler linear path.

We can investigate the cycloid curve between the points A and B, in which A is a
(0,0) and B is at (5,-5). Because y2 ≤ y1, the curve will end up being an inverted
brachistochrone. However, since it is more difficult to calculate the parametric equations for
such a curve, I will solve this issue by reflecting x2,y2 on the x axis. Therefore, we have to
look for the cycloid between the points (0,0) and (5,5), and then reflect it on the x axis to get
the brachistochrone curve.

Solve for this equation as follows:

5=r ¿

5=r ¿

5
r=
1−cos t

10
 5
5= ( 1−cos t)
¿

1−cos t=t−sin t

1−t=cos t−sin t

To find t, I graphed both functions using Desmos and found the intersection point

as per the graph, t ≈ 2.41201 . From here, one can easily solve for r using

5
r=
1−cos t

5
r= ≈ 2.8645864390264
1−cos 2.41201

To find the equations for the brachistochrone curve, we must reflect this curve across the
x-axis to get

x=2.86 ¿

y=−2.86 ¿

this yields

11
Now that a path has been drawn, we must establish if it is indeed the fastest time curve. In
order to do this, I will use the time equation we had made earlier
x
1+ y '2
T =∫ √ dx
0 √ 2 gy
a modification of y ( y ' 2+1 ) =2 r states:
dy 2 r− y
dx
= y'=
y √
2r − y
dx=dy∗
√ y
substituting this into the travel time equation gives
t
T=
1
∫ √ 1+ y '2 dy
√2 g 0 √ y
2r − y

T=
1
∫
√2 g 0
t
√ 1+
y
y

2r − y
∗1
dy

√ y

y +2 r− y

√
t
1 y
¿ ∫
√2 g 0 y∗2 r− y
∗dy
y

2r
¿
1
t

∫
√2 g 0 √ y
∗1
2r − y
dy

12
 t
1 2r
¿ ∫ √
√2 g 0 y (2r − y )
dy

This equation can be rewritten as

t
1 2r
¿ ∫
√
√2 g 0 −1 {( r− y )2−r 2 }
dy

Recalling that y=r (1−cos t ), we can notice that the denominator

{( r − y )2−r 2 }=( r−r+ r∗cos t )2−r 2
¿¿
¿ r 2 cos2 t−r 2
¿ r 2 (cos 2 t−1)
¿−r 2 sin2 t
The integral can now be simplified to
x
1 2r
¿ ∫
√
√2 g 0 −r sin 2 t
2

x
dy

1 √ 2 r dy
¿ ∫
√2 g 0 r sin t
solve the integral by substituting dy =r sint dt into the equation
t
1 √2 r ∗¿ r sin t dt ¿
∫
√ 2 g 0 r sin t
t
1
∫ √2 r∗dt
√2 g 0
r
T=
√ g
∗t

Using this equation, we can find the shortest travel time for any curve, given its gravitational
acceleration. This will work for the equations of the brachistochrone given that r, t are
variables of the equations.
Now, we can use this equation to determine the travel time for the brachistochrone curve
based on the parameters set. This curve will be defined based on Earth conditions, whereby

m
the gravitational acceleration 9.81 2 .
s
Calculate t
5=2.86 ¿
t=2.4136138662007

13
Using the travel time equation, we can calculate:
2.8645864390264
T=
√ 9.81
∗2.4136138662007

T =1.3042602375541
A comparison of this value to the travel time for a straight line path will be beneficial,
considering that we cannot calculate the time for all possible curves. The co-ordinates will
remain the same so that we can compare the time taken by both paths for the same distance.
For the calculation of a straight line path, one needs to know that
distance= √ 52 +(−5)2=√ 50
π
acceleration=g sin =6.9367175234401
4
1
d=v i t+ a t 2
2
vi =0 , so
1
√ 50= ∗6.9367175234401∗t 2
2
t=1.4278431229271 seconds
time difference is:
t line −t brachistochrone =1.4278431229271−1.3042602375541=0.123582885373
Δt
%Gain= ∗100
t linear
0.123582885373
¿ ∗100
1.4278431229271
¿ 8.6552 %
Although a 0.1 second time difference may seem marginal, as the distance increases, so will
the time increment. As seen, the brachistochrone path is roughly 8.7% faster than a normal
linear path, proving that it indeed is the fastest time path.
co-ordinates (x,y) timebrachistochrone timeline time difference time difference(%)
(0,0) to (5,5) 1.304 1.428 0.124
8.7 %
(0,0) to (1000,1000) 18.43 20.19 8.7 %
1.76
As seen in this table, as the distance increases the time increment increases as well. The
percentage increase in the time that the linear path takes remains 8.7%.
4.0 FURTHER APPLICATIONS AND CONCLUSIONS
The most defining factor of the brachistochrone problem is that it paved the way for
the development calculus of variations, which has become an integral component of modern
day mathematics and physics. Because of its many applications the real world, the

14
brachistochrone has become a fundamental aspect of design and engineering. It is used to
design rollercoasters, in surfing, and ski jumps.
With respect to this particular investigation, I have successfully found mathematical
evidence that the brachistochrone curve is the curve of fastest time. However, my
investigation could have been improved if I had made a physical model to test my
mathematical values. Through the course of this exploration, I have learned about the various
properties of the cycloid, advanced calculus, and about the strange nature of the
brachistochrone itself.

15
Bibliography
Weisstein, Eric W. "Euclid's Postulates." From MathWorld--A Wolfram Web Resource.
http://mathworld.wolfram.com/EuclidsPostulates.html

Wiki.Math.Ntnu.No, 2020, https://wiki.math.ntnu.no/_media/tma4180/2015v/calcvar.pdf.

Accessed 20 Feb 2020.

Physics.Usu.Edu, 2020,
http://www.physics.usu.edu/Wheeler/ClassicalMechanics/CMBrachistochrone.pdf. Accessed 20
Feb 2020.

16