Computational Physics: Lecture – III

Graphical method

Finding the solutions of Algebraic and Transcendental equations

𝑓(𝑥) = 0 Algebraic equation

𝑓(𝑥) = 𝑔(𝑥) Transcendental equation

Algebraic equation

𝑓(𝑥) = (1 + 𝑥)! 𝑓(𝑥) = (1 + 𝑥)"

Root at 𝑥 = −1 Root at 𝑥 = −1

Transcendental equation

𝑥 " = 3𝑥 − 1 sin(𝑥) = 0.3 𝑥

The solutions to the equations are points where the two lines intersect.
Root finding (graphical method)

Consider the algebraic equation

𝑓(𝑥) = 𝑥 " − 2 𝑥 ! − 2 𝑥 + 2

For better estimation of the roots, the function should

be plotted in three different regions
𝑥# ± 𝜖, 𝑥! ± 𝜖, and 𝑥" ± 𝜖

One can do the same thing for transcendental

equations

Bisection method

Here, our aim is to find the roots of the function

𝑓(𝑥) in the interval [𝑥$ , 𝑥% ] with the following
conditions.

𝑥$ < 𝑥%

Either 𝑓(𝑥$ ) < 0 and 𝑓(𝑥% ) > 0

or
𝑓(𝑥$ ) > 0 and 𝑓(𝑥% ) < 0.

Such that 𝑓(𝑥$ )𝑓(𝑥% ) < 0

Find the midpoint between [𝑥$ , 𝑥% ]

𝑥$ + 𝑥%
𝑥& =
2

If 𝑓(𝑥$ )𝑓(𝑥& ) < 0 then the root is between [𝑥$ , 𝑥& ] otherwise the root is in between [𝑥& , 𝑥% ].

One need to continue this process, of finding the midpoints, until the condition |𝑓(𝑥& )| ≤ 𝜖,
where 𝜖 is user specified tolerance.
Example:

Consider the equation 𝑓(𝑥) = 𝑥 " − 𝑥 − 2.

First, we need to find the values of 𝑥$ and 𝑥% such that 𝑓(𝑥$ )𝑓(𝑥% ) < 0.

For the above function 𝑥$ = 1 and 𝑥% = 2 satisfy this criterion. 𝑓(1)𝑓(2) = −6 < 0

Since the above function is continuous, there must be a root within the interval [1, 2]
#' !
The midpoint is 𝑥& = !
= 1.5, and 𝑓(1.5)𝑓(2) < 0. So, the root is within the interval
[1.5, 2].

This process needs to be continued until |𝑓(𝑥& )| ≤ 𝜖.

Newtons – Raphson method (to speed up the Bisection method)

Let us say the function 𝑓(𝑥) has a root at 𝑥) . The Taylor expansion of the function is

𝑓(𝑥 + ∆) ≈ 𝑓(𝑥) + ∆ 𝑓 * (𝑥) + 𝑜(𝜀 ! )

where 𝑥 = 𝑥) − ∆

𝑓(𝑥) ) ≈ 𝑓(𝑥) + ∆ 𝑓 * (𝑥) + 𝑜(𝜀 ! )

Since 𝑓(𝑥) ) = 0, we get

𝑓(𝑥)
∆≈−
𝑓 * (𝑥)

For example, 𝑓(𝑥) = 𝐴 𝑥 − 𝐵

𝑓(𝑥) 𝐴𝑥−𝐵 𝐵
∆≈ − =− = − C𝑥 − D
𝑓 (𝑥)
* 𝐴 𝐴
𝐵
𝑥+∆=
𝐴

The function 𝑓(𝑥) has a root at

𝐵
𝑥) =
𝐴

The above example is simple.

Considers a nonlinear function, for example

𝑓(𝑥) = (𝑥 − 1)(𝑥 − 2)
"
To begin with, calculate ∆ at 𝑥 = +.
 3
𝑓 E𝑥 = 4G 5/16
∆≈ − =
3 3/2
𝑓 * E𝑥 = G
4

5
∆≈ = 0.20833
24

However, the exact value 𝑥) = 1 (2 is also a root but focus on the first root which is 1)
"
So, ∆ = 𝑥) − 𝑥 = 1 − + = 0.25. The values 0.25 and 0.20833 are close but not the same!

Here, the idea is to start at some location 𝑥, and try to step towards the root using ∆ to guide
the choice of step size.

For a function 𝑓(𝑥), using a starting point 𝑥# , we update 𝑥 according to

𝑓(𝑥# )
𝑥! = 𝑥# + J− K
𝑓 * (𝑥# )

In this step ∆ = 𝑥! − 𝑥#
𝑓(𝑥! )
𝑥" = 𝑥! −
𝑓 * (𝑥! )

After 𝑛 + 1 steps
𝑓(𝑥, )
𝑥,'# = 𝑥, −
𝑓 * (𝑥, )

We can terminate this iteration when the condition |𝑓(𝑥,'# )| < 𝜀, where 𝜀 is the user defined
tolerance.
Secant method

In the Newton – Raphson’s method we used

𝑓(𝑥, )
𝑥,'# = 𝑥, −
𝑓 * (𝑥, )

In the Secant method, we replace 𝑓 * (𝑥, ) with the following

𝑓(𝑥, ) − 𝑓(𝑥, − ℎ)
𝑓 * (𝑥, ) = lim
-→) ℎ

𝑓(𝑥, ) − 𝑓(𝑥,/# )
𝑓 * (𝑥, ) ≈
𝑥, − 𝑥,/#

Thus, we get

𝑥, − 𝑥,/#
𝑥,'# = 𝑥, − P Q 𝑓(𝑥, )
𝑓(𝑥, ) − 𝑓(𝑥,/# )