Cardiff School of Computer Science and Informatics

Coursework Assessment Pro-forma

Module Code: CM2208
Module Title: Scientific Computing
Lecturer: Prof. Paul Rosin
Assessment Title: Solving one variable non-linear equations
Assessment Number: 2
Date Set: Monday 22nd March 2021 (week 8)
Submission date and Time: Monday 3rd May 2021 at 9:30am (week 11)
Return Date: Monday 24th May 2021

This coursework is worth 35% of the total marks available for this module. If coursework is
submitted late (and where there are no extenuating circumstances):

1. If the assessment is submitted no later than 24 hours after the deadline, the mark for the
assessment will be capped at the minimum pass mark;

2. If the assessment is submitted more than 24 hours after the deadline, a mark of 0 will be
given for the assessment.

Your submission must include the official Coursework Submission Cover sheet, which can be
found here:
https://docs.cs.cf.ac.uk/downloads/coursework/Coversheet.pdf

Submission Instructions
Description Type Name
Cover sheet Compulsory One pdf file [Student number].pdf
Report Compulsory One pdf file Report [Student number].pdf
Source code Compulsory One or more Matlab files Code [Student number].zip
packaged as a single zip file

Any code submitted will be run on a system equivalent to those available in the School laboratory
and must be submitted as stipulated in the instructions above.

Any deviation from the submission instructions above (including the number and types of files
submitted) will result in a mark of zero for the assessment.

Staff reserve the right to invite students to a meeting to discuss coursework sub-
missions

Assignment
Modify the code Newton.m which is provided on Learning Central and has been covered in class
during the sessions on solving one variable non-linear equations.

Note (applying to the following tasks):

1. This assignment does not involve writing a GUI (e.g. using AppDesigner).

1
 2. Your code needs to be able to handle functions that are not just polynomials (as demon-
strated in some of the examples below).
3. Use the code for Newton.m and Bisection.m that are provided on Learning Central as
the base versions for your solutions.
4. Set the maximum allowed number of iterations to 100.
5. Set the error tolerance to 1 × 10−5 .

Task 1 (5% weight): Newton.m uses this update equation:

f (pn−1 )
qn−1 = pn = pn−1 −
f 0 (pn−1 )

which we will denote the result of by qn−1 . Write a modified version of Newton.m to implement
Ostrowski’s method called Ostrowski.m, which improves the convergence rate of Newton’s
method by replacing the update with the following update equation:

f (pn−1 ) f (pn−1 ) − f (qn−1 )

pn = pn−1 − · .
f 0 (pn−1 ) f (pn−1 ) − 2f (qn−1 )

Task 2 (15% weight): Write a program visualiseConvergence1.m that uses Newton.m

and Ostrowski.m or modified versions of them. It should plot the given non-linear function
f (xi ) colour coded such that plotting f (xi ) blue indicates that initialising the root finding with
p0 = xi led to the root finding algorithm converging, whereas red indicates that p0 = xi led to
the root finding algorithm not converging within the maximum allowed number of iterations.
An example of the expected output is shown below for f (x) = xe−x .

Note:
1. Set the range of x to [−10, 10].
2. Sample the curve at 1000 points.

Task 3 (20% weight): Write a modified version of Newton.m called NewtonMulti.m to extract
multiple roots using the method of deflation. Deflation operates by solving for the first root r1
of f1 (x) = f (x) and then modifying the function to become f2 (x) = f1 (x)/(x−r1 ). This process
is repeated to extract the set of roots ri from fi (x) = fi−1 (x)/(x − ri−1 ). After extracting ri
from fi (x) this solution should be refined by updating the initialisation to p0 = ri and rerunning
Newton’s method on the modified function. (Note that at each iteration the function becomes
progressively more complicated.) Plot the function and the positions of the roots as shown in
x5 x4 x3 2
the example shown below where f (x) = 1024 + 3256 − 564 − 1516x + x + 12.

2
 Note:
1. Plot output in the range of x in [−10, 10].
2. Look for a maximum of 10 roots. If less than 10 roots can be found then (automatically)
stop the root finding process.

Task 4 (10% weight): Write a modified version of Bisection.m called BisectionInitialise.m

that improves the initialisation of the Bisection Method. Given the initial range [xmin , xmax ]
repeatedly subdivide this range to find a range that is suitable for the Bisection Method; i.e.
f (xmin ) and f (xmax ) should have opposite signs. To be efficient first test if f (xmin ) and f (xmax )
have suitable signs, and if not then test the midpoint f ( xmin +x 2
max
) against f (xmax ). Repeat
this process for each of the subintervals until a suitable initialisation is found or a pre-specified
number has been considered.
Note:
1
1. Stop subdividing the intervals once they have been reduced to 220
th of the original range.

Task 5 (10% weight): Combine your methods developed in Tasks 1 and 4 to produce a more
effective root finding algorithm that 1/ does not require careful specification of the bound-
ing range for the root, and 2/ is efficient. Your program RootFindingImproved.m first runs
BisectionInitialise.m followed by 5 iterations of Bisection.m, to produce a solution that
is refined by Ostrowski.m. This program just needs to find a single root.

Task 6 (30% weight): Write a program visualiseConvergence2.m that uses Newton.m

and Ostrowski.m or modified versions of them. Whereas the previous tasks assumed that
the function was real valued, this task will use complex values for x and the values of f (x).
Consider x = a + bi, Newton.m can be run without change. Create a visualisation of the basins
of convergence for the root finding methods using x0 = a + bi over a range of a and b values
used for initialisation.
Consider 1/ the values of the roots and 2/ the number of iterations n, required to find a root.
For the former, assign a unique colour to each of the roots detected and allocate this colour to
each pixel that converges to this root. For the latter, perform log mapping, i.e. log(n), and
rescale to the full dynamic range. Create images for these two elements of visualisation, and
output the result of blending the two images with equal weight (i.e. average).
An example of the expected output is shown below for f (x) = (x3 − 1)(x2 + 2)e−x and
f (x) = x16 − 1.

3
Note:
1. Set the range of x to [−1, 1].
2. Set the sampling rate of a and b such that the image dimensions are 400 × 400.
3. If the root finding method does not converge then set the colour to mid gray.
4. When checking for duplication of roots use a high tolerance threshold of 1 × 10−3 to avoid
proliferation of roots.
5. The colours do not need to be consistent across different root finding methods.

Please note: You are allowed to use the Geometric Processing toolbox introduced in the module
and third-party libraries, as long as you clearly reference the sources in your report.

You must supply a report on your submission which provides a short written description (1–2
pages of text plus diagrams, screenshots etc.) conveying all the appropriate information to
demonstrate its operation.

Learning Outcomes Assessed

1. Demonstrate an awareness of basic Scientific Computing with MATLAB

2. Demonstrate an awareness of basic Numerical Analysis Techniques with MATLAB

4
Criteria for assessment
Credit will be awarded against the following criteria.

1. 5% – Ostrowski’s method (Task 1)

2. 15% – Visualise convergence in 1D (Task 2)
3. 20% – Extract multiple roots using Newton’s method with deflation (Task 3)
4. 10% – Improved initialisation of the Bisection method (Task 4)
5. 10% – Combined method for robust and effective root finding (Task 5)
6. 30% – Visualise convergence in complex plane (Task 6)
7. 10% – Report describing the operation of your program and your extension of the basic
algorithm

and the amount of credit will be awarded according to the following indicators:

1. 1st: the submission fully addresses the stated requirement for the Part,
as well as meeting the excellence indicators below.
2. 2.1: the submission fully addresses the stated requirement for the Part,
but has weaknesses in terms of the weakness indicators below.
3. 2.2: the submission partially addresses the stated requirement for the Part.
4. 3rd: the submission minimally addresses the stated requirement for the Part.
5. Fail: the submission does not adequately address the stated requirement for the Part.

Factor Weakness indicator Excellence indicator

Approach Does not adopt a professional Adopts appropriate methods with
or defensible approach full justification for the choices made
Insight and Little or no insight Has developed considerable
understanding and understanding insight and understanding

Feedback and suggestion for future learning

Feedback on your coursework will address the above criteria. Feedback and marks will be
returned on Monday 24th May 2021 via email.