MAST30028: Numerical and Symbolic Mathematics

two assignments worth 20% each and final exam worth 60%

Overall the lectures were very dry and it was rather difficult to learn anything from watching them the first time around. They contained a lot of non-important excess information that only really served to obfuscate what is actually important and make it harder to understand what was going on. The lecturers were only useful to me after having read the relevant textbook chapters and having spent a considerable amount of time attempting the assignment questions.

The lecturer was also quite dry in his delivery and wasn't able to intonate his voice in a way that emphasised what was important in the lectures which further complicated things.

Due to this I relied quite heavily on the textbook, which luckily is quite well written.

The exam was a bit of a curveball, the theory behind each of the questions in the exam was essentially the same as what was done in the past exams but the approach required and way the questions were asked were very different from the past exams.

There's another reply about this subject this semester I would like to address because I don't entirely agree with it:

The assignments were a bit heavy but they did a fantastic job at teaching the content, and they were worth 20% each so I feel like that justifies the time that was required to complete them.

Outputting everything in the right format and putting everything together was a bit tedious so I agree on that part, but it's also sort of just what creating a report entails and is a skill and of itself. It's also the exact sort of thing required in a lot of jobs anyway so
¯\_(ツ)_/¯

I also disagree entirely that Matlab isn't a good language, I think it's a fantastic language. It's easy to use, has a really nice dynamic environment, has just the right amount of syntax so that programming doesn't take forever and it's really clear what everything is doing, is really fast and also has plenty of ready functions available without having to import a whole bunch of stuff like in python.

I agree that some of the questions were a bit vague but honestly all you had to do was go to the labs or consultation and ask about it he would spend as much time explaining and clarifying it to you as was needed.

One thing I took particular notice of is that the commenter criticised the marking for requiring the use of vectorisation (rather than a bunch of for loops and if statements):
Vectorisation makes the program FAR faster and also condenses what would otherwise at times take 5-10+ lines of code and a large amount of effort into a single, easily understood line of code.
That's a pretty basic concept and important lesson that strikes at the heart of what you want to accomplish in programming and numerical computing.
Overall I'd sum it up at crappy teaching, great subject.

Yes, with screen capture etc.

Hailong Guo

Yes, 2 past exams available

3.5

two 1 hour lectures and one 2 hour lab class a week

2020 sem2

83

2 assignments (each worth 20%) and 2 x 90 minute exams (each worth 30%)

This is a pretty tough subject and keeping up to date is a necessity.

The first half of the subject is the numerical component taught by Steven Carnie. Here you start off learning how to use MATLAB and then you learn about how computers process numbers, error analysis and numerical stability applied to solving linear equations and ODE's.

The assignment is doable, although it is still pretty tough. The exam in the other hand was far too long to finish in time so everybody was marked out of 40 instead of 50.

The second half of the subject is the symbolic part taught by Jan de Gier. This part is considered easier, although the trade-off is that MATLAB doesn't crap itself anywhere near as often as Mathematica. Again you start off learning how to use Mathematica and then you look at writing short lines of code that do the same as Mathematica's inbuilt functions, then you look at cobweb plots and chaos associated with them, bifurcation diagrams and then finally solving ODE's symbolically. My favourite part to this was learning that the solution to:



Is:



Then you learn perturbation theory, boundary layer method and finally WKB approximation.

The assignment for the symbolic part isn't too hard, the exam is much easier than the MATLAB one, but again, Mathematica can play up on you.
With both exams, you're allowed to use lecture notes on the LMS and any code the lecturer has provided on the LMS. Make sure you know where they are so you can USE THEM in exams! Doing this will speed up the pace at which you complete the exams!

No

Assoc Prof Jan De Gier (Symbolic Component - Mathematica), Assoc Prof Steven Carnie (Numerical Component - MATLAB)

Yes (MATLAB component only), you're given exams for 2010 and 2011

5/5

C. Moler, Numerical Computing with Matlab, SIAM, 2004.

1 or 2 x 1 hour lectures, 1 x 3 hour computer lab and 1 x 1 hour computer lab

2012

H1 [80]