Postgraduate / Master's · Advanced Mathematics
Numerical Optimisation
What is it
Numerical optimisation studies algorithms for finding the minimum or maximum of a function, especially in high-dimensional or complex settings where analytical solutions don't exist — covering gradient-based methods, convergence guarantees, and constrained optimisation.
Why it matters
Nearly all modern machine learning is, at its core, a numerical optimisation problem — training any model means numerically minimising some loss function, making a genuine understanding of optimisation algorithms essential for research in machine learning and applied mathematics.
Exam tip
When an optimisation algorithm fails to converge, check the learning rate or step size first before assuming the algorithm itself is wrong — an inappropriately large step size is the most common practical cause of divergence in gradient-based methods.
Related topics
Want help mastering Numerical Optimisation?
Tell us about the student's goals and confidence — we'll design a personalised plan.