Skip to main content
CBE NEXUS • GLOBAL MATHEMATICS

Applied and Specialist Mathematics

Two progressive pathways covering engineering, finance, actuarial science, economics, computing, data science, mathematical physics, control systems, decision science and environmental or biological modelling.

How the two levels differ

Undergraduate: develops mathematical foundations, standard analytical and computational techniques, model formulation, interpretation and applied problem solving.

Advanced postgraduate: develops deeper theory, proof-based reasoning, stochastic and optimisation methods, advanced computation, research literature skills and independent model evaluation.

These are broad international topic frameworks, not a claim that every university teaches every topic in this exact order. Institutions differ in prerequisites, course titles, depth, software and assessment. Match formal study to the current programme handbook and module descriptors.

Part A — Undergraduate Applied and Specialist Mathematics

20 core areas for bachelor's degree study and applied specialisation

01

Engineering mathematics and dimensional analysis

Units, dimensions, vector quantities, scaling, error propagation and dimensional consistency in engineering models.

02

Calculus for engineering systems

Differentiation and integration for rates, work, energy, optimisation, transient response and engineering design.

03

Linear algebra for data and engineering

Vectors, matrices, eigenvalues, least squares, systems of equations and matrix methods in engineering and data analysis.

04

Differential equations in physical systems

First- and second-order ODEs, coupled systems, initial/boundary conditions and introductory PDE models.

05

Numerical methods and simulation

Root finding, interpolation, numerical differentiation/integration, ODE solvers, stability, error and computational simulation.

06

Statistics for quality control and experiments

Descriptive and inferential statistics, design of experiments, regression, control charts, process capability and reliability.

07

Operations research and decision science

Model formulation, resource allocation, networks, inventory, queuing, scheduling and decision models.

08

Linear, integer and nonlinear programming

Feasible regions, simplex concepts, integer models, nonlinear constraints, sensitivity and optimisation software.

09

Financial mathematics and compound interest

Simple and compound interest, discounting, annuities, amortisation, effective rates and time value of money.

10

Investment mathematics and cash-flow analysis

Net present value, internal rate of return, bond and equity measures, portfolio returns and project cash flows.

11

Actuarial mathematics and risk

Survival models, life tables, premiums, loss distributions, reserves, risk pooling and introductory ruin concepts.

12

Economics mathematics and econometrics

Optimisation in economics, elasticity, comparative statics, regression, time series and causal inference foundations.

13

Computer science mathematics and algorithms

Asymptotic analysis, recurrence relations, algorithm correctness, complexity and mathematical foundations of computation.

14

Discrete structures, logic and graph algorithms

Sets, relations, combinatorics, Boolean logic, trees, networks, graph traversal and shortest-path methods.

15

Cryptography and elementary number theory

Divisibility, congruences, modular arithmetic, primes, Euclidean algorithm, RSA foundations and secure computation concepts.

16

Data science, probability and statistical learning

Random variables, distributions, estimation, regression, classification, validation, uncertainty and statistical learning foundations.

17

Mathematical physics and modelling

Vector calculus, conservation laws, dimensional models, oscillations, waves, fields and mathematical formulation of physical systems.

18

Control theory and systems mathematics

Linear systems, state-space models, feedback, stability, transfer functions and introductory controller design.

19

Game theory and decision analysis

Strategic-form games, Nash equilibrium, sequential decisions, utility, uncertainty, bargaining and multi-criteria decisions.

20

Environmental, biological and population modelling

Compartment models, population growth, epidemiological dynamics, ecological interactions, parameter estimation and sensitivity.

Part B — Advanced Postgraduate Applied and Specialist Mathematics

20 advanced extensions for Master's, doctoral coursework and research preparation

01

Advanced engineering mathematics and asymptotic methods

Advanced dimensional analysis, perturbation methods, asymptotic expansions, tensor methods and model reduction for coupled engineering systems.

02

Advanced calculus and variational methods for engineering

Vector and tensor calculus, integral theorems, calculus of variations, constrained functionals and optimal design.

03

Advanced computational linear algebra

Spectral theory for matrices, conditioning, sparse systems, Krylov methods, SVD, matrix perturbation and large-scale inverse problems.

04

Advanced differential equations and mathematical physics

Weak formulations, boundary-value problems, Green functions, nonlinear systems, bifurcation and analytical properties of PDE models.

05

Numerical analysis, high-performance simulation and uncertainty

Finite difference/element/volume methods, convergence and stability proofs, adaptive meshes, Monte Carlo simulation and uncertainty quantification.

06

Advanced statistical quality engineering and experimental design

Multivariate process control, sequential monitoring, robust design, Bayesian experiments, reliability growth and industrial data streams.

07

Stochastic operations research and decision science

Markov decision processes, stochastic programming, queueing networks, robust decisions and dynamic optimisation.

08

Advanced optimisation: convex, integer and nonlinear methods

Duality, KKT conditions, decomposition, branch-and-bound, cutting planes, nonconvex optimisation and convergence analysis.

09

Quantitative finance and advanced interest-rate models

Term-structure mathematics, duration/convexity, stochastic discounting, arbitrage-free pricing and interest-rate dynamics.

10

Investment theory, portfolio optimisation and derivatives

Mean-variance theory, factor models, CAPM foundations, option pricing, risk-neutral valuation and cash-flow optimisation.

11

Actuarial risk theory and insurance mathematics

Collective risk models, credibility, survival analysis, reserving, ruin probabilities, reinsurance and solvency capital.

12

Advanced econometrics and mathematical economics

Identification, simultaneous equations, panel data, time-series processes, causal inference and equilibrium modelling.

13

Algorithmic mathematics and theoretical computer science

Randomised algorithms, complexity classes, approximation, optimisation algorithms, formal verification and computational number theory.

14

Advanced discrete mathematics and graph optimisation

Extremal and algebraic graph theory, network flows, matching, matroids, spectral graphs and approximation algorithms.

15

Modern cryptography and computational number theory

Finite fields, elliptic curves, lattices, complexity assumptions, public-key constructions and cryptographic proof techniques.

16

Statistical learning theory and advanced data science

Generalisation bounds, regularisation, Bayesian learning, high-dimensional statistics, causal machine learning and model uncertainty.

17

Advanced mathematical physics and inverse modelling

Operator methods, spectral problems, nonlinear waves, inverse problems, PDE-constrained inference and parameter identifiability.

18

Nonlinear control and dynamical systems

Lyapunov stability, nonlinear feedback, controllability, observability, optimal control and robust/adaptive control.

19

Advanced game theory and mechanism design

Bayesian and evolutionary games, repeated games, auctions, mechanism design, information asymmetry and equilibrium refinements.

20

Advanced environmental, biological and population dynamics

Reaction-diffusion systems, spatial ecology, infectious-disease dynamics, optimal harvesting, climate-risk models and parameter inference.

Part C — Undergraduate applied mathematics problem bank

These are original, calculation-led university practice problems. Show equations, units, assumptions and interpretation. Unless a question states otherwise, give exact values before rounding and justify every modelling choice.

Engineering, calculus and linear algebra

  1. Dimensional analysis (10 marks). A pendulum period is modelled as T = kLagb, where L is length and g is gravitational acceleration. Use dimensions to determine a and b, and state the dimensions of k. Explain what dimensional analysis cannot determine about k.
  2. Engineering optimisation (10 marks). An open-top rectangular tank must hold 12 m3. Its base has length twice its width. Express the surface area in terms of the width w, differentiate, and determine the dimensions that minimise material area. Verify that the stationary point is a minimum.
  3. Linear systems and least squares (10 marks). Solve the overdetermined system represented by A = [[1,0],[1,1],[1,2]] and b = [1,2,2]T in the least-squares sense. Calculate the residual vector and explain its orthogonality to the column space of A.
  4. Eigenvalue stability (10 marks). For x′ = Ax with A = [[-2,1],[0,-3]], find the eigenvalues and general solution. Classify the equilibrium at the origin and explain the long-term behaviour of each component.
  5. Forced differential equation (10 marks). A damped oscillator satisfies y″ + 4y′ + 13y = 10cos(2t), with y(0)=0 and y′(0)=0. Find the complementary solution, construct a steady-state particular solution, and use the initial conditions to determine the constants.
  6. Numerical root finding (8 marks). Apply Newton's method to f(x)=x3−2x−5, starting at x0=2. Perform three iterations, tabulate xn and f(xn), and estimate the absolute error using the last two iterates.

Statistics, optimisation and finance

  1. Regression and inference (10 marks). For observations (1,2), (2,3), (3,5), (4,4), fit y = a + bx by least squares. Calculate a, b and R2; interpret the slope and discuss why a high R2 does not establish causation.
  2. Linear programming (10 marks). Maximise P = 5x + 4y subject to 2x+y≤8, x+2y≤8, x≥0, y≥0. Find all feasible vertices, evaluate P at each, identify the optimum and state which constraints are binding.
  3. Compound interest and amortisation (8 marks). A loan of KSh 480,000 is repaid by equal monthly payments over 36 months at a nominal annual rate of 12% compounded monthly. Derive the monthly payment formula, calculate the payment, and determine the total interest paid.
  4. Portfolio risk (8 marks). Two assets have expected returns 8% and 13%, standard deviations 10% and 20%, and correlation 0.25. For a portfolio investing 60% in asset 1 and 40% in asset 2, calculate expected return and standard deviation. Explain the contribution of covariance to portfolio risk.

Computing, networks and mathematical modelling

  1. Algorithmic complexity (8 marks). A nested algorithm performs, for each i=1,…,n, an inner loop of j=1,…,i. Derive the exact number of inner-loop executions and its asymptotic order. Compare it with an algorithm performing n log2n operations.
  2. Graph theory and shortest paths (8 marks). A directed network has edges A→B (4), A→C (2), C→B (1), B→D (5), C→D (8), and D→E (2). Use Dijkstra's algorithm from A to find the shortest path and distance to E. Show the settled-node order.
  3. Modular arithmetic (8 marks). Solve 7x ≡ 5 (mod 26). Find the multiplicative inverse used, give the least non-negative solution and verify it by substitution.
  4. Population dynamics (10 marks). A population follows dP/dt = 0.4P(1−P/1000). Solve the logistic equation for P(0)=100, determine the time at which P reaches 500, and interpret the carrying capacity.
  5. Compartment model (10 marks). A two-compartment tracer model is x′=−3x+y, y′=2x−2y, with x(0)=1 and y(0)=0. Find the eigenvalues, determine whether total tracer x+y is conserved, and explain the physical meaning of the result.

Hidden worked answers — reveal after attempting

Select “Show answer” to reveal the solution; select again to hide it.

Show answer 1

b=−1/2, a=1/2; T=k√(L/g), k dimensionless. Dimensional analysis cannot determine k (2π for the small-angle pendulum).

Show answer 2

h=6/w²; for an open-top tank S=2w²+6wh=2w²+36/w. S′=0 gives w³=9, so w=∛9 m, length 2w, height 6/w²; S″=4+72/w³>0.

Show answer 3

AᵀA=[[3,3],[3,5]], Aᵀb=[5,7]ᵀ; least-squares solution (a,b)=(2/3,1). Residual (1/3,1/3,−2/3)ᵀ and Aᵀr=0.

Show answer 4

Eigenvalues −2,−3; x₂=C₂e⁻³ᵗ, x₁=C₁e⁻²ᵗ−C₂e⁻³ᵗ. Origin is an asymptotically stable node.

Show answer 5

y_c=e⁻²ᵗ(C cos3t+D sin3t). Particular y_p=Acos2t+Bsin2t with A=90/145, B=80/145. Use y(0)=0 and y′(0)=0: C=−A, D=(2C−2B)/3.

Show answer 6

Newton iterates: x₁=2.1, x₂≈2.094568121, x₃≈2.094551482; root ≈2.09455148.

Show answer 7

x̄=2.5, ȳ=3.5, Sxx=5, Sxy=4; slope 0.8, intercept 1.5, R²=0.64. Association is not proof of causation.

Show answer 8

Vertices (0,0),(4,0),(0,4),(8/3,8/3); P values 0,20,16,24. Maximum 24 at (8/3,8/3); both constraints bind.

Show answer 9

i=.01,n=36; M=4800/[1−1.01⁻³⁶]≈KSh 15,929. Total interest ≈KSh 93,444 (rounding may vary).

Show answer 10

Expected return 10%; variance .0124 and standard deviation √.0124≈11.14%. Covariance term is 2w₁w₂ρσ₁σ₂.

Show answer 11

Count Σi=n(n+1)/2, hence Θ(n²); n log₂n is asymptotically smaller.

Show answer 12

Shortest path A→C→B→D→E; distance 2+1+5+2=10. Settled order A,C,B,D,E.

Show answer 13

7⁻¹≡15 mod26; x≡15·5≡23 mod26; 7·23≡5 mod26.

Show answer 14

P(t)=1000/[1+9e⁻⁰·⁴ᵗ]; P=500 at t=ln9/.4≈5.493. Carrying capacity 1000.

Show answer 15

Eigenvalues −1,−4; stable. (x+y)′=−x−y, so tracer total decays as e⁻ᵗ and is not conserved.

Part D — Advanced postgraduate applied and specialist problem bank

These problems require proof, derivation, computational design or critical model evaluation at Master's/doctoral level. State regularity and feasibility assumptions explicitly; numerical evidence alone is not a proof.

Advanced analysis, computation and optimisation

  1. Perturbation analysis (12 marks). For εy″+y′+y=0 on 0≤t≤1 with y(0)=1 and y′(0)=0, derive the outer expansion through O(ε). Identify the boundary layer needed to satisfy the derivative condition and construct a uniformly valid first-order approximation.
  2. Conditioning and SVD (12 marks). Let A = [[1,1],[1,1+δ]], δ>0. Derive det(A), explain why the system becomes ill-conditioned as δ→0, and obtain the leading-order behaviour of the smallest singular value. Discuss the effect on least-squares parameter uncertainty.
  3. Finite-difference stability (12 marks). Apply forward Euler in time and centred differences in space to ut=κuxx. Use a Fourier-mode analysis to derive the stability restriction on r=κΔt/(Δx)2. Explain why consistency alone is insufficient for convergence.
  4. Constrained optimisation (12 marks). Minimise f(x,y)=x2+2y2 subject to x+y=3 and x≥0,y≥0. Derive the KKT conditions, solve the problem, and verify constraint qualification and complementary slackness.
  5. Integer programming (12 marks). A facility may be opened at either site A or B, with binary variables zA,zB. Opening costs are 40 and 30; each unit of demand costs 5 from A and 8 from B. Formulate a mixed-integer model for demand 10 with exactly one facility open, capacity 12 at either site, and prove which facility is optimal.

Stochastic systems, finance and actuarial science

  1. Markov decision process (12 marks). A two-state discounted decision process has actions with specified one-step costs c(s,a) and transition matrix Pa. Write the Bellman optimality equations for discount factor β∈(0,1), state conditions for a contraction, and outline a policy-iteration algorithm. Explain how you would verify the computed policy.
  2. Option pricing (12 marks). In a one-period binomial model, S0=100, u=1.2, d=0.9, strike K=105 and risk-free gross return R=1.05. Compute the risk-neutral probability, call payoff in each state and no-arbitrage call value. Check the no-arbitrage condition d<R<u.
  3. Ruin probability (12 marks). In a classical risk model with initial reserve u, premium rate c and claim process N(t), derive the surplus process. State the net profit condition and formulate the integro-differential equation for ultimate ruin probability under an explicitly stated claim-size distribution.
  4. Bayesian experimental design (12 marks). Let observations satisfy Y|θ ~ Normal(θ,σ²) with known σ² and prior θ~Normal(μ₀,τ₀²). Derive the posterior distribution after n independent observations and determine how posterior variance changes with n. Propose a design criterion for selecting the next observation precision.
  5. Time-series econometrics (12 marks). For yt=φyt−1+εt, with white-noise innovations of variance σ², derive the stationarity condition, unconditional variance and autocorrelation function. Explain the consequences of fitting this stationary model when |φ|≥1.

Control, networks and scientific modelling

  1. Lyapunov stability (12 marks). For x′=Ax with A=[[-1,2],[-3,-4]], determine whether A is Hurwitz. Solve ATP+PA=−I for symmetric P and use V(x)=xTPx to establish asymptotic stability.
  2. Game theory (10 marks). Two firms choose high or low output. Their payoff matrix is [[(8,8),(2,10)],[(10,2),(4,4)]], with rows and columns ordered High, Low. Identify best responses, pure Nash equilibria and whether either firm has a dominant strategy.
  3. Reaction-diffusion stability (12 marks). For ut=D uxx+f(u), linearise about a spatially uniform steady state u*. Derive the growth rate of a Fourier perturbation and state the condition under which diffusion stabilises or destabilises a mode.
  4. Inverse problem and identifiability (12 marks). Observations satisfy y(t)=a exp(−kt)+b. Determine which parameters are structurally identifiable from ideal continuous noiseless data, identify degenerate cases, and propose a sensitivity-based experiment to distinguish a from b.
  5. Uncertainty quantification (12 marks). A model output is Q=g(X₁,X₂) with correlated inputs. Derive a first-order variance approximation using the gradient and covariance matrix. Explain when a Monte Carlo estimate is preferable and design a convergence diagnostic.

Hidden worked answers — reveal after attempting

Select “Show answer” to reveal the solution; select again to hide it.

Show answer 1

Outer solution begins y₀=e⁻ᵗ, y₁′+y₁=−e⁻ᵗ. Fast layer τ=t/ε has decaying e⁻ᵗ/ε mode to satisfy y′(0)=0; match inner/outer expansions for a uniform composite approximation.

Show answer 2

det A=δ; as δ→0, σmin≈δ/2, σmax≈2, κ₂≈4/δ. Data perturbations are amplified and parameter uncertainty grows.

Show answer 3

Fourier amplification G=1−4r sin²(kΔx/2); |G|≤1 for all modes gives 0≤r≤1/2. Consistency without stability does not ensure convergence.

Show answer 4

KKT: 2x+λ−μ=0, 4y+λ−ν=0, x+y=3, μx=νy=0. Solution x=2,y=1, objective 6; strict convexity gives unique minimum.

Show answer 5

Binary choice: A cost 40+5(10)=90; B cost 30+8(10)=110. Open A; capacity 12 covers demand 10.

Show answer 6

Bellman: V(s)=minₐ[c(s,a)+βΣP(s′|s,a)V(s′)]. β<1 gives contraction in sup norm; policy iteration evaluates then improves until unchanged.

Show answer 7

Call payoffs 15,0; q=(1.05−.9)/(1.2−.9)=.5; price=(.5·15)/1.05≈7.1429. No-arbitrage .9<1.05<1.2.

Show answer 8

Surplus U(t)=u+ct−ΣXᵢ; net profit c>λE[X]. For exponential claims density βe⁻ᵝˣ, cψ′=λ[ψ−∫₀ᵘψ(u−x)βe⁻ᵝˣdx−e⁻ᵝᵘ].

Show answer 9

Posterior precision τₙ⁻²=τ₀⁻²+nσ⁻²; mean τₙ²(μ₀/τ₀²+Σyᵢ/σ²). Variance decreases with n; choose observations by expected information gain or posterior variance reduction.

Show answer 10

AR(1) stationary iff |φ|<1; variance σ²/(1−φ²), ACF φᵏ. For |φ|≥1 stationary AR inference is invalid.

Show answer 11

Eigenvalues of A have real part −2.5, so Hurwitz. Solving Lyapunov gives P=[[.35,.05],[.05,.15]], positive definite; V=xᵀPx proves asymptotic stability.

Show answer 12

Low is strictly dominant for each firm; unique Nash equilibrium (Low,Low), payoffs (4,4).

Show answer 13

Linear growth rate σ(k)=f′(u*)−Dk². Positive diffusion suppresses high-frequency modes; scalar diffusion stabilises rather than destabilises a stable uniform state.

Show answer 14

For a≠0,k>0, b=lim y(t), a=y(0)−b, k=−d/dt ln(y−b); degenerate a=0 makes k unidentifiable, k=0 identifies only a+b. Sample early and late times.

Show answer 15

Delta method Var(Q)≈∇g(μ)ᵀΣ∇g(μ). Use Monte Carlo for nonlinear/non-Gaussian/tail cases; monitor batch standard errors and stability as sample size increases.

\n

Study and assessment expectations

Official Education References

Use these official organisations for curriculum guidance, examinations and further learning information.