DASE7034 Operational Research
DASE7034 Overview
- The University of Hong Kong
- First Semester, 2026-27
- Postgraduate course
- Mathematics
- Two-hour written examination
What this course asks you to do
Operational Research turns a decision into a model that can be inspected, solved and challenged. The course starts with the philosophy of model building: objectives, relevant data, assumptions, controllable and uncontrollable variables, problem boundaries, validation and implementation.
- Assessed by Two coursework reports worth twenty per cent each and a written examination worth sixty per cent.
- Core method Define the decision, state the assumptions, compute the solution and finish with a certificate.
- Hardest step Choosing the model boundary and the stopping or stability rule before beginning arithmetic.
What you get in the full DASE7034 Exam Bible
- 12 chapters, 48 pages — every examinable topic
- 29 original practice questions, each with a model answer in the Written Examination format
- 12 worked examples, solved the way DASE7034 marks
- 19 computed figures, redrawn from the course’s own numbers
- Exam-morning plan for the Written Examination (60%) — what to revise last, where marks are lost
Free on this page: assessment table, chapter map, one worked example, key terms, FAQ. The rest is in the full exam bible.
How DASE7034 is assessed
| Component | Weight | Format |
|---|---|---|
| Coursework 1: Game Theory | 20% | Individual report of about 4,000 words plus figures and tables |
| Coursework 2: Application of OR in Evacuation Planning | 20% | Individual essay of about 4,000 words plus figures and tables |
| Written Examination | 60% | Two-hour examination measuring overall subject knowledge |
The published weights total 100%. The coursework brief states a ten per cent daily late discount, while the course outline says late assignments are normally not accepted. Confirm the operative rule and the current examination instructions on Moodle.
Assessment structure
The segment widths reproduce the published weights and total 100%.
What other students are asking about DASE7034
- Where can I find lecture notes or study guides for DASE7034 at HKU?
- The DASE7034 study guide in AskSia Library covers Operational Research in 12 chapters, with the assessment table, a chapter-by-chapter topic map, key terms and a worked example, free to read on this page.
- Is there a final exam in DASE7034, and how much is it worth?
- Yes. In the offering this AskSia Library guide is built from, Written Examination carries 60% of the DASE7034 mark; the full split is Coursework 1: Game Theory 20%, Coursework 2: Application of OR in Evacuation Planning 20%, Written Examination 60%.
- What does DASE7034 cover on linear programming formulation and geometry?
- In AskSia Library's DASE7034 guide, the Linear Programming Formulation and Geometry chapter explains linearity, feasible regions, corner points, binding constraints and optimality.
- What does DASE7034 cover on simplex foundations and pivot logic?
- In AskSia Library's DASE7034 guide, the Simplex Foundations and Pivot Logic chapter explains standard form, basic feasible solutions, ratio tests and pivots.
- What does DASE7034 cover on simplex edge cases and artificial variables?
- In AskSia Library's DASE7034 guide, the Simplex Edge Cases and Artificial Variables chapter explains alternate optima, unboundedness, infeasibility, degeneracy and artificial variables.
Current DASE7034 dates
| Date | Item | Control |
|---|---|---|
| 30 November 2026 | Coursework 1 and Coursework 2 | Both Moodle assignment pages and the coursework brief state a 5 pm deadline. |
Dates are as published in the current coursework brief and Moodle assignment pages. Confirm exact deadlines and submission settings in the live LMS.
What DASE7034 covers
Twelve chapters move from model construction through linear and integer optimisation to network algorithms and queueing analysis.
Operational Research Models and Decisions
decision variables, objectives, constraints, boundaries, validation and implementationOperational Research begins by identifying a decision, the performance measure, the controllable variables, the relevant data and the boundary of the system.
- Decision framing and objectives
- Controllable and external variables
- Constraints and data requirements
- Model boundaries and assumptions
- Validation and implementation
Linear Programming Formulation and Geometry
linearity, feasible regions, corner points, binding constraints and optimalityLinear programming represents a linear objective over a feasible set defined by linear equalities or inequalities and variable domains.
- Linearity assumptions
- Feasible and infeasible solutions
- Graphing half-planes
- Corner-point optimality
- Binding constraints and slack
Simplex Foundations and Pivot Logic
standard form, basic feasible solutions, ratio tests and pivotsThe simplex method expresses LP geometry through bases and pivots. Standard form introduces slack or transformed variables, a basis identifies a basic solution and nonnegativity determines whether it is feasible.
- Standard-form transformations
- Basic and non-basic variables
- Basic feasible solutions
- Entering and leaving variables
- Pivot and optimality tests
Simplex Edge Cases and Artificial Variables
alternate optima, unboundedness, infeasibility, degeneracy and artificial variablesTerminal tableaus can describe alternate optimal solutions, unbounded improvement or infeasibility.
- Alternate optimal bases
- Unbounded improving directions
- Artificial starting variables
- Infeasibility certificates
- Degenerate pivots
Sensitivity Analysis and Shadow Prices
allowable ranges, reduced costs, shadow prices and local updatesSensitivity analysis describes local changes around an optimal basis.
- Objective coefficient ranges
- Constraint RHS ranges
- Reduced costs
- Shadow prices
- Limits of local analysis
Integer Programming Formulation
binary logic, linking constraints, valid bounds and discrete decisionsInteger programming represents indivisible counts, yes-or-no choices and logical relationships.
- Pure and mixed integer models
- Binary decision variables
- Logical implications
- Big-M linking constraints
- LP relaxation bounds
Relaxations, Integrality and Branch-and-Bound
LP bounds, incumbents, pruning, optimality gaps and network integralityBranch-and-bound partitions the integer search space while LP relaxations provide bounds. A feasible integer solution becomes the incumbent, and a node closes when its relaxation is infeasible, already integer or unable to improve the incumbent.
- Relaxation bounds
- Incumbent solutions
- Fractional branching
- Pruning rules
- Minimum-cost flow integrality
Maximum Flow and Minimum Cuts
capacities, residual networks, augmenting paths and cut certificatesMaximum flow sends the greatest feasible quantity from a source to a sink through capacitated arcs. Residual capacity records unused forward capacity and cancellable reverse flow, allowing augmenting paths to revise earlier choices.
- Source and sink structure
- Arc capacity and conservation
- Residual networks
- Augmenting path bottlenecks
- Maximum-flow minimum-cut certificate
Shortest Paths and Dijkstra's Algorithm
tentative labels, permanent nodes, relaxation and path reconstructionShortest-path models minimise additive arc cost. Dijkstra's algorithm applies when every arc cost is nonnegative: the smallest tentative label becomes permanent, its outbound arcs are relaxed and predecessor labels record the route.
- Shortest-path variants
- Tentative and permanent labels
- Outbound arc relaxation
- Predecessor backtracking
- Nonnegative-cost condition
Queueing Foundations and Little's Law
arrival and service rates, utilisation, Poisson processes and steady stateQueueing analysis separates arrivals, waiting, service and departures. A/B/s notation identifies inter-arrival and service-time distributions plus the number of servers.
- Queue anatomy and discipline
- Arrival and service rates
- Utilisation and stability
- Little's Law
- Poisson and birth-death processes
Single and Multi-Server Queueing Models
M/M/1, M/M/s, finite capacity, waiting and capacity planningM/M/1 assumes Poisson arrivals, exponential service, one server and a stable load; M/M/s extends the process to parallel servers and requires its own probability structure. Finite-capacity and general-service models change the state process again.
- M/M/1 steady state
- M/M/s pooled service
- Finite system capacity
- Waiting-cost trade-offs
- Model assumption checks
Queueing Variability and Jackson Networks
traffic equations, effective rates, variability and node stabilityOpen queueing networks combine external arrivals with routed customer flow. Traffic equations determine each node's effective arrival rate before node stability and waiting can be assessed.
- External and routed arrivals
- Traffic equations
- Node stability
- Jackson network decomposition
- Arrival and service variability
That beginning controls every later technique. A numerical optimum is useful only when the variables represent a real choice, the constraints represent real limits and the recommendation survives a check against the system.
From linear programming to integer decisions
Linear programming supplies the central language of objectives, feasible regions, binding constraints and corner-point optimality.
The simplex chapters turn that geometry into an algebraic method and show how a tableau diagnoses alternate optima, unboundedness and infeasibility. Sensitivity analysis then asks what can change without changing the current basis, separating objective-coefficient ranges, reduced costs, RHS ranges and shadow prices.
Integer programming adds binary and discrete decisions, logical constraints, LP relaxations and branch-and-bound certificates.
Networks and stochastic service systems
The network chapters compare two different optimisation structures. Maximum flow uses capacities, residual networks and augmenting paths, with a minimum cut as its optimality certificate.
Shortest paths use arc costs, tentative labels and predecessors, with Dijkstra's algorithm justified by nonnegative costs. Queueing begins with arrival and service rates, utilisation, Little's Law, exponential memorylessness, Poisson counting and birth-death processes.
It then develops single-server, multi-server and open queueing-network models, always testing stability before reporting a steady-state average.
Assessment control
The published outline assigns sixty per cent to a two-hour written examination and forty per cent to in-course assessment.
The current coursework brief resolves the latter into two individual reports worth twenty per cent each: one on game theory and one applying an OR technique to evacuation planning. Both reports are due at 5 pm on 30 November 2026. The brief and outline publish different late-submission rules, so the live Moodle setting should control.
The official course information for this offering does not publish the examination sections, question types, permitted materials or per-question allocations.
How to revise with this guide
For each chapter, identify the model family before reading the worked solution.
Recompute every intermediate number, then cover the answer and explain the certificate aloud: binding constraints for LP, a bound gap for integer programming, a cut for maximum flow, a reconstructed path for shortest paths, or stability and Little's Law for queueing. Finish by saying what evidence or assumption could change the recommendation.
That sequence rehearses both computation and interpretation without inventing a paper format the University has not published.
Worked example · free
Choose the correct certificate for an OR result
- 3Match each model family to its feasibility or assumption check.
- 4Name the model-specific optimality or stopping certificate.
- 3State the interpretation that closes the answer.
Key terms
- Decision Variable
- A controllable quantity whose value the model chooses.
- Feasible Region
- The set of all decisions that satisfy every constraint and domain rule.
- Binding Constraint
- A constraint that holds as an equality at the solution under discussion.
- Basic Feasible Solution
- A nonnegative basic solution corresponding to a corner of a standard-form LP.
- Reduced Cost
- The objective improvement required before a non-basic variable can enter an optimal basis.
- Shadow Price
- The local change in optimal objective per unit increase in a constraint RHS while the basis remains valid.
- LP Relaxation
- The model obtained by removing integer restrictions, used to provide a bound for an integer program.
- Incumbent
- The best feasible integer solution found so far in a branch-and-bound search.
- Residual Capacity
- The additional net flow an arc direction can carry after accounting for unused capacity and cancellable reverse flow.
- Minimum Cut
- A source-sink partition with the smallest total forward capacity across the partition.
- Utilisation
- The arrival rate divided by total service capacity for the queueing model.
- Traffic Equation
- A flow-balance equation that combines external and routed arrivals to obtain a node's effective rate.
DASE7034 FAQ
What is the published assessment structure?
The current outline assigns sixty per cent to a two-hour Written Examination and forty per cent to in-course assessment. The current coursework brief resolves the in-course portion into two individual reports worth twenty per cent each, one on game theory and one on evacuation planning.
What is known about the written examination?
The outline states that the Written Examination lasts two hours, carries sixty per cent and measures overall subject knowledge. The official course information for this offering does not publish sections, question types, permitted materials or per-question marks, so confirm those operational details on Moodle.
How should I choose an OR method?
Start from the decision structure. Continuous linear activities suggest LP; discrete openings or choices require integer variables; source-sink capacity suggests maximum flow; additive route costs suggest shortest paths; and random arrivals competing for service suggest queueing. Then verify each method's assumptions.
Why is a certificate part of the answer?
A numerical result is not self-verifying. Binding constraints and optimality tests support an LP, a bound gap supports an integer solution, a matching cut supports maximum flow, predecessor backtracking supports a shortest route, and stability plus Little's Law supports a queueing result.
Can I rely on the published late penalty?
The course documents conflict. The coursework brief states a ten per cent daily discount, while the course outline says late assignments are normally not accepted. Check the current Moodle setting before the deadline and do not plan submission timing around either interpretation.
Are the exercise questions past papers?
The outline says tutorials may go over past examination questions, but the course provides exercise sheets rather than a separately labelled past paper or sample final. This guide therefore uses original practice shaped by the taught methods and does not imply access to a past exam.
What should I check in a queueing answer?
Convert arrival and service quantities to the same time unit, distinguish queue measures from system measures, test the stability condition for the chosen model and use Little's Law as a cross-check. For networks, solve effective arrival rates before computing node delays.
Questions students ask in each DASE7034 chapter
- Why should implementation be considered before solving?
- A solution that depends on unavailable data, authority the decision maker does not hold or an action the organisation cannot execute is only a theoretical benchmark. Early implementation checks help choose variables and constraints that correspond to a usable decision. Chapter 01 · Operational Research Models and Decisions
- Why can an optimum be checked at corners?
- For a linear objective over a polyhedral feasible region, any existing optimum has at least one corner-point representative. The objective cannot form an isolated interior peak because its level sets are parallel lines or hyperplanes. Chapter 02 · Linear Programming Formulation and Geometry
- Why is the smallest positive ratio selected?
- Increasing the entering variable reduces current basic variables. The smallest eligible ratio identifies the first one to reach zero, so making that variable leave preserves nonnegativity and keeps the next basis feasible. Chapter 03 · Simplex Foundations and Pivot Logic
- What does a zero reduced cost mean at optimum?
- For an eligible non-basic variable, it means the variable can enter without changing the objective, potentially producing another optimal corner. Pivoting reveals a second endpoint of an optimal edge or confirms degeneracy. Chapter 04 · Simplex Edge Cases and Artificial Variables
- Do quantities stay fixed after an RHS change?
- Not necessarily. The basis can remain the same while the values of basic variables change with the RHS. The shadow price updates the objective locally but does not by itself provide the new quantity vector. Chapter 05 · Sensitivity Analysis and Shadow Prices
- Why is rounding a relaxed solution unsafe?
- Independent rounding can violate capacities, balances or logical implications, and even a feasible rounded point need not be the best integer solution. Integer optimisation uses bounds and search to preserve feasibility and certify quality. Chapter 06 · Integer Programming Formulation
- Which branches cover a fractional integer value?
- If an integer variable has relaxed value 2.5, use x less than or equal to 2 and x greater than or equal to 3. The branches cover every integer possibility and exclude the current fractional point. Chapter 07 · Relaxations, Integrality and Branch-and-Bound
- Why are reverse residual arcs needed?
- They allow later augmentations to cancel part of an earlier flow and reroute it. Without reverse capacity, a locally attractive first path can trap the algorithm below the true maximum. Chapter 08 · Maximum Flow and Minimum Cuts
- Why must arc costs be nonnegative for Dijkstra?
- A later path containing a negative arc could undercut a node already declared permanent. Nonnegative costs guarantee that extending paths through unselected nodes cannot reduce the current smallest tentative label. Chapter 09 · Shortest Paths and Dijkstra's Algorithm
- What is the difference between W and Wq?
- W is total expected time in the system and includes service, while Wq is expected waiting time before service. Under the basic model, their difference is mean service time. Chapter 10 · Queueing Foundations and Little's Law
- Can total service capacity replace mu in the one-server formula?
- No. In a multi-server system, service capacity depends on how many customers are present until all servers are busy. Use the M/M/s probability structure rather than substituting s times mu into M/M/1 waiting formulas. Chapter 11 · Single and Multi-Server Queueing Models
- Why are traffic equations solved simultaneously?
- A node can receive customers from several other nodes, including feedback loops. Its outgoing throughput affects other arrivals, so effective rates are linked and generally must be balanced together. Chapter 12 · Queueing Variability and Jackson Networks
How to study for the exam
Build a model-selection grid with five rows: linear programming, integer programming, maximum flow, shortest path and queueing. For every practice problem, fill the row before calculating: decision, objective, assumptions, algorithm and certificate. Recompute one worked example per study session and change one input to predict the direction of the result.
In LP, move an objective coefficient or RHS and state whether the basis is protected. In networks, change a capacity or arc cost and identify which path or cut may change. In queueing, change arrival rate or server count and test stability before estimating delay. Keep units visible on every rate and coefficient.
Finish each answer with one operational sentence naming the binding resource, bottleneck arc, route condition or capacity margin, plus the evidence that could overturn the recommendation. During the final revision pass, practise the published two-hour duration without assuming a section structure or permitted-material rule that the official course information for this offering does not state.
Your AI Mathematics tutor for DASE7034
Stuck on a hard DASE7034 question? Sia is AskSia’s AI Mathematics tutor — ask any DASE7034 Operational Research question and get a clear, step-by-step explanation grounded in how the course is actually taught and assessed. Read this whole study guide free, then take your hardest questions to Sia.