eduKate Sengkang · LEARNING HALL · ALGORITHMS LEARNING ROADMAP
A method becomes useful when you can see it, rebuild it, test it and choose it for the right reason.
Trace → Reconstruct → Design → Justify → Evaluate → Transfer
This is the learner-facing progression map. It helps a student choose the next level of algorithmic work without duplicating the technical estate. The Algorithms & Computing Hub owns the field map, and the Complete Algorithms & Computing Index owns the recovered registry.
What This Hub Owns
An algorithm is a controlled procedure for moving from a stated problem to an output. Computational thinking is the wider discipline of representing the problem, choosing what information matters, decomposing the work, preserving the right conditions, checking the result and deciding whether the method remains suitable when constraints change.
This hub owns the navigation and learning progression across that estate. It does not replace the individual articles, Mathematics, Science, MindOS, the Student Interface or professional computing practice. Those neighbouring rooms keep their own definitions and responsibilities. The hub connects them only where the learning route genuinely crosses.
The useful question is not only “What does this algorithm do?” It is also “What problem contract makes it appropriate, what invariant keeps it safe, what evidence shows it works, and what conditions would make another method better?”
Choose the Learning Stage, Not the Most Impressive Title
A research algorithm is not automatically the best starting point. Enter where the learner can make the next move independently.
1 · Trace
See inputs, outputs, changing state, decisions and stopping conditions. Work through small examples before hiding the process inside code.
2 · Reconstruct
Remove the worked answer. Rebuild the method from the problem contract, delayed memory, a trace, pseudocode or changed input.
3 · Design & Justify
Choose a strategy, state assumptions, build an invariant, reason about termination and complexity, and search for counterexamples.
4 · Evaluate
Compare methods under real workloads, memory limits, numerical error, concurrency, reliability, maintainability and operational risk.
The Technical Corridors
Each corridor opens a family of related problems. Use the representative entrances here, then move to the complete index for the full estate.
Foundations & Proof
Representation, pseudocode, recursion, complexity, invariants, induction, amortised analysis and counterexamples.
Search, Sort & Selection
Ordering, locating, partitioning, ranking and choosing under different time, comparison and memory constraints.
Data Structures
Trees, heaps, hashing, tries, compressed indexes, persistent structures and the cost of maintaining useful state.
Graphs, Paths & Flows
Connectivity, traversal, shortest paths, spanning trees, cuts, flows, matchings and changing networks.
Strings, Parsing & Automata
Pattern matching, grammars, suffix structures, compressed text indexes, finite automata and language processing.
Optimisation & Operations Research
Linear, integer, convex and combinatorial optimisation; assignments, scheduling, approximation and constrained choice.
Numerical & Scientific Computing
Linear algebra, root finding, integration, differential equations, transforms, approximation and numerical stability.
Probability, Sampling & Statistics
Random choices, streaming summaries, Monte Carlo methods, exact sampling, estimation and uncertainty-aware computation.
Geometry, Graphics & Vision
Geometric predicates, spatial indexes, paths, meshes, curves, raster operations, image analysis and robust visual computation.
Distributed Systems & Concurrency
Consensus, elections, clocks, snapshots, replication, mutual exclusion, fault models and concurrent progress.
Systems, Databases & Networks
Memory, storage, scheduling, recovery, joins, routing, congestion control and production performance.
Machine Learning & AI
Inference, learning, search, ranking, clustering, recommendation, control and sample-efficient optimisation.
Algebra, Number Theory & Coding
Exact arithmetic, polynomial and finite-field methods, cryptographic computation, error correction and symbolic structure.
Bioinformatics & Sequence Methods
Sequence alignment, phylogenetic reconstruction, RNA structure and computation applied to biological evidence.
Research & Specialist Methods
Advanced algorithms whose value becomes clearer after the foundations are stable: specialised decompositions, oracles, exact methods and frontier engineering.
How to Use One Algorithm Article
Do not read an algorithm article as a catalogue of impressive terms. First state the problem in plain language. Then identify the input, output, assumptions, changing state and stopping condition. Work through one small example. Name the invariant or preserved relationship. Search for a case that might break the method. Only after that should implementation details, complexity and performance engineering become the centre of attention.
Before
Predict the output, route or failure point before seeing the answer.
During
Track state changes and explain why each move remains legal.
After
Reconstruct without looking, vary the input and compare an alternative.
Return
Use the result to decide whether to practise, repair, prove or move on.
The Boundary Matters
Algorithms often cross other disciplines, but a connection does not erase ownership. A Fourier transform can be studied as an algorithm, a mathematical transformation, a signal-processing tool or part of a scientific instrument. Sequence alignment can be computational while the biological claim still belongs to Biology. Optimisation can be mathematical while the operational decision still belongs to the system using it.
This hub therefore routes by the learner’s job. Enter here when the job is to understand, reconstruct, design, prove, compare or implement a computational method. Enter the neighbouring hub when the job is to establish a scientific claim, learn a school subject, diagnose a learner state or operate a professional system.
Start Small Enough to See. Continue Far Enough to Judge.
The complete estate is large. The learner does not need to consume it in order. Choose one problem family, make the state visible, reconstruct one method, test a changed case and let the return determine the next route.
