Start with a single loop. Climb through the classic design strategies and the patterns senior engineers rely on, and finish with the algorithms inside today's databases, search engines and research papers. Watch each idea move, write it yourself in pseudocode that is run and checked for real, and measure how fast it is.
Your first problem, Largest Element, takes about five minutes. No account and no installs: progress stays in this browser. Prefer a plan? Follow the Learning Path.
Chapters 0–18, step-by-step simulations and a problem bank, checked in your browser, in eight levels. Grounded in A. Levitin, Introduction to the Design and Analysis of Algorithms, 3rd ed., then extended well beyond it.
Loading your saved progress…
Every Arena problem has a level. Climb in order or jump in where you are: each station says what you will be able to do once you finish it.
The Algorithm Atlas follows one problem at a time from the first idea anyone tries to the newest research result, with what each step fixed and where it runs today. Point at a station to read it; open it to see the whole ladder.
Loading the Atlas…
Read it, watch it, write it, test it, trace its history and check yourself. Every page links to the others, so you can switch whenever one way stops working for you.
Over 200 problems in eight levels. Write Forge Pseudocode, JavaScript or Python, or snap shuffled blocks into order; hidden tests run in your browser and count the operations your solution performs.
Run any Forge Pseudocode you write. Step through it line by line and watch its variables and arrays change, then send it to the Growth Lab, which runs it on larger and larger inputs and fits the curve.
Nineteen chapters, each built around Build Cards: the problem in one sentence, a story, a picture, the pseudocode, the math, and a check-yourself question.
Dozens of step-through simulations. Type your own input, step forward and back, and read why each step happened.
Evolution ladders for many problems, from the first idea to today's research, every claim with a source.
Multiple-choice questions with explanations and a flashcard mode, plus one printable page of sums, recurrences and complexities.
Also on hand: a 12-week study plan problem sets and two practice exams with full solutions a tested Python library that counts its own operations
Each algorithm is learned the same five ways, in the same order. If you get stuck at any step, go back one, usually to the picture.
Read the lesson's Build Card: the problem in one sentence, a real-life story, then the idea.
Lessons by chapterRun the simulation on your own input. Step forward and back; every frame explains what happened and why.
SimulationsAssemble the algorithm from shuffled blocks of pseudocode. Order and indentation are the whole puzzle.
Blocks modeWrite it from a blank page. The Arena runs your pseudocode, checks it against hidden tests and counts its operations.
Practice ArenaSolved it? Every problem ends with the harder follow-up an experienced engineer would ask next.
Levels 0–7Chapters 1–12 follow the classic design strategies in textbook order. Chapters 13–17 add the tools and habits of working engineers, and Chapter 18 reaches the research frontier.
Everyone has a favorite first door. Walk in through yours, then try the others: the understanding that lasts is the one you have seen from several sides.
< to ≤ and predict what fails.Anany Levitin, Introduction to the Design and Analysis of Algorithms, 3rd edition (Pearson, 2012). Chapters 1–12 follow its chapters and cite sections like “Levitin §4.2”. All explanations, examples and pseudocode here are written fresh for Algorithm Forge; get the book for the full treatment and exercises. The Atlas and Chapter 18 cite the original papers.
A free, open learning resource for anyone who wants to build algorithms from scratch. Source on GitHub.