ABC
Analysis · Concept hubDeep story

Markov Chains

1906 CE20th-century Russia (Markov)

Through Markov Chains: How can repeated signals emerge from a single uncertain event?

Travel from dice and insurance to data, noise, and learning algorithms—measuring uncertainty and turning it into decisions.

This voyage is an editorial path for understanding, not a claim of direct historical influence or sole invention.

Understand it in one breath

"The current state contains the information needed for the distribution of the next state." Markov formalized the theory in 1906 and in 1913 analyzed vowel–consonant transitions in Pushkin. Not every chain converges. A finite irreducible, aperiodic chain converges to a unique stationary distribution regardless of its starting state. This structure appears in PageRank, speech recognition, and reinforcement learning.

At a glance

→ Next

Sunny

Cloudy

Rainy

Sunny

0.7

0.2

0.1

Cloudy

0.3

0.4

0.3

Rainy

0.2

0.3

0.5

Stationary distribution (convergence)

0.46

0.28

0.26

Repeated multiplication by the transition matrix reaches the same stationary distribution ≈ (0.46, 0.28, 0.26) regardless of the starting state — the ergodic theorem. PageRank performs exactly this calculation.

Concept

A stochastic process where the next state depends only on the present. Foundation of PageRank, GPT, speech recognition.

Key formula

P(Xn+1=jXn=i,,X0)=P(Xn+1=jXn=i)P(X_{n+1} = j \mid X_n = i,\, \ldots,\, X_0) = P(X_{n+1} = j \mid X_n = i)

Ports in time

This concept was not invented in one instant

Follow the scenes to see problems, notation, standards of proof, and applications changing across different times and places.

1
AD 1906Scene 1 / 3Continue through the world of this year

From literature to probability — Pushkin’s verse

Andrey Markov analyzed sequences of vowels and consonants in Pushkin’s Eugene Onegin, demonstrating a probability model with dependence only on the current state.

No reliable place is given, so time continues without an invented pin

Continue through the world of this year
2
AD 1953Scene 2 / 3Continue through the world of this year

MCMC — the Metropolis algorithm

Work at Los Alamos introduced Markov chain Monte Carlo, making it possible to draw samples from distributions too difficult to handle directly.

No reliable place is given, so time continues without an invented pin

Continue through the world of this year
3
AD 1998Scene 3 / 3Continue through the world of this year

PageRank — a Markov chain on the web

Google’s random-surfer model treated hyperlinks as transition probabilities, using a Markov chain to rank pages across the web.

No reliable place is given, so time continues without an invented pin

Continue through the world of this year

Modern applications

Token prediction, reinforcement learning, MCMC Bayesian inference, and analysis of how thoroughly a deck of cards has been shuffled.

Beyond MathVoyage

Curated sources and problems. Bring one discovery back from OEIS, Project Euler, MathOverflow, or arXiv.

No concept belongs to one person

Follow people who played different roles

These are not inventor credits. They are different ports: opening a problem, sharpening a language, or carrying it into another world.

Number lenses

A concept looks different when its world of numbers changes

These numbers are editorial lenses for the voyage, not required prerequisites.

Concept genealogy

What supports it, and what does it open?

Concepts arriving from before

Current port

Markov Chains

Concepts opened from here

Only direct editorial links are shown; this is not a complete learning order or historical influence line.