JSDoc: Class: Chain

Constructor

new Chain(optionsopt)

Parameters:

Name Type Attributes Default Description

options

object

{}

Options object

Properties

Name	Type	Attributes	Default	Description
`corpus`	Array.<Array.<any>>	<optional>	[]	Sample runs of the process
`order`	number	<optional>	0	Size of the chain's memory
`useTokenMap`	boolean	<optional>	false	Whether to map token to states
`model`	Map.<Tuple.<any>, any>	<optional>		Prebuilt state space

Source:

chain.js, line 33

Members

(readonly) initialState :Tuple.<any>

Initial state with BEGIN tokens.

Type:

Tuple.<any>

Source:

chain.js, line 78

(readonly) model :Map.<Tuple.<any>, any>

Map of chain states.

Type:

Map.<Tuple.<any>, any>

Source:

chain.js, line 69

(readonly) order :number

Order of chain.

Type:

number

Source:

chain.js, line 60

(readonly) tokenMap :Map.<any, Tuple.<any>>

Map of token to state.

Type:

Map.<any, Tuple.<any>>

Source:

chain.js, line 90

Methods

(static) fromJSON(jsonChain, options) → {Chain}

Creates a Chain from a JSON string.

Parameters:

Name	Type	Description
`jsonChain`	string	A chain serialised with `Chain.toJSON`
`options`	object	Additional options to Chain constructor

Source:

chain.js, line 330

Returns:

A new chain instance

Type: Chain

(static) seed(run, chainopt)

Updates a model from a single run.

Parameters:

Name Type Attributes Default Description

run

Array.<any>

Array of tokens

chain

object

{}

Chain object

Properties

Name	Type	Attributes	Description
`model`	Map.<Tuple.<any>, any>		Model to update
`initialState`	Tuple.<any>		Starting tuple
`order`	number		Order of chain
`tokenMap`	Map.<any, Tuple.<any>>	<optional>	Map of token to states

Source:

chain.js, line 105

run(optionsopt) → {Array.<Array.<any>>}

Walks the Markov chain and returns all steps.

The returned step array will look like:

[ [backward_steps], [starting_tokens], [forward_steps] ]

The starting tokens are only returned when forward or backward steps were actually generated from a subset of the {options.tokens} parameter.

Parameters:

Name Type Attributes Description

options

object

Options object

Properties

Name	Type	Attributes	Default	Description
`tokens`	Array.<any>	<optional>	[]	Starting state tokens
`backSearch`	boolean	<optional>	true	Should walk back
`useTokenMap`	boolean	<optional>	true	Whether to use token map
`runMissingTokens`	boolean	<optional>	true	Whether to answer when tokens are not in model

Source:

chain.js, line 267

Returns:

Array with back root and forward steps

Type: Array.<Array.<any>>

toJSON() → {Array.<any>}

Serialises the chain into a JSONable array.

The returned array will look like:

[ [ [state], [ [next, count], ...], [ [prev, count], ...] ], ...]

Source:

chain.js, line 302

See:

https://mdn.io/stringify#toJSON()_behavior

Returns:

JSON array

Type: Array.<any>

(generator) walkBackward(fromStateopt) → {any}

Generates successive states until the chain reaches a BEGIN.

Parameters:

Name	Type	Attributes	Description
`fromState`	Tuple.<any>	<optional>	Starting state of the chain walk

Source:

chain.js, line 210

Yields:

Next preceeding step of the chain

Type: any

(generator) walkForward(fromStateopt) → {any}

Generates successive states until the chain reaches an END.

Parameters:

Name	Type	Attributes	Description
`fromState`	Tuple.<any>	<optional>	Begin state of the chain walk

Source:

chain.js, line 200

Yields:

Next succeding step of the chain

Type: any