jssm 5.153.0 → 5.154.0

package/jssm.es6.d.ts

@@ -524,6 +524,34 @@ declare type JssmEditorConfig = {
     stochastic_run_count?: number;
     panels?: Array<string>;
 };
+/** Which stochastic view a run batch produces. */
+declare type JssmStochasticMode = 'montecarlo' | 'steady_state';
+/** Options for {@link Machine.stochastic_summary} / {@link Machine.stochastic_runs}. */
+declare type JssmStochasticOptions = {
+    mode?: JssmStochasticMode;
+    runs?: number;
+    max_steps?: number;
+    seed?: number;
+};
+/** One walk's result, yielded by {@link Machine.stochastic_runs}. */
+declare type JssmStochasticRun = {
+    states: Array<string>;
+    edges: Array<string>;
+    length: number;
+    terminated: boolean;
+};
+/** Aggregate statistics over a stochastic run batch. */
+declare type JssmStochasticSummary = {
+    mode: JssmStochasticMode;
+    runs: number;
+    seed: number;
+    state_visits: Map<string, number>;
+    state_visit_fraction: Map<string, number>;
+    edge_traversals: Map<string, number>;
+    path_lengths?: Array<number>;
+    terminal_reached?: number;
+    capped?: number;
+};
 declare type JssmGenericConfig<StateType, DataType> = {
     graph_layout?: JssmLayout;
     complete?: Array<StateType>;
@@ -1991,6 +2019,10 @@ declare function transfer_state_properties(state_decl: JssmStateDeclaration): Js
  *
  */
 declare function state_style_condense(jssk: JssmStateStyleKeyList, machine?: any): JssmStateConfig;
+/** Default number of independent Monte-Carlo runs when none is declared. */
+declare const STOCHASTIC_DEFAULT_RUNS = 1000;
+/** Default per-run step cap (montecarlo) / walk length (steady_state). */
+declare const STOCHASTIC_DEFAULT_MAX_STEPS = 1000;
 declare class Machine<mDT> {
     _state: StateType;
     _states: Map<StateType, JssmGenericState>;
@@ -2919,6 +2951,65 @@ declare class Machine<mDT> {
      *  @returns A `Map` from state name to visit count.
      */
     probabilistic_histo_walk(n: number): Map<StateType, number>;
+    /** One non-destructive weighted-random walk over the graph from `start`.
+     *
+     *  Reads the graph and advances the PRNG only — it never calls
+     *  {@link Machine.transition}, so it fires no hooks, mutates no machine
+     *  state, and touches no `data`.  A state with no probabilistic exits
+     *  (a terminal, or a forced-only `~>` state) ends the walk.
+     *
+     *  @param start - State to begin the walk from.
+     *  @param max_steps - Maximum transitions before the walk is step-capped.
+     *  @returns The {@link JssmStochasticRun} for this walk.
+     */
+    private _stochastic_one_walk;
+    /** Lazily yield one {@link JssmStochasticRun} at a time.
+     *
+     *  In `montecarlo` mode (default) yields `runs` independent walks from the
+     *  current state, each ending at a terminal or after `max_steps`.  In
+     *  `steady_state` mode yields exactly one walk of `max_steps` steps.  This
+     *  is the lazy engine behind {@link Machine.stochastic_summary}; the
+     *  fsl-stochastic panel drives it across animation frames.
+     *
+     *  Passing `seed` reseeds the machine for reproducible runs.  Unlike
+     *  {@link Machine.stochastic_summary}, the generator does NOT restore the
+     *  prior seed afterward — a direct caller's machine is left reseeded.
+     *
+     *  @param opts - {@link JssmStochasticOptions}.
+     *  @returns A generator of per-run results.
+     *
+     *  @example
+     *  const m = sm`a 'go' -> b 'go' -> c;`;
+     *  [...m.stochastic_runs({ runs: 2, seed: 1 })].length;  // => 2
+     */
+    stochastic_runs(opts?: JssmStochasticOptions): Generator<JssmStochasticRun>;
+    /** Run many weighted-random walks and return aggregate statistics.
+     *
+     *  Honors `%` transition probabilities (via the existing probabilistic
+     *  machinery).  Non-destructive: the machine's current state and
+     *  {@link Machine.rng_seed} are restored before returning, so calling this
+     *  never perturbs the live machine.  `montecarlo` mode (default) reports
+     *  per-run `path_lengths`, `terminal_reached`, and `capped`; `steady_state`
+     *  mode runs one long walk and omits those fields.
+     *
+     *  Timing (`after`) decorations and data-guard conditions are not modeled
+     *  by this sampler; it walks the probabilistic graph topology.
+     *
+     *  @param opts - {@link JssmStochasticOptions}.  `runs` defaults to the
+     *  machine's declared `editor: { stochastic_run_count }` (fsl#1334) when
+     *  present, otherwise {@link STOCHASTIC_DEFAULT_RUNS}.
+     *  @returns A {@link JssmStochasticSummary}.
+     *
+     *  @see Machine.stochastic_runs
+     *  @see Machine.probabilistic_walk
+     *  @see Machine.editor_config
+     *
+     *  @example
+     *  const m = sm`a 'go' -> b 'go' -> c;`;
+     *  const s = m.stochastic_summary({ runs: 100, seed: 1 });
+     *  s.terminal_reached;  // => 100
+     */
+    stochastic_summary(opts?: JssmStochasticOptions): JssmStochasticSummary;
     /********
      *
      *  List all actions available from this state.  Please note that the order of
@@ -4691,4 +4782,4 @@ declare function compareVersions(v1: string, v2: string): number;
  */
 declare function deserialize<mDT>(machine_string: string, ser: JssmSerialization<mDT>): Machine<mDT>;
 
-export { FslDirections, Machine, abstract_everything_hook_step, abstract_hook_step, action_label_chars, arrow_direction, arrow_left_kind, arrow_right_kind, build_time, compareVersions, compile, jssm_constants_d as constants, deserialize, find_repeated, from, fslCompletions, fslDiagnostics, fslSemanticSpans, gen_splitmix32, gviz_shapes, histograph, is_hook_complex_result, is_hook_rejection, make, named_colors, wrap_parse as parse, seq, shapes, sleep, sm, state_name_chars, state_name_first_chars, state_style_condense, transfer_state_properties, unique, version, weighted_histo_key, weighted_rand_select, weighted_sample_select };
+export { FslDirections, Machine, STOCHASTIC_DEFAULT_MAX_STEPS, STOCHASTIC_DEFAULT_RUNS, abstract_everything_hook_step, abstract_hook_step, action_label_chars, arrow_direction, arrow_left_kind, arrow_right_kind, build_time, compareVersions, compile, jssm_constants_d as constants, deserialize, find_repeated, from, fslCompletions, fslDiagnostics, fslSemanticSpans, gen_splitmix32, gviz_shapes, histograph, is_hook_complex_result, is_hook_rejection, make, named_colors, wrap_parse as parse, seq, shapes, sleep, sm, state_name_chars, state_name_first_chars, state_style_condense, transfer_state_properties, unique, version, weighted_histo_key, weighted_rand_select, weighted_sample_select };

package/jssm_viz.es5.d.cts

@@ -519,6 +519,34 @@ declare type JssmEditorConfig = {
     stochastic_run_count?: number;
     panels?: Array<string>;
 };
+/** Which stochastic view a run batch produces. */
+declare type JssmStochasticMode = 'montecarlo' | 'steady_state';
+/** Options for {@link Machine.stochastic_summary} / {@link Machine.stochastic_runs}. */
+declare type JssmStochasticOptions = {
+    mode?: JssmStochasticMode;
+    runs?: number;
+    max_steps?: number;
+    seed?: number;
+};
+/** One walk's result, yielded by {@link Machine.stochastic_runs}. */
+declare type JssmStochasticRun = {
+    states: Array<string>;
+    edges: Array<string>;
+    length: number;
+    terminated: boolean;
+};
+/** Aggregate statistics over a stochastic run batch. */
+declare type JssmStochasticSummary = {
+    mode: JssmStochasticMode;
+    runs: number;
+    seed: number;
+    state_visits: Map<string, number>;
+    state_visit_fraction: Map<string, number>;
+    edge_traversals: Map<string, number>;
+    path_lengths?: Array<number>;
+    terminal_reached?: number;
+    capped?: number;
+};
 declare type JssmGenericConfig<StateType, DataType> = {
     graph_layout?: JssmLayout;
     complete?: Array<StateType>;
@@ -2156,6 +2184,65 @@ declare class Machine<mDT> {
      *  @returns A `Map` from state name to visit count.
      */
     probabilistic_histo_walk(n: number): Map<StateType, number>;
+    /** One non-destructive weighted-random walk over the graph from `start`.
+     *
+     *  Reads the graph and advances the PRNG only — it never calls
+     *  {@link Machine.transition}, so it fires no hooks, mutates no machine
+     *  state, and touches no `data`.  A state with no probabilistic exits
+     *  (a terminal, or a forced-only `~>` state) ends the walk.
+     *
+     *  @param start - State to begin the walk from.
+     *  @param max_steps - Maximum transitions before the walk is step-capped.
+     *  @returns The {@link JssmStochasticRun} for this walk.
+     */
+    private _stochastic_one_walk;
+    /** Lazily yield one {@link JssmStochasticRun} at a time.
+     *
+     *  In `montecarlo` mode (default) yields `runs` independent walks from the
+     *  current state, each ending at a terminal or after `max_steps`.  In
+     *  `steady_state` mode yields exactly one walk of `max_steps` steps.  This
+     *  is the lazy engine behind {@link Machine.stochastic_summary}; the
+     *  fsl-stochastic panel drives it across animation frames.
+     *
+     *  Passing `seed` reseeds the machine for reproducible runs.  Unlike
+     *  {@link Machine.stochastic_summary}, the generator does NOT restore the
+     *  prior seed afterward — a direct caller's machine is left reseeded.
+     *
+     *  @param opts - {@link JssmStochasticOptions}.
+     *  @returns A generator of per-run results.
+     *
+     *  @example
+     *  const m = sm`a 'go' -> b 'go' -> c;`;
+     *  [...m.stochastic_runs({ runs: 2, seed: 1 })].length;  // => 2
+     */
+    stochastic_runs(opts?: JssmStochasticOptions): Generator<JssmStochasticRun>;
+    /** Run many weighted-random walks and return aggregate statistics.
+     *
+     *  Honors `%` transition probabilities (via the existing probabilistic
+     *  machinery).  Non-destructive: the machine's current state and
+     *  {@link Machine.rng_seed} are restored before returning, so calling this
+     *  never perturbs the live machine.  `montecarlo` mode (default) reports
+     *  per-run `path_lengths`, `terminal_reached`, and `capped`; `steady_state`
+     *  mode runs one long walk and omits those fields.
+     *
+     *  Timing (`after`) decorations and data-guard conditions are not modeled
+     *  by this sampler; it walks the probabilistic graph topology.
+     *
+     *  @param opts - {@link JssmStochasticOptions}.  `runs` defaults to the
+     *  machine's declared `editor: { stochastic_run_count }` (fsl#1334) when
+     *  present, otherwise {@link STOCHASTIC_DEFAULT_RUNS}.
+     *  @returns A {@link JssmStochasticSummary}.
+     *
+     *  @see Machine.stochastic_runs
+     *  @see Machine.probabilistic_walk
+     *  @see Machine.editor_config
+     *
+     *  @example
+     *  const m = sm`a 'go' -> b 'go' -> c;`;
+     *  const s = m.stochastic_summary({ runs: 100, seed: 1 });
+     *  s.terminal_reached;  // => 100
+     */
+    stochastic_summary(opts?: JssmStochasticOptions): JssmStochasticSummary;
     /********
      *
      *  List all actions available from this state.  Please note that the order of

package/jssm_viz.es6.d.ts

@@ -519,6 +519,34 @@ declare type JssmEditorConfig = {
     stochastic_run_count?: number;
     panels?: Array<string>;
 };
+/** Which stochastic view a run batch produces. */
+declare type JssmStochasticMode = 'montecarlo' | 'steady_state';
+/** Options for {@link Machine.stochastic_summary} / {@link Machine.stochastic_runs}. */
+declare type JssmStochasticOptions = {
+    mode?: JssmStochasticMode;
+    runs?: number;
+    max_steps?: number;
+    seed?: number;
+};
+/** One walk's result, yielded by {@link Machine.stochastic_runs}. */
+declare type JssmStochasticRun = {
+    states: Array<string>;
+    edges: Array<string>;
+    length: number;
+    terminated: boolean;
+};
+/** Aggregate statistics over a stochastic run batch. */
+declare type JssmStochasticSummary = {
+    mode: JssmStochasticMode;
+    runs: number;
+    seed: number;
+    state_visits: Map<string, number>;
+    state_visit_fraction: Map<string, number>;
+    edge_traversals: Map<string, number>;
+    path_lengths?: Array<number>;
+    terminal_reached?: number;
+    capped?: number;
+};
 declare type JssmGenericConfig<StateType, DataType> = {
     graph_layout?: JssmLayout;
     complete?: Array<StateType>;
@@ -2156,6 +2184,65 @@ declare class Machine<mDT> {
      *  @returns A `Map` from state name to visit count.
      */
     probabilistic_histo_walk(n: number): Map<StateType, number>;
+    /** One non-destructive weighted-random walk over the graph from `start`.
+     *
+     *  Reads the graph and advances the PRNG only — it never calls
+     *  {@link Machine.transition}, so it fires no hooks, mutates no machine
+     *  state, and touches no `data`.  A state with no probabilistic exits
+     *  (a terminal, or a forced-only `~>` state) ends the walk.
+     *
+     *  @param start - State to begin the walk from.
+     *  @param max_steps - Maximum transitions before the walk is step-capped.
+     *  @returns The {@link JssmStochasticRun} for this walk.
+     */
+    private _stochastic_one_walk;
+    /** Lazily yield one {@link JssmStochasticRun} at a time.
+     *
+     *  In `montecarlo` mode (default) yields `runs` independent walks from the
+     *  current state, each ending at a terminal or after `max_steps`.  In
+     *  `steady_state` mode yields exactly one walk of `max_steps` steps.  This
+     *  is the lazy engine behind {@link Machine.stochastic_summary}; the
+     *  fsl-stochastic panel drives it across animation frames.
+     *
+     *  Passing `seed` reseeds the machine for reproducible runs.  Unlike
+     *  {@link Machine.stochastic_summary}, the generator does NOT restore the
+     *  prior seed afterward — a direct caller's machine is left reseeded.
+     *
+     *  @param opts - {@link JssmStochasticOptions}.
+     *  @returns A generator of per-run results.
+     *
+     *  @example
+     *  const m = sm`a 'go' -> b 'go' -> c;`;
+     *  [...m.stochastic_runs({ runs: 2, seed: 1 })].length;  // => 2
+     */
+    stochastic_runs(opts?: JssmStochasticOptions): Generator<JssmStochasticRun>;
+    /** Run many weighted-random walks and return aggregate statistics.
+     *
+     *  Honors `%` transition probabilities (via the existing probabilistic
+     *  machinery).  Non-destructive: the machine's current state and
+     *  {@link Machine.rng_seed} are restored before returning, so calling this
+     *  never perturbs the live machine.  `montecarlo` mode (default) reports
+     *  per-run `path_lengths`, `terminal_reached`, and `capped`; `steady_state`
+     *  mode runs one long walk and omits those fields.
+     *
+     *  Timing (`after`) decorations and data-guard conditions are not modeled
+     *  by this sampler; it walks the probabilistic graph topology.
+     *
+     *  @param opts - {@link JssmStochasticOptions}.  `runs` defaults to the
+     *  machine's declared `editor: { stochastic_run_count }` (fsl#1334) when
+     *  present, otherwise {@link STOCHASTIC_DEFAULT_RUNS}.
+     *  @returns A {@link JssmStochasticSummary}.
+     *
+     *  @see Machine.stochastic_runs
+     *  @see Machine.probabilistic_walk
+     *  @see Machine.editor_config
+     *
+     *  @example
+     *  const m = sm`a 'go' -> b 'go' -> c;`;
+     *  const s = m.stochastic_summary({ runs: 100, seed: 1 });
+     *  s.terminal_reached;  // => 100
+     */
+    stochastic_summary(opts?: JssmStochasticOptions): JssmStochasticSummary;
     /********
      *
      *  List all actions available from this state.  Please note that the order of

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jssm",
-  "version": "5.153.0",
+  "version": "5.154.0",
   "engines": {
     "node": ">=10.0.0"
   },