Lattice API

Lattice API Quick Reference

Detailed API documentation for the major lattice subsystems. Use /lattice-search for discovery; use this skill for API details once you know what subsystem you need.

FP Toolkit (lattice/fp/)

Monads, parsers, streams, zippers, game theory, symbolic math, control systems (state-space, Kalman filters, PID, stability analysis), term rewriting.

(li 'fp)   ; Full description
(le 'fp)   ; All exports

Regex/FSM (lattice/fp/parsing/)

Regex→NFA compilation via Thompson's construction.

Features:

• Quantifier ranges {n,m}
• Anchors ^/$
• Lookahead (?=...)/(?!...)

Note: Anchors and lookahead require position-aware NFA execution; nfa->dfa and fsm-minimize skip conversion when assertions present. Use regex-accepts? or fsm-accepts? which auto-detect and use appropriate runner.

(regex-accepts? "^foo" "foobar")  ; #t
(regex-accepts? "foo$" "foobar")  ; #f

Game Theory (lattice/fp/game/)

Rich set of ready-to-use algorithms for cooperative games, voting, matching, and fair division.

These are pure functions — import them into boundary code for applications like QA triage, resource allocation, or voting systems.

Example: Shapley Value

(load "lattice/fp/game/coop-games.ss")

;; Define a cooperative game: 3 players, characteristic function
(define game (make-coop-game 3
  (lambda (coalition)
    (cond
      [(equal? coalition '(1 2 3)) 100]  ; Grand coalition
      [(equal? coalition '(1 2)) 70]
      [(equal? coalition '(1 3)) 50]
      [(equal? coalition '(2 3)) 60]
      [else 0]))))

(shapley-value game)  ; => fair allocation to each player

Statistics (lattice/statistics/)

(li 'statistics)  ; Full description
(le 'statistics)  ; All exports

CLP(FD) (lattice/fp/clp/)

cKanren-style constraint logic programming with finite domains.

Features:

• Finite domain constraints
• Arithmetic constraints (fd-<, fd-+, etc.)
• Global constraints (all-different)
• Intelligent search strategies

Classic problems: N-Queens, Sudoku, cryptarithmetic

(load "lattice/fp/clp/clpfd.ss")

;; N-Queens
(run* (q)
  (fresh (q1 q2 q3 q4)
    (== q (list q1 q2 q3 q4))
    (fd-dom q1 '(1 2 3 4))
    (fd-dom q2 '(1 2 3 4))
    (fd-dom q3 '(1 2 3 4))
    (fd-dom q4 '(1 2 3 4))
    (all-different q)
    (queens-safe q)))

SAT/MaxSAT (lattice/fp/sat/)

CDCL SAT solver with clause learning, Two-Watched Literals, VSIDS branching. MaxSAT extension for optimization.

Example: Graph Coloring

(load "lattice/fp/sat/sat.ss")
(load "lattice/fp/sat/applications.ss")

;; 3-color a graph
(define edges '((1 . 2) (2 . 3) (3 . 1)))
(define coloring-cnf (graph-coloring 3 3 edges))
(sat-solve coloring-cnf)  ; => 'sat or 'unsat
(sat-model coloring-cnf)  ; => variable assignments

Optics (lattice/fp/optics/)

Complete optics tower for composable data access.

Operators

Example: Nested Access

(load "lattice/fp/optics/optics.ss")

;; View nested position
(^. body body-pos-lens)

;; Modify x-coordinate of position
(& body (%~ (>>> body-pos-lens vec2-x-lens) add1))

;; Get all velocities from world
(^.. world (>>> world-all-bodies body-vel-lens))

Traced Optics (lattice/autodiff/traced-optics.ss)

Compute gradients through optic-focused paths — combines autodiff with optics.

(load "lattice/autodiff/traced-optics.ss")

;; Gradient of loss w.r.t. nested parameter via optic composition
(optic-gradient loss-fn (>>> outer-lens inner-lens) structure)

;; Gradient descent step at optic focus
(optimize-at lens-fst '(5.0 . ignored) (lambda (p) (traced-sq (car p))) 0.1)
;; => (4.0 . ignored)  ; 5 - 0.1 * 2 * 5 = 4

;; Gradients for all traversal targets
(optic-gradient-list loss-fn traversal-each '(1 2 3))

Skill Manifests

Each lattice skill has a manifest.sexp declaring metadata:

(skill <name>
  (version "x.y.z")
  (tier 0-2)                       ; 0=foundational, 1=intermediate, 2+=advanced
  (path "lattice/<name>")
  (purity total|partial)           ; total=pure, partial=may have effects
  (stability stable|experimental)
  (fuel-bound "O(...)")            ; Complexity bound
  (deps (<skill> ...))             ; Skill-level dependencies
  (description "...")
  (keywords (<keyword> ...))       ; For search
  (aliases (<alias> ...))          ; Alternative names
  (exports (<module> <symbol> ...) ...)
  (modules (<name> "<file>" "<desc>") ...))