eyeling 1.34.6 → 2.35.0

package/README.md

@@ -224,6 +224,25 @@ The CLI renders the `log:outputString` values directly:
 hello
 ```
 
+### Predicate memoization
+
+Eyeling supports a predicate-local performance directive for backward rules:
+
+```n3
+@prefix log: <http://www.w3.org/2000/10/swap/log#> .
+
+:predicate log:memoize true .
+```
+
+A top-level `log:memoize true` triple is treated as an engine directive rather than an RDF fact. It asks the backward prover to reuse completed answers for that predicate during the current reasoning run. This is especially useful for recursive predicates with overlapping subgoals, such as the direct Fibonacci recurrence or Collatz trajectories where many starting values share suffixes:
+
+```n3
+:fibonacci log:memoize true .
+:collatz log:memoize true .
+```
+
+Memoization is intentionally selective. It can slow down pure generators, predicates with mostly unique states, or small linear recursions because maintaining the table costs more than recomputation. Prefer it for dynamic-programming-shaped predicates that are called repeatedly with the same bound subject or object. The examples `examples/fibonacci.n3`, `examples/collatz-1000.n3`, and `examples/fundamental-theorem-arithmetic.n3` show cases where the directive helps.
+
 ---
 
 ## Command-line interface
@@ -743,6 +762,8 @@ Eyeling supports both native N3 list terms and materialized RDF collections. Ano
 
 Formula-aware built-ins make Eyeling useful for meta-reasoning. `log:includes`, `log:notIncludes`, `log:collectAllIn`, and `log:forAllIn` prove goals inside formula scopes. `log:query` selects output triples, while `log:outputString` writes selected string values directly to stdout.
 
+`log:memoize` is a top-level performance directive for backward predicates, not an RDF-producing fact and not a general-purpose built-in. Use it when a recursive predicate has overlapping subgoals whose completed answers are likely to be requested again.
+
 `log:semantics`, `log:content`, and related built-ins may dereference sources. Use `--enforce-https` or `{ enforceHttps: true }` in environments where HTTP-to-HTTPS rewriting is required.
 
 
@@ -901,6 +922,8 @@ The backward prover solves rule bodies. It can match current facts, use backward
 { :example :works true } .
 ```
 
+For selected predicates, `:predicate log:memoize true .` enables tabling of completed backward answers. This preserves the normal declarative reading of the backward rules but can turn repeated recursive work into table lookups. The directive is removed from the data facts before reasoning, so it does not appear in normal output.
+
 ### Dynamic rules
 
 Eyeling treats top-level `log:implies` and `log:impliedBy` as rule forms and can activate derived implication facts as live rules during reasoning. This supports programs that derive rules as part of their logic.
@@ -1042,6 +1065,16 @@ The repository contains more than two hundred N3 examples under `examples/`, RDF
 | `examples/rdf-messages.n3` | RDF Message Log replay. |
 | `examples/context-schema-audit.n3` | Quoted-context schema validation with `log:includes` and list arity checks. |
 
+### Memoization examples
+
+These examples demonstrate `log:memoize` on predicates that benefit from reusing completed backward-rule answers:
+
+| Example | Memoized predicate | Why it helps |
+|---|---|---|
+| `examples/fibonacci.n3` | `:fibonacci` | The direct recurrence repeatedly asks for the same smaller Fibonacci numbers. |
+| `examples/collatz-1000.n3` | `:collatz` | Many starting values share Collatz suffix trajectories. |
+| `examples/fundamental-theorem-arithmetic.n3` | `:smallestDivisorFrom`, `:factorSmallest` | Factorization and independent checks reuse trial-division and factor-list subgoals. |
+
 ### Running all examples through tests
 
 ```bash

package/dist/browser/eyeling.browser.js

@@ -6242,7 +6242,7 @@ function runParsedDocumentOnce(mergedDocument, { rdfMode = false, outputPrefixes
   let outTriples = [];
   let queryDerived = [];
   if (hasQueries) {
-    const res = engine.forwardChainAndCollectLogQueryConclusions(facts, frules, brules, qrules, { prefixes });
+    const res = engine.forwardChainAndCollectLogQueryConclusions(facts, frules, brules, qrules, null, { captureExplanations: false, prefixes });
     derived = res.derived;
     outTriples = res.queryTriples;
     queryDerived = res.queryDerived || [];
@@ -6929,7 +6929,7 @@ function factsContainOutputStrings(triplesForOutput) {
   let outDerived = [];
 
   if (hasQueries) {
-    const res = engine.forwardChainAndCollectLogQueryConclusions(facts, frules, brules, qrules, { prefixes });
+    const res = engine.forwardChainAndCollectLogQueryConclusions(facts, frules, brules, qrules, null, { captureExplanations: engine.getProofCommentsEnabled(), prefixes });
     derived = res.derived;
     outTriples = res.queryTriples;
     outDerived = res.queryDerived;
@@ -7507,6 +7507,7 @@ const LOG_COLLECT_ALL_IN_IRI = LOG_NS + 'collectAllIn';
 const LOG_FOR_ALL_IN_IRI = LOG_NS + 'forAllIn';
 const LOG_INCLUDES_IRI = LOG_NS + 'includes';
 const LOG_NOT_INCLUDES_IRI = LOG_NS + 'notIncludes';
+const LOG_MEMOIZE_IRI = LOG_NS + 'memoize';
 const LOG_IMPLIES_IRI = LOG_NS + 'implies';
 const LOG_IMPLIED_BY_IRI = LOG_NS + 'impliedBy';
 
@@ -7608,6 +7609,7 @@ const FACT_INDEX_ARRAY_MAP_FIELDS = ['__byPred', '__byPNonFastS', '__byPNonFastO
 const FACT_INDEX_NESTED_ARRAY_MAP_FIELDS = ['__byPS', '__byPO'];
 const FACT_INDEX_MAP_FIELDS = FACT_INDEX_ARRAY_MAP_FIELDS.concat(FACT_INDEX_NESTED_ARRAY_MAP_FIELDS);
 const FACT_INDEX_REQUIRED_FIELDS = FACT_INDEX_ARRAY_FIELDS.concat(FACT_INDEX_MAP_FIELDS, ['__keySet']);
+const FACT_INDEX_ALL_FIELDS = FACT_INDEX_REQUIRED_FIELDS.concat(['__keySetComplete']);
 
 let version = 'dev';
 try {
@@ -8675,7 +8677,13 @@ function termLookupKey(t) {
     return t.__tid;
   }
   if (t instanceof Blank) return t.__tid;
-  if (t instanceof Literal) return literalLookupKey(t);
+  if (t instanceof Literal) {
+    const cached = t.__lookupKey;
+    if (cached !== undefined) return cached;
+    const key = literalLookupKey(t);
+    Object.defineProperty(t, '__lookupKey', { value: key, enumerable: false });
+    return key;
+  }
 
   if (t instanceof ListTerm) {
     const cached = t.__lookupKey;
@@ -8713,6 +8721,13 @@ function tripleFastKey(tr) {
   return ks + '\t' + kp + '\t' + ko;
 }
 
+function resetFactIndexes(facts) {
+  if (!facts) return;
+  for (const name of FACT_INDEX_ALL_FIELDS) {
+    try { delete facts[name]; } catch {}
+  }
+}
+
 function ensureFactIndexes(facts) {
   if (FACT_INDEX_REQUIRED_FIELDS.every((name) => facts[name])) return;
 
@@ -9143,6 +9158,157 @@ function isLogImpliedBy(p) {
   return p instanceof Iri && p.value === LOG_IMPLIED_BY_IRI;
 }
 
+function isLogMemoize(p) {
+  return p instanceof Iri && p.value === LOG_MEMOIZE_IRI;
+}
+
+function isTruthyMemoizeObject(o) {
+  if (!(o instanceof Literal)) return false;
+  if (o.value === 'true') return true;
+  const info = parseBooleanLiteralInfo(o);
+  return !!(info && info.value === true);
+}
+
+// ===========================================================================
+// Predicate memoization declarations
+// ===========================================================================
+//
+// Eyeling accepts top-level declarations of the form:
+//   :predicate log:memoize true.
+// They are engine directives, not data facts.  Each declaration asks the
+// backward prover to table completed answers for that predicate during the
+// current reasoning scope.  The hint is deliberately predicate-local and
+// conservative: unbound calls and finite-result probes still use the normal
+// path, and cached answers are invalidated when the fact/rule scope changes.
+
+function __extractLogMemoizeDeclarations(facts) {
+  const memoized = new Set();
+  if (!Array.isArray(facts) || facts.length === 0) return memoized;
+
+  let write = 0;
+  let removed = false;
+  for (let read = 0; read < facts.length; read++) {
+    const tr = facts[read];
+    if (tr && isLogMemoize(tr.p) && tr.s instanceof Iri && isTruthyMemoizeObject(tr.o)) {
+      memoized.add(tr.s.__tid);
+      removed = true;
+      continue;
+    }
+    if (write !== read) facts[write] = tr;
+    write++;
+  }
+
+  if (removed) {
+    facts.length = write;
+    resetFactIndexes(facts);
+  }
+  return memoized;
+}
+
+function __attachMemoizedPredicates(scopeCarrier, memoized) {
+  if (!scopeCarrier || !memoized || memoized.size === 0) return;
+  __setHiddenWritable(scopeCarrier, 'memoizedPredicates', memoized);
+}
+
+function __memoizedPredicateSet(facts, backRules) {
+  return (facts && facts.memoizedPredicates) || (backRules && backRules.memoizedPredicates) || null;
+}
+
+function __predicateIsMemoized(facts, backRules, p) {
+  if (!(p instanceof Iri)) return false;
+  const memoized = __memoizedPredicateSet(facts, backRules);
+  return !!(memoized && memoized.has(p.__tid));
+}
+
+function __makePredicateMemoTable() {
+  return {
+    scopeVersion: null,
+    entries: new Map(),
+  };
+}
+
+function __ensurePredicateMemoTable(facts, backRules) {
+  let table = (facts && facts.predicateMemoTable) || (backRules && backRules.predicateMemoTable) || null;
+  if (!table) table = __makePredicateMemoTable();
+  __setHiddenWritable(facts, 'predicateMemoTable', table);
+  __setHiddenWritable(backRules, 'predicateMemoTable', table);
+
+  const version = goalTableScopeVersion(facts, backRules);
+  if (table.scopeVersion !== version) {
+    table.scopeVersion = version;
+    table.entries.clear();
+  }
+  return table;
+}
+
+function __memoKeyPart(t) {
+  if (__termHasVarOrBlank(t)) return { bound: false, text: '_' };
+  const fast = termLookupKey(t);
+  if (fast !== null) return { bound: true, text: encodeLookupKeyPart(fast) };
+  return { bound: true, text: skolemKeyFromTerm(t) };
+}
+
+function __predicateMemoKey(goal) {
+  if (!(goal && goal.p instanceof Iri)) return null;
+  const sPart = __memoKeyPart(goal.s);
+  const oPart = __memoKeyPart(goal.o);
+  if (!sPart.bound && !oPart.bound) return null;
+  return `${goal.p.__tid}|${sPart.text}|${oPart.text}`;
+}
+
+function __makePredicateMemoEntry() {
+  return {
+    computing: false,
+    complete: false,
+    unsafe: false,
+    answers: [],
+    answerKeys: new Set(),
+  };
+}
+
+function __getPredicateMemoEntry(facts, backRules, key) {
+  const table = __ensurePredicateMemoTable(facts, backRules);
+  let entry = table.entries.get(key);
+  if (!entry) {
+    entry = __makePredicateMemoEntry();
+    table.entries.set(key, entry);
+  }
+  return entry;
+}
+
+function __canStorePredicateMemo(maxResults) {
+  return !(typeof maxResults === 'number' && maxResults > 0);
+}
+
+function __canMemoizeAnswerTerm(t) {
+  if (t instanceof Var || t instanceof OpenListTerm) return false;
+  if (t instanceof ListTerm) return t.elems.every(__canMemoizeAnswerTerm);
+  if (t instanceof GraphTerm) return t.triples.every((tr) => __canMemoizeAnswerTerm(tr.s) && __canMemoizeAnswerTerm(tr.p) && __canMemoizeAnswerTerm(tr.o));
+  return true;
+}
+
+function __memoAnswerKeyPart(t) {
+  const fast = termLookupKey(t);
+  if (fast !== null) return encodeLookupKeyPart(fast);
+  return skolemKeyFromTerm(t);
+}
+
+function __memoAnswerKey(tr) {
+  return __memoAnswerKeyPart(tr.s) + '\t' + __memoAnswerKeyPart(tr.p) + '\t' + __memoAnswerKeyPart(tr.o);
+}
+
+function __storePredicateMemoAnswer(entry, rawGoal, subst) {
+  const answer = applySubstTriple(rawGoal, subst || {});
+  if (!__canMemoizeAnswerTerm(answer.s) || !__canMemoizeAnswerTerm(answer.p) || !__canMemoizeAnswerTerm(answer.o)) {
+    entry.unsafe = true;
+    return;
+  }
+  const key = __memoAnswerKey(answer);
+  if (entry.answerKeys.has(key)) return;
+  entry.answerKeys.add(key);
+  entry.answers.push(answer);
+}
+
 // ===========================================================================
 // Completed-goal answer tables (minimal tabling)
 // ===========================================================================
@@ -10056,6 +10222,47 @@ function proveGoals(goals, subst, facts, backRules, depth, visited, varGen, maxR
       continue;
     }
 
+    if (frame.kind === 'memoComplete') {
+      frame.entry.computing = false;
+      if (frame.entry.unsafe) {
+        frame.table.entries.delete(frame.key);
+      } else {
+        frame.entry.complete = true;
+      }
+      continue;
+    }
+
+    if (frame.kind === 'memoAnswerIter') {
+      const answers = frame.entry.answers;
+      while (frame.idx < answers.length && results.length < max) {
+        const answer = answers[frame.idx++];
+        const mark = trail.length;
+        if (!unifyTripleTrail(frame.goal0, answer)) {
+          undoTo(mark);
+          continue;
+        }
+
+        if (!frame.restGoals.length) {
+          results.push(gcCompactForGoals(substMut, [], answerVars));
+          undoTo(mark);
+          if (results.length >= max) return results;
+          continue;
+        }
+
+        stack.push(frame);
+        stack.push({ kind: 'undo', substMark: mark, visitedMark: visitedTrail.length });
+        stack.push({
+          kind: 'node',
+          goalsNow: frame.restGoals,
+          curDepth: frame.curDepth + 1,
+          canDeferBuiltins: frame.canDeferBuiltins,
+          deferCount: 0,
+        });
+        break;
+      }
+      continue;
+    }
+
     if (frame.kind === 'deltaIter') {
       const deltas = frame.deltas;
       while (frame.idx < deltas.length && results.length < max) {
@@ -10197,7 +10404,21 @@ function proveGoals(goals, subst, facts, backRules, depth, visited, varGen, maxR
     }
 
     const rawGoal = goalsNow[0];
-    const restGoals = goalsNow.length > 1 ? goalsNow.slice(1) : [];
+    let restGoals = goalsNow.length > 1 ? goalsNow.slice(1) : [];
+
+    if (rawGoal && rawGoal.__kind === 'memoStore') {
+      __storePredicateMemoAnswer(rawGoal.entry, rawGoal.goal, substMut);
+      if (rawGoal.stop && restGoals.length === 0) continue;
+      stack.push({
+        kind: 'node',
+        goalsNow: restGoals,
+        curDepth: frame.curDepth,
+        canDeferBuiltins: frame.canDeferBuiltins,
+        deferCount: 0,
+      });
+      continue;
+    }
+
     const goal0 = applySubstTriple(rawGoal, substMut);
 
     // 1) Builtins
@@ -10293,8 +10514,46 @@ function proveGoals(goals, subst, facts, backRules, depth, visited, varGen, maxR
     const goalKey = tripleKeyForVisited(goal0);
     const goalWasVisited = visitedCounts.has(goalKey);
 
+    let memoCompleteFrame = null;
+    let memoReplayFrame = null;
+    if (__predicateIsMemoized(facts, backRules, goal0.p)) {
+      const memoKey = __predicateMemoKey(goal0);
+      if (memoKey !== null) {
+        const memoEntry = __getPredicateMemoEntry(facts, backRules, memoKey);
+        if (memoEntry.complete) {
+          stack.push({
+            kind: 'memoAnswerIter',
+            entry: memoEntry,
+            idx: 0,
+            goal0,
+            restGoals,
+            curDepth: frame.curDepth,
+            canDeferBuiltins: frame.canDeferBuiltins,
+          });
+          continue;
+        }
+        if (!memoEntry.computing && __canStorePredicateMemo(maxResults)) {
+          memoEntry.computing = true;
+          const table = __ensurePredicateMemoTable(facts, backRules);
+          memoReplayFrame = {
+            kind: 'memoAnswerIter',
+            entry: memoEntry,
+            idx: 0,
+            goal0,
+            restGoals,
+            curDepth: frame.curDepth,
+            canDeferBuiltins: frame.canDeferBuiltins,
+          };
+          memoCompleteFrame = { kind: 'memoComplete', entry: memoEntry, table, key: memoKey };
+          restGoals = [{ __kind: 'memoStore', entry: memoEntry, goal: rawGoal, stop: true }];
+        }
+      }
+    }
+
     // 3) Backward rules (indexed by head predicate) — explored first
     if (goal0.p instanceof Iri) {
+      if (memoReplayFrame) stack.push(memoReplayFrame);
+      if (memoCompleteFrame) stack.push(memoCompleteFrame);
       ensureBackRuleIndexes(backRules);
       const candRules = (backRules.__byHeadPred.get(goal0.p.__tid) || []).concat(backRules.__wildHeadPred);
 
@@ -10326,6 +10585,8 @@ function proveGoals(goals, subst, facts, backRules, depth, visited, varGen, maxR
         });
       }
     } else {
+      if (memoReplayFrame) stack.push(memoReplayFrame);
+      if (memoCompleteFrame) stack.push(memoCompleteFrame);
       // No IRI predicate: rule indexing doesn't apply; only try all facts.
       stack.push({
         kind: 'factIter',
@@ -10531,6 +10792,10 @@ function __exitReasoning(code, message /* optional */) {
 function forwardChain(facts, forwardRules, backRules, onDerived /* optional */, opts = {}) {
   enterReasoningRun();
   try {
+    const memoizedPredicates = __extractLogMemoizeDeclarations(facts);
+    __attachMemoizedPredicates(facts, memoizedPredicates);
+    __attachMemoizedPredicates(backRules, memoizedPredicates);
+
     ensureFactIndexes(facts);
     ensureBackRuleIndexes(backRules);
 
@@ -11169,7 +11434,7 @@ function reasonStream(input, opts = {}) {
   if (Array.isArray(logQueryRules) && logQueryRules.length) {
     // Query-selection mode: derive full closure, then output only the unique
     // instantiated conclusion triples of the log:query directives.
-    const res = forwardChainAndCollectLogQueryConclusions(facts, frules, brules, logQueryRules, { prefixes });
+    const res = forwardChainAndCollectLogQueryConclusions(facts, frules, brules, logQueryRules, null, { captureExplanations: proof, prefixes });
     derived = res.derived;
     queryTriples = res.queryTriples;
     queryDerived = res.queryDerived;
@@ -11203,7 +11468,7 @@ function reasonStream(input, opts = {}) {
           onDerived(payload);
         }
       },
-      { prefixes },
+      { captureExplanations: proof, prefixes },
     );
   }
 
@@ -14850,7 +15115,7 @@ function parseN3Text(text, opts = {}) {
   }
 
   const tokens = lex(text, { rdf });
-  const parser = new Parser(tokens);
+  const parser = new Parser(tokens, { sourceOffsets: sourceLocations });
   if (baseIri) parser.prefixes.setBase(baseIri);
   const [prefixes, triples, frules, brules, logQueryRules] = parser.parseDocument();
 
@@ -15125,7 +15390,7 @@ function annotateSourceOffset(obj, offset) {
 }
 
 class Parser {
-  constructor(tokens) {
+  constructor(tokens, options = {}) {
     this.toks = tokens;
     this.pos = 0;
     this.prefixes = PrefixEnv.newDefault();
@@ -15139,6 +15404,7 @@ class Parser {
     // makes derived output read naturally, e.g. ':s :p _:b.' preceding
     // '_:b :q :r.'
     this.pendingTriplesAfter = [];
+    this.trackSourceOffsets = !!(options && options.sourceOffsets);
   }
 
   peek() {
@@ -15151,6 +15417,10 @@ class Parser {
     return tok;
   }
 
+  annotateSourceOffset(obj, offset) {
+    return this.trackSourceOffsets ? annotateSourceOffset(obj, offset) : obj;
+  }
+
   fail(message, tok = this.peek()) {
     const off = tok && typeof tok.offset === 'number' ? tok.offset : null;
     throw new N3SyntaxError(message, off);
@@ -15255,12 +15525,12 @@ class Parser {
         this.next();
         const second = this.parseTerm();
         this.expectDot();
-        forwardRules.push(annotateSourceOffset(this.makeRule(first, second, true), statementOffset));
+        forwardRules.push(this.annotateSourceOffset(this.makeRule(first, second, true), statementOffset));
       } else if (this.peek().typ === 'OpImpliedBy') {
         this.next();
         const second = this.parseTerm();
         this.expectDot();
-        backwardRules.push(annotateSourceOffset(this.makeRule(first, second, false), statementOffset));
+        backwardRules.push(this.annotateSourceOffset(this.makeRule(first, second, false), statementOffset));
       } else {
         let more;
 
@@ -15279,16 +15549,16 @@ class Parser {
 
         // normalize explicit log:implies / log:impliedBy at top-level
         for (const tr0 of more) {
-          const tr = annotateSourceOffset(tr0, statementOffset);
+          const tr = this.annotateSourceOffset(tr0, statementOffset);
           if (isLogImplies(tr.p) && tr.s instanceof GraphTerm && tr.o instanceof GraphTerm) {
-            forwardRules.push(annotateSourceOffset(this.makeRule(tr.s, tr.o, true), statementOffset));
+            forwardRules.push(this.annotateSourceOffset(this.makeRule(tr.s, tr.o, true), statementOffset));
           } else if (isLogImpliedBy(tr.p) && tr.s instanceof GraphTerm && tr.o instanceof GraphTerm) {
-            backwardRules.push(annotateSourceOffset(this.makeRule(tr.s, tr.o, false), statementOffset));
+            backwardRules.push(this.annotateSourceOffset(this.makeRule(tr.s, tr.o, false), statementOffset));
           } else if (isLogQuery(tr.p) && tr.s instanceof GraphTerm && tr.o instanceof GraphTerm) {
             // Output-selection directive: { premise } log:query { conclusion }.
             // When present at top-level, eyeling prints only the instantiated conclusion
             // triples (unique) instead of all newly derived facts.
-            logQueries.push(annotateSourceOffset(this.makeRule(tr.s, tr.o, true), statementOffset));
+            logQueries.push(this.annotateSourceOffset(this.makeRule(tr.s, tr.o, true), statementOffset));
           } else {
             triples.push(tr);
           }

package/examples/collatz-1000.n3

@@ -19,6 +19,10 @@
 @prefix :     <http://example.org/collatz#> .
 @prefix math: <http://www.w3.org/2000/10/swap/math#> .
 @prefix list: <http://www.w3.org/2000/10/swap/list#> .
+@prefix log:  <http://www.w3.org/2000/10/swap/log#> .
+
+# Memoize the overlapping Collatz suffixes shared by many starting values.
+:collatz log:memoize true .
 
 # -----------------------------------------
 # Query / materialization of the test suite

package/examples/fibonacci.n3

@@ -3,18 +3,24 @@
 # See https://en.wikipedia.org/wiki/Fibonacci_number
 # ==================================================
 
+@prefix log: <http://www.w3.org/2000/10/swap/log#>.
 @prefix math: <http://www.w3.org/2000/10/swap/math#>.
 @prefix : <https://eyereasoner.github.io/eye/reasoning#>.
 
-{?X :fibonacci ?Y} <= {(?X 0 1) :fib ?Y}.
+# Memoize the overlapping recursive subproblems.  The direct definition below
+# intentionally exposes the classic recurrence so repeated calls such as
+# F(10000), F(1000), F(100), ... can reuse completed :fibonacci answers.
+:fibonacci log:memoize true.
 
-{(0 ?A ?B) :fib ?A} <= true.
-{(1 ?A ?B) :fib ?B} <= true.
-{(?X ?A ?B) :fib ?Y} <= {
+{0 :fibonacci 0} <= true.
+{1 :fibonacci 1} <= true.
+{?X :fibonacci ?Y} <= {
     ?X math:greaterThan 1.
-    (?X 1) math:difference ?D.
-    (?A ?B) math:sum ?S.
-    (?D ?B ?S) :fib ?Y.
+    (?X 1) math:difference ?X1.
+    (?X 2) math:difference ?X2.
+    ?X1 :fibonacci ?Y1.
+    ?X2 :fibonacci ?Y2.
+    (?Y1 ?Y2) math:sum ?Y.
 }.
 
 {
@@ -34,4 +40,3 @@
         10000 :fibonacci ?F10000.
     }.
 }.
-

package/examples/fundamental-theorem-arithmetic.n3

@@ -34,6 +34,12 @@
 # Backward helper relations
 # -------------------------
 
+# Memoize repeated factorization subgoals.  The example asks for the same
+# smallest-divisor and factor-list computations through several independent
+# checks, so tabling completed answers avoids recomputing trial division.
+:smallestDivisorFrom log:memoize true .
+:factorSmallest log:memoize true .
+
 # divides0(A,B): A divides B in the positive integers.
 { (?A ?B) :divides0 true }
 <=