Drop log2q from the UI; keep it as recorded data only

Per review, log2q is display-only noise (the hint-count laws depend only on
(n,h)). Removed the params-table column; kept the transcribed value in PARAM_SETS
and PAPER-NOTES.md for completeness, with the positivity test retained. The
"Validated hints" column stays. Known Gaps reverts to "q is not modeled."

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: systemslibrarian

BUILD-NOTES.md

@@ -42,9 +42,10 @@ emits with base `/crypto-lab-lwe-hints/`.
 
 ## Notes for Paul
 
-1. **`log₂ q` now surfaced.** Each `PARAM_SETS` row carries its Table 1 modulus
-   bit-size (`log2q`), shown in the in-app params table. It is informational only —
-   the hint-count laws don't depend on `q` — and a test pins it positive per row.
+1. **`log₂ q` recorded, not displayed.** Each `PARAM_SETS` row carries its Table 1
+   modulus bit-size (`log2q`) and it's transcribed in `PAPER-NOTES.md`, but it is
+   intentionally NOT shown in the UI (it doesn't affect the hint-count laws, which
+   depend only on `(n, h)`). A test still pins it positive per row.
 2. **Hint type now explicit per row.** Each row carries a `validatedHints` note
    transcribed from the paper: the `(2¹⁵,32)` anchor is approximate + perfect; the
    OpenFHE `(2¹⁵,192)` row is perfect-only (approximate not yet validated there);

index.html

@@ -227,7 +227,7 @@ <h2>Parameters &amp; sources</h2>
           <table id="param-table">
             <thead>
               <tr>
-                <th>Regime</th><th>n</th><th>log₂q</th><th>h</th>
+                <th>Regime</th><th>n</th><th>h</th>
                 <th>Prior hints</th><th>New hints</th>
                 <th>Validated hints</th><th>Source</th>
               </tr>
@@ -263,9 +263,9 @@ <h2>Known gaps (honest by construction)</h2>
             anchor covers approximate + perfect.
           </li>
           <li>
-            <strong>q is shown but not modeled.</strong> The <code>log₂ q</code>
-            column above is transcribed from Table 1 for transparency; the
-            hint-count laws don't depend on the modulus, so the math keys off
+            <strong>q is not modeled.</strong> Table 1 lists a modulus bit-size
+            <code>log₂ q</code> per row (recorded in <code>PAPER-NOTES.md</code>);
+            the hint-count laws don't depend on the modulus, so the demo keys off
             <code>(n, h)</code> only.
           </li>
           <li>

src/main.ts

@@ -137,8 +137,7 @@ function renderTable(): void {
         ? '<span class="badge badge-paper">paper</span>'
         : '<span class="badge badge-heuristic">model</span>';
     return (
-      `<tr><td>${s.label}</td><td>2^${Math.round(Math.log2(s.n))}</td>` +
-      `<td>${s.log2q}</td><td>${s.h}</td>` +
+      `<tr><td>${s.label}</td><td>2^${Math.round(Math.log2(s.n))}</td><td>${s.h}</td>` +
       `<td>${fmt(s.hintsPrior)}</td><td>${fmt(s.hintsNew)}</td>` +
       `<td>${s.validatedHints}</td>` +
       `<td>${badge}<br><span class="muted">${s.cite}</span></td></tr>`

src/model.ts

@@ -45,8 +45,9 @@ export interface ParamSet {
   n: number;
   /** Hamming weight h (number of nonzero entries in the sparse ternary secret). */
   h: number;
-  /** Modulus bit-size log2(q) for this regime (Table 1). Informational: the
-   *  hint-count laws do NOT depend on q. Shown for transparency, not used in math. */
+  /** Modulus bit-size log2(q) for this regime (Table 1). Recorded for
+   *  completeness (and PAPER-NOTES.md); NOT displayed in the UI and NOT used in
+   *  any math — the hint-count laws depend only on (n, h). */
   log2q: number;
   /** Hints prior work needed (~n/2). */
   hintsPrior: number;