Experiments · E248d

With the temperature grid extended to 30 K, does the ordering rung get every known answer right?

Yes. All nine hold: the 95 K separation now fails, a 60 K separation passes, and every ordering answer is unchanged. The 95 K margin is thin, and is stated.

In the log: extend rung 1's grid down to 30 K (pre-registered 2026-09-27 23:21)

confirmedDate 2026-09-27 23:21, as written in the logrung 1 · ordering0 predictions · 1 result paragraphEXPERIMENTS.md lines 17479–17494, lines 17496–17519

Pre-registration

The pre-registration, as written

E248d — extend rung 1's grid down to 30 K (pre-registered 2026-09-27 23:21)

Change. The geometric grid continues from 60 K down to 30 K at the same ratio (0.92). Then any separation with T_c ≥ 90 K, whose peak reads at ≥ 0.64 × 90 ≈ 58 K, lies inside the sweep with room below it. Cost: about 8 more temperatures per leg, most of them cheap and near-frozen.

Known answers, on the new grid (fresh sampling; the Isings are cheap):

  • clustering Ising 95 K, seeds 8 and 9: p_rung1 ≤ 0.1 (the E248c false pass must go);
  • clustering Ising 60 K, seeds 8 and 9: p_rung1 ≥ 0.9 (a separation below the window is allowed, and must not be failed);
  • clustering Ising 300 K, seeds 8 and 9: corrected T within 10 %;
  • the ordering Isings K1, K4, K2a, K5 and K2b and the ring, seeds 6 and 7: E247's bars.
  • Mo–Ta on e6 is not re-sampled. Its transition is at 1120 K, and the added temperatures lie in its frozen, ordered region, so its verdict cannot move. This is stated, not tested.

Prediction: every bar holds.

Results

EXPERIMENTS.md · line 17496

E248d RESULT (2026-09-28 00:51). runs/e248d/, all on the 30 K grid, read under E247 with the separation factor where the cold state clusters. Every bar holds; the prediction holds.

known answer clusters p_rung1 T read bar
K1 Ising 1410 K no 0.000 1425 p ≤ 0.01 holds
K4 Ising 1050 K no 0.000 1076 p ≤ 0.01 holds
K2a Ising 500 K no 0.000 512 p ≤ 0.01 holds
K5 Ising 90 K no 0.160 92.6 0.1–0.9 holds
K2b Ising 50 K no 1.000 50.5 (now a peak inside the sweep) p ≥ 0.9 holds
1D ring no 0.000 broad, not established p < 0.5 holds
separation 95 K yes 0.000 89.6, broad: not established p ≤ 0.1 holds (was the E248c false pass)
separation 60 K yes 1.000 56.7 p ≥ 0.9 holds
separation 300 K yes 0.000 304.5 within 10 % holds
  • The ordering answers read exactly as before above 60 K (same seeds and temperatures).
  • The 50 K Ising, read before as "still rising at the floor", is now a resolved peak at 50.5 K.
  • Stated margin: the 95 K separation reads 89.6 K after correction, just under 90 K. It fails because its broad feature reaches the window, not on its temperature alone. Separation on this cell always reads broad (FWHM/T 0.61–0.71 at every temperature tried), so the rule is systematic rather than lucky, but the margin is thin. A factor fitted per temperature, or a larger cell, would widen it.

The review's first rung-1 question is closed. Rung 1 now fails decomposition inside the window, passes it below the window, and reads its temperature within 5 % from 60 to 300 K.

The full record

This entry is written in 2 separate places in the log, shown here in log order.

EXPERIMENTS.md · lines 17479–17494

E248d — extend rung 1's grid down to 30 K (pre-registered 2026-09-27 23:21)

Change. The geometric grid continues from 60 K down to 30 K at the same ratio (0.92). Then any separation with T_c ≥ 90 K, whose peak reads at ≥ 0.64 × 90 ≈ 58 K, lies inside the sweep with room below it. Cost: about 8 more temperatures per leg, most of them cheap and near-frozen.

Known answers, on the new grid (fresh sampling; the Isings are cheap):

  • clustering Ising 95 K, seeds 8 and 9: p_rung1 ≤ 0.1 (the E248c false pass must go);
  • clustering Ising 60 K, seeds 8 and 9: p_rung1 ≥ 0.9 (a separation below the window is allowed, and must not be failed);
  • clustering Ising 300 K, seeds 8 and 9: corrected T within 10 %;
  • the ordering Isings K1, K4, K2a, K5 and K2b and the ring, seeds 6 and 7: E247's bars.
  • Mo–Ta on e6 is not re-sampled. Its transition is at 1120 K, and the added temperatures lie in its frozen, ordered region, so its verdict cannot move. This is stated, not tested.

Prediction: every bar holds.

EXPERIMENTS.md · lines 17496–17519

E248d RESULT (2026-09-28 00:51). runs/e248d/, all on the 30 K grid, read under E247 with the separation factor where the cold state clusters. Every bar holds; the prediction holds.

known answer clusters p_rung1 T read bar
K1 Ising 1410 K no 0.000 1425 p ≤ 0.01 holds
K4 Ising 1050 K no 0.000 1076 p ≤ 0.01 holds
K2a Ising 500 K no 0.000 512 p ≤ 0.01 holds
K5 Ising 90 K no 0.160 92.6 0.1–0.9 holds
K2b Ising 50 K no 1.000 50.5 (now a peak inside the sweep) p ≥ 0.9 holds
1D ring no 0.000 broad, not established p < 0.5 holds
separation 95 K yes 0.000 89.6, broad: not established p ≤ 0.1 holds (was the E248c false pass)
separation 60 K yes 1.000 56.7 p ≥ 0.9 holds
separation 300 K yes 0.000 304.5 within 10 % holds
  • The ordering answers read exactly as before above 60 K (same seeds and temperatures).
  • The 50 K Ising, read before as "still rising at the floor", is now a resolved peak at 50.5 K.
  • Stated margin: the 95 K separation reads 89.6 K after correction, just under 90 K. It fails because its broad feature reaches the window, not on its temperature alone. Separation on this cell always reads broad (FWHM/T 0.61–0.71 at every temperature tried), so the rule is systematic rather than lucky, but the margin is thin. A factor fitted per temperature, or a larger cell, would widen it.

The review's first rung-1 question is closed. Rung 1 now fails decomposition inside the window, passes it below the window, and reads its temperature within 5 % from 60 to 300 K.

Related entries

Built with PRISMWebsite and visualizations made using Claude