================================================================================ CIPHER V9 MELTING RESISTANCE SCORING — 2026-04-04 First blind test of eigenvalue-derived melt_resist vs. actual melting points Source: CIPHER_V9_SWEEP_2026-04-04.json (field: melt_resist) No external calibration. No curve fitting. Pure geometric prediction. ================================================================================ METHODOLOGY ----------- - melt_resist is the cipher's predicted melting resistance (higher = harder to melt) - Known melting points (K) from standard reference data - Spearman rank correlation: tests whether the ORDERING is correct (we care about ranking, not absolute values — no calibration was applied) - Tied predicted values receive averaged ranks - Simple Spearman formula: rho = 1 - 6*sum(d_i^2) / (n*(n^2-1)) Note: ties present in predicted ranks; correction is small (<0.005 rho) ================================================================================ PERIOD 4 D-BLOCK (Sc through Zn) ================================================================================ RAW DATA: Element Z d_pos final_arch cross_term melt_resist Actual_Tm(K) ------- -- ----- ---------- ---------- ----------- ------------ Sc 21 1 HCP root 0.6319 1814 Ti 22 2 HCP root 0.6319 1941 V 23 3 BCC consonant 1.2000 2183 Cr 24 4 BCC consonant 1.2000 2180 Mn 25 5 BCC tension 0.9000 1519 Fe 26 6 BCC tension 0.9000 1811 Co 27 7 HCP tension 0.5688 1768 Ni 28 8 FCC consonant 0.5417 1728 Cu 29 9 FCC consonant 0.5417 1358 Zn 30 10 HCP tension 0.5688 693 PREDICTED ORDER (descending melt_resist): 1-2. V, Cr (1.2000) -- tied 3-4. Mn, Fe (0.9000) -- tied 5-6. Sc, Ti (0.6319) -- tied 7-8. Co, Zn (0.5688) -- tied 9-10. Ni, Cu (0.5417) -- tied ACTUAL ORDER (descending Tm): 1. V (2183 K) 2. Cr (2180 K) 3. Ti (1941 K) 4. Sc (1814 K) 5. Fe (1811 K) 6. Co (1768 K) 7. Ni (1728 K) 8. Mn (1519 K) 9. Cu (1358 K) 10. Zn (693 K) SPEARMAN RANK CORRELATION: Element Pred_Rank Actual_Rank d d^2 ------- --------- ----------- ------ ------ Sc 5.5 4 +1.5 2.25 Ti 5.5 3 +2.5 6.25 V 1.5 1 +0.5 0.25 Cr 1.5 2 -0.5 0.25 Mn 3.5 8 -4.5 20.25 Fe 3.5 5 -1.5 2.25 Co 7.5 6 +1.5 2.25 Ni 9.5 7 +2.5 6.25 Cu 9.5 9 +0.5 0.25 Zn 7.5 10 -2.5 6.25 Sum(d^2) = 46.50 n = 10 rho = 1 - 6(46.50) / (10)(99) = 1 - 279/990 = 0.718 >>> PERIOD 4 SPEARMAN rho = +0.718 <<< HITS (correct ordering calls): + V and Cr correctly identified as the TOP TWO (actual #1 and #2) + Cu correctly placed near the bottom (pred 9-10, actual 9) + Zn correctly placed at or near bottom (pred 7-8, actual 10) + General shape: high in early d-block, declining toward end = CORRECT MISSES (significant misplacements): - Mn: predicted rank 3-4, actual rank 8. WORST MISS (off by 4.5 ranks) Mn has the half-filled d5 anomaly — known problematic element. Cipher reads BCC+tension but cannot distinguish the d5 stability dip. - Fe: predicted rank 3-4, actual rank 5. Grouped with Mn by symmetry. - Ti: predicted rank 5-6, actual rank 3. Slightly undervalued. - Ni: predicted rank 9-10, actual rank 7. Slightly undervalued. DIAGNOSIS: The cipher groups elements by archetype (BCC > HCP > FCC) and cross_term (consonant > tension > root). This captures the BROAD shape but creates degenerate pairs that cannot be split without additional physics. The Mn anomaly (half-filled d-shell stability reducing cohesion despite high symmetry) is the biggest single failure — this is the "cooling phase" the cipher cannot yet read. ================================================================================ PERIOD 5 D-BLOCK (Y through Cd) ================================================================================ RAW DATA: Element Z d_pos final_arch cross_term melt_resist Actual_Tm(K) ------- -- ----- ---------- ---------- ----------- ------------ Y 39 1 HCP root 0.6319 1799 Zr 40 2 HCP root 0.6319 2128 Nb 41 3 BCC consonant 1.2000 2750 Mo 42 4 BCC consonant 1.2000 2896 Tc 43 5 HCP* consonant 1.3000 2430 Ru 44 6 HCP* consonant 1.3000 2607 Rh 45 7 FCC root 0.4514 2237 Pd 46 8 FCC consonant 0.5417 1828 Ag 47 9 FCC consonant 0.5417 1235 Cd 48 10 HCP tension 0.5688 594 (* = corrected from BCC base) PREDICTED ORDER (descending melt_resist): 1-2. Tc, Ru (1.3000) -- tied 3-4. Nb, Mo (1.2000) -- tied 5-6. Y, Zr (0.6319) -- tied 7. Cd (0.5688) 8-9. Pd, Ag (0.5417) -- tied 10. Rh (0.4514) ACTUAL ORDER (descending Tm): 1. Mo (2896 K) 2. Nb (2750 K) 3. Ru (2607 K) 4. Tc (2430 K) 5. Rh (2237 K) 6. Zr (2128 K) 7. Pd (1828 K) 8. Y (1799 K) 9. Ag (1235 K) 10. Cd (594 K) SPEARMAN RANK CORRELATION: Element Pred_Rank Actual_Rank d d^2 ------- --------- ----------- ------ ------ Y 5.5 8 -2.5 6.25 Zr 5.5 6 -0.5 0.25 Nb 3.5 2 +1.5 2.25 Mo 3.5 1 +2.5 6.25 Tc 1.5 4 -2.5 6.25 Ru 1.5 3 -1.5 2.25 Rh 10.0 5 +5.0 25.00 Pd 8.5 7 +1.5 2.25 Ag 8.5 9 -0.5 0.25 Cd 7.0 10 -3.0 9.00 Sum(d^2) = 60.00 n = 10 rho = 1 - 6(60.00) / (10)(99) = 1 - 360/990 = 0.636 >>> PERIOD 5 SPEARMAN rho = +0.636 <<< HITS: + Nb and Mo correctly identified in the top tier (actual #1 and #2) + Tc and Ru correctly identified as very high (actual #3 and #4) + Ag correctly placed near bottom (pred 8-9, actual 9) + Cd correctly placed near bottom (pred 7, actual 10) + The top-4 elements are all in the predicted top-4 (just reordered) MISSES: - Rh: predicted LAST (rank 10), actual rank 5. WORST MISS (off by 5 ranks) Rh is FCC with cross_term "root" — the cipher reads it as geometrically weak, but Rh has very strong metallic bonding (4d8 config). This is the FCC blind spot: cipher treats all FCC as low-resistance, but late-transition FCC metals can have very high cohesive energy. - Tc vs Mo ordering inverted: cipher gives Tc > Mo (due to HCP correction bonus), but actually Mo > Tc. The correction overcorrects. - Cd: predicted rank 7, actual rank 10. Should be lower. DIAGNOSIS: The HCP correction (giving Tc and Ru a 1.3 bonus) pushes them ABOVE Mo/Nb, which is wrong — Mo is the actual period-5 peak. The correction captures that these elements are refractory but overshoots. The Rh failure is structural: FCC + root cross_term = lowest tier, but Rh is actually mid-pack. The cipher has no mechanism to distinguish "noble metal FCC" from "soft metal FCC." ================================================================================ PERIOD 6 D-BLOCK (Hf through Hg) ================================================================================ RAW DATA: Element Z d_pos final_arch cross_term melt_resist Actual_Tm(K) ------- -- ----- ---------- ---------- ----------- ------------ Hf 72 2 HCP root 0.6319 2506 Ta 73 3 BCC consonant 1.2000 3290 W 74 4 BCC consonant 1.2000 3695 Re 75 5 HCP* consonant 1.3000 3459 Os 76 6 HCP* consonant 1.3000 3306 Ir 77 7 FCC root 0.4514 2739 Pt 78 8 FCC consonant 0.5417 2041 Au 79 9 FCC consonant 0.5417 1337 Hg 80 10 A7* root 0.6250 234 (* = corrected from base archetype) PREDICTED ORDER (descending melt_resist): 1-2. Re, Os (1.3000) -- tied 3-4. Ta, W (1.2000) -- tied 5. Hf (0.6319) 6. Hg (0.6250) 7-8. Pt, Au (0.5417) -- tied 9. Ir (0.4514) ACTUAL ORDER (descending Tm): 1. W (3695 K) 2. Re (3459 K) 3. Os (3306 K) 4. Ta (3290 K) 5. Ir (2739 K) 6. Hf (2506 K) 7. Pt (2041 K) 8. Au (1337 K) 9. Hg (234 K) SPEARMAN RANK CORRELATION: Element Pred_Rank Actual_Rank d d^2 ------- --------- ----------- ------ ------ Hf 5.0 6 -1.0 1.00 Ta 3.5 4 -0.5 0.25 W 3.5 1 +2.5 6.25 Re 1.5 2 -0.5 0.25 Os 1.5 3 -1.5 2.25 Ir 9.0 5 +4.0 16.00 Pt 7.5 7 +0.5 0.25 Au 7.5 8 -0.5 0.25 Hg 6.0 9 -3.0 9.00 Sum(d^2) = 35.50 n = 9 rho = 1 - 6(35.50) / (9)(80) = 1 - 213/720 = 0.704 >>> PERIOD 6 SPEARMAN rho = +0.704 <<< HITS: + Re and Os correctly identified in the top tier (actual #2 and #3) + Ta correctly identified as very high (pred 3-4, actual 4) + Pt correctly placed at rank 7 (actual 7) — near-perfect + Au correctly placed at rank 8 (actual 8) — near-perfect + Hf correctly placed mid-pack (pred 5, actual 6) + The bottom three (Pt, Au, Hg) are correctly ordered relative to each other MISSES: - W: predicted rank 3-4 (tied with Ta), actual rank 1. SIGNIFICANT MISS. W is THE highest-melting elemental metal. The cipher ties it with Ta because both are BCC+consonant. Cannot distinguish W's exceptional cohesive energy from Ta. - Ir: predicted LAST (rank 9), actual rank 5. WORST MISS (off by 4 ranks) Same FCC blind spot as Rh in period 5. Ir is FCC+root = bottom tier in the cipher, but Ir has extraordinary cohesive strength (5d7). - Hg: predicted rank 6, actual rank 9. Off by 3 ranks. The A7 correction gives Hg a melt_resist of 0.625, which is too high. Hg is the ONLY metal that is liquid at room temperature (234 K). The correction helps structurally but hurts the melting prediction. DIAGNOSIS: Same pattern as period 5: the HCP correction pushes Re/Os above W, which inverts the peak. W should be #1 but the cipher cannot separate it from Ta. The Ir failure mirrors Rh exactly — FCC noble metals are systematically undervalued. Hg's A7 correction helps the structure assignment but gives it too much melting credit. ================================================================================ CROSS-PERIOD SUMMARY ================================================================================ Period n Spearman rho Interpretation ------ -- ------------ -------------- 4 10 +0.718 Strong positive correlation 5 10 +0.636 Moderate-strong positive correlation 6 9 +0.704 Strong positive correlation OVERALL WEIGHTED AVERAGE: rho = (10*0.718 + 10*0.636 + 9*0.704) / 29 = (7.18 + 6.36 + 6.336) / 29 = 19.876 / 29 = +0.685 For a ZERO-PARAMETER geometric prediction with NO calibration against melting point data, rho ~ 0.69 across 29 d-block elements is a meaningful signal. This is well above chance (rho = 0 expected for random ordering). ================================================================================ SYSTEMATIC PATTERNS (DIAGNOSTIC) ================================================================================ 1. THE ARCHETYPE STAIRCASE WORKS The cipher's melt_resist hierarchy is: corrected-HCP (1.3) > BCC-consonant (1.2) > BCC-tension (0.9) > HCP-root (0.63) > HCP-tension (0.57) > FCC-consonant (0.54) > FCC-root (0.45). This staircase roughly tracks actual melting point trends across d-block periods. The broad shape is correct: mid-d-block elements (d_pos 3-6) melt highest, ends melt lowest. 2. THE FCC BLIND SPOT (Rh, Ir) Rh (period 5, d_pos=7) and Ir (period 6, d_pos=7) are both FCC with cross_term "root" — the cipher's lowest melting tier. But both are actually among the HIGHEST melting elements in their period (Rh #5/10, Ir #5/9). The cipher treats FCC geometry as inherently low-resistance to melting, which fails for noble metals with strong d-band cohesion. This is the SAME blind spot identified in the cooling-phase analysis: the cipher reads the accumulation/peak geometry but not the reorganization strength of late d-block FCC metals. 3. THE HCP CORRECTION OVERCORRECTS Elements corrected from BCC to HCP (Tc, Ru, Re, Os) get melt_resist = 1.3, which is ABOVE the BCC-consonant value of 1.2. In reality: - Period 5: Mo (1.2, BCC) > Tc (1.3, corrected) — WRONG, Mo melts higher - Period 6: W (1.2, BCC) > Re (1.3, corrected) — WRONG, W melts higher The correction captures that these elements are refractory but should not exceed the BCC peak. The correction should bring them TO the BCC level, not ABOVE it. 4. DEGENERATE PAIRS LIMIT RESOLUTION The cipher produces only ~6 distinct melt_resist values across 10 elements. Most elements are tied in pairs or groups. This means the cipher can place elements into the RIGHT TIER but cannot order within tiers. A finer-grained metric (e.g., incorporating ramp_frac as a continuous modifier rather than discrete archetype bins) would break these ties. 5. THE Mn ANOMALY (HALF-FILLED d-SHELL) Mn (d_pos=5) has the half-filled 3d5 configuration. Its melting point (1519 K) is anomalously LOW for a BCC element — lower than Sc and Ti (both HCP). The cipher assigns Mn to BCC-tension with melt_resist = 0.9, which is reasonable for a BCC element but too high relative to reality. The half-filled shell stability reduces cohesive energy in a way the cipher cannot capture from geometry alone. 6. THE Hg ANOMALY (LIQUID METAL) Hg (234 K) is the only metal liquid at room temperature. The cipher gives it melt_resist = 0.625 (via A7 correction), placing it ABOVE the FCC metals Pt and Au. In reality Hg is 6x lower than Au. The cipher reads Hg's distorted structure but cannot read the relativistic contraction that weakens Hg's metallic bonding. This is an inherent limit: the cipher reads geometry, but Hg's weakness comes from ELECTRONIC structure (6s inert pair effect). ================================================================================ RECOMMENDATIONS FOR CIPHER V10 ================================================================================ 1. CAP THE HCP CORRECTION at 1.2 (equal to BCC-consonant, not above it). This alone would fix the Mo/Tc and W/Re inversions. 2. ADD A d-ELECTRON COUNT MODIFIER to break degenerate pairs. The ramp_frac field is already computed — use it as a continuous tie-breaker within archetype groups. Elements with ramp_frac closer to the bulge should get higher melt_resist. 3. CREATE A LATE-FCC EXCEPTION for d_pos = 7 (Rh, Ir position). These elements have anomalously strong bonding for FCC metals. The cipher could flag d_pos=7 FCC as "cohesive FCC" vs. d_pos=8-10 as "noble/soft FCC." 4. ADD A RELATIVISTIC FLAG for period 6, d_pos >= 9. Hg's anomaly is fundamentally electronic (6s inert pair) and may not be capturable from geometry alone. A flag acknowledging this limit would be honest rather than forcing a geometric explanation. 5. CONSIDER THE HALF-FILLED SHELL as a geometric feature. d_pos=5 has {5}-fold symmetry in the d-orbital occupation. If the cipher already recognizes {5} as a stabilizing geometry (cf. fullerene work), it could treat d_pos=5 as a local minimum in cohesive energy — the {5} stabilizes the ELECTRON CONFIGURATION but destabilizes the LATTICE (reduced bonding overlap). ================================================================================ VERDICT ================================================================================ This is a PASS with caveats. For a first-principles geometric prediction with zero calibration: - rho ~ 0.69 across 29 elements is a real signal, not noise - The broad shape (peak at d_pos 3-6, decline at ends) is correct - The top tier (refractory metals) is consistently identified - The bottom tier (Cu, Ag, Au, Zn, Cd, Hg) is mostly correct But: - 6 distinct values for 10 elements = too coarse - FCC noble metals (Rh, Ir) are systematically misplaced - The HCP correction overshoots the BCC peak - Special cases (Mn half-shell, Hg relativistic) are out of reach The cipher reads the GEOMETRY of melting correctly. What it cannot read is the ENERGETICS within a given geometry — which is exactly the "cooling phase gap" identified in earlier analysis. Next test: apply recommendations 1-3 (cap correction, ramp_frac modifier, late-FCC exception) and re-score. Expected improvement: rho > 0.80. ================================================================================ END OF SCORING ================================================================================