← Catalogue · The r₁₂ frontier
P30 Partial 3/5 Frontière r₁₂

Cusp de Kato

Le rapport de cusp de Kato atteint 1,9 contre 2,0 exact : la queue corrélée reste approximative. Première mesure de la distance à la frontière r₁₂.

◫ Simulation figure(s)

Figure P30

JSON results — converted to tables

Machine-readable artefacts frozen by SHA-256. Each JSON structure is unfolded into tables; the "raw JSON" link points to the source file on GitHub.

p30_cusp.json 10 table(s) · 40 rows Raw JSON on GitHub
ChampValeur
chantierP30
titreCusp de Kato sature — reponse a deux corps contrainte, 0 parametre

Explanation — context & formalism

P30 — The Kato Cusp

Field: Atomic / Quantum Chemistry Verdict: ⚠️ Partial Success (3/5)

Problem

Is the two-body response derivable? First test: impose the Kato cusp as a kinematic constraint (value 1/2 derived from 1/r₁₂ divergence, zero free parameter) in a saturating factor u(r) = c r/(1+r).

Results

ConfigurationE (Ha)% of HF→exact gap
HF (1-body bound)−2.84770
Split-ζ P27 (MC)−2.848551.5
Jastrow smooth P28 (c_opt = 0)44.3
Cusp saturated c = 1/2, Φ HF fixed−2.793~57
Cusp saturated, (a,b) re-optimized−2.793 (1.73, 1.73)197*
Exact (Pekeris)−2.9037100

*Percentage misleading: the comparison point is the best bound (−2.8485), which the result does not cross.

Key finding

The cusp works at fixed Φ, and reverses at free Φ: when (a,b) are free, the two-body factor replaces radial correlation instead of adding to it. The global minimum prefers (1.73, 1.73), symmetric, without inner-outer screening. Angular and radial correlations are not additive in a fixed-range trial function: they compete.

Failures (documented)

  1. Bound not improved: −2.793 > −2.8485 (P27) — variational criterion C1 fails.
  2. Lever inverted: at optimized (a,b), c_opt = 0 — consequence of (1), not independent signal.
  3. Cause: saturating factor range fixed at 1 a₀; imposing cusp 1/2 at all distances over-correlates the pair at large separation, where the true correlation factor vanishes.

Reading

The cusp is necessary but not sufficient; the kinematic constraint fixes only contact, not range. Freeing range would cure energy but reintroduce an adjusted parameter — exactly the betrayal the method refuses. The result is not confused, it is clear: the cusp is derivable, the range is not; no single-term factor adds angular to radial.

Verdict

CriterionResult
Control validated
Capture > 55% at fixed Φ
Zero parameter
P27 bound crossed
Lever inverted at free Φ

Document chain

Document source

Reports (PDF verdict notes)

</> Simulation — Python scripts

p30_cusp.py

⌗ Cross-references

§ Related glossary entries

Cusp de Kato — Condition exacte sur la dérivée de la fonction d'onde au contact électron–noyau (rapport = 2 pour He). Atteint : 1,9 — la queue corrélée reste approximative.
Frontière r₁₂ — Limite constitutive de la machine : la réponse corrélée continue à deux corps (r₁₂ explicite) n'est pas dérivable sans liberté de forme. Déclarée en P31, loi en P32, renforcée en P33.
Queue asymptotique √(2I) — Décroissance exacte de la densité à grande distance, contrôlée par le potentiel d'ionisation I. Contrainte exacte cumulée en P33.

See the full glossary →