inZOR-ND vs Random Search: Comparative Validation for CASSCF Active Space Discovery on Cr₂ with Multiple Geometries

Read full text Summarise with VIRAT
Authors: Dumitru Novic
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

We present a controlled comparative validation of the inZOR-ND bio-adaptive discovery engine against random search for automatic CASSCF(8,8) active space selection. Building on two previous studies — N₂ dissociation (6-31G, C(18,6) = 18,564 candidates) and Cr₂ single-geometry (STO-3G, C(36,8) = 30.26M candidates) — we design a harder benchmark: Cr₂ with 3 simultaneous geometries (R = 1.68, 2.0, 2.5 Å), requiring the active space to converge at all bond lengths. The multi-geometry constraint reduces the convergence rate from ~100% (single geometry) to only 8.8%, and eliminates solution degeneracy: from ~77 degenerate optima (single geometry) to exactly 1. Under these conditions, inZOR-ND finds the unique optimal active space with fitness 2064.493, while random search stagnates at 2064.437 — a gap of 0.056 Hartree (35 kcal/mol), far beyond chemical accuracy (1 kcal/mol). Key results: - inZOR-ND best fitness: 2064.493 vs Random: 2064.437 - Energy gap: 0.056 Eh = 35 kcal/mol (35× chemical accuracy) - inZOR-ND convergence rate: 8.8% vs Random: 1.1% (8× higher) - Only 1 degenerate optimal solution out of 30.26M candidates - 11 progressive fitness improvements (inZOR-ND) vs stagnation (Random) - inZOR-ND engine: unmodified - Third study in series: N₂ → Cr₂ → Cr₂ comparative ``` https://dumitrunovic-svg.github.io/inZOR-ND/tests/active_space_comparative/index.html

Fields of Study & Subjects

Advanced Chemical Physics Studies Machine Learning in Materials Science Computational Drug Discovery Methods

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
20 March 2026
Language
EN
Document type
Preprint