Using quantum information theory tools to measure the quality of active spaces

• Grant number: 21/09907-5
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate (Direct)
• Start date: May 31, 2022
• End date: October 30, 2022
• Field of knowledge: Physical Sciences and Mathematics - Chemistry - Physical-Chemistry
• Principal Investigator: Yuri Alexandre Aoto
• Grantee: Matheus Morato Ferreira de Moraes
• Supervisor: Katharina Boguslawski
• Host Institution: Centro de Matemática, Computação e Cognição (CMCC). Universidade Federal do ABC (UFABC). Ministério da Educação (Brasil). Santo André , SP, Brazil
• Institution abroad: Nicolaus Copernicus University, Poland
• Associated to the scholarship: 18/14629-1 - Diatomic molecules containing cobalt and their excited states, BP.DD

Abstract

During the first years of PhD project we proposed a set of approaches to generate multiconfigurational reference wave functions able to describe both ground and excited states for molecules containing transition metals, without the need of a double d-shell active space. The use of these wave functions as reference in more correlated methods results in accurate descriptions of the electronic structure with a considerable computational cost reduction. Based on the initial good results and the potential for improvement, this study aim at increase the robustness and accuracy of these approaches. The use of orbital correlation and parameters calculated using concepts from quantum information theory will be explored to obtain a quantitative measure, or diagnostic, of quality for reference wave functions and, consequently, for active spaces. To ground the conclusions, those measures will be also compared with more accurate theoretical results, obtained using the density-matrix renormalization group (DMRG) method. This algorithm has shown good results, but is not widely applied in the Brazilian theoretical chemistry community. Is expected that one or more numerical diagnostics can be used to measure the quality of the generated reference wave functions. What should allow the reduction of necessary calculations to assure the accuracy of the final results. These approaches may be useful tools to study electronic properties of large complexes what, nowadays, is impracticable with highly accurate methodologies. (AU)