Banca de QUALIFICAÇÃO: VINÍCIUS NUNES DA ROCHA

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
STUDENT : VINÍCIUS NUNES DA ROCHA
DATE: 09/06/2025
TIME: 09:00
LOCAL: Ambiente remoto
TITLE:

Machine Learning Implementation in Quantum Chemical Methods for Investigation of Photochemical Reactions of Interstellar Glycine

KEY WORDS:

astrochemistry, photochemistry, glycine ice, machine learning, ab initio, DFT

PAGES: 29
BIG AREA: Ciências Exatas e da Terra
AREA: Química
SUBÁREA: Físico-Química
SPECIALTY: Química Teórica
SUMMARY:

Experimental and theoretical studies investigate the interstellar origin of glycine, focusing on its formation, stability, and photodegradation in environments simulating the interstellar medium. Advances in theoretical and computational methods allow the investigation of the photochemical dynamics of molecules, combining classical and quantum approaches to treat excited states and non-adiabatic processes. Machine Learning has revolutionized Computational Chemistry by enabling the efficient estimation of complex molecular properties, including those of photochemical systems, based on high-quality data and well-validated models. Interstellar evidence of glycine and methylamine reinforces the relevance of astrochemistry, while Machine Learning continues to revolutionize the modeling of photochemical and quantum-dynamic processes with promising and still underexplored applications. Recent literature reveals a growing diversity of approaches, from deep neural networks to hybrid models with DFT or Coupled Cluster, covering excited systems, radicals, and non-adiabatic dynamics. This project proposes the development of an innovative ML-QC hybrid methodology to estimate electronic energies, rate coefficients, and PEHSs for glycine and its radical cation, integrating physicochemical data, reference models, and statistical validation. Machine Learning models, grounded in statistical theory, will be trained with data (e.g., chemical descriptors) to estimate complex patterns, with a focus on supervised, unsupervised, and hybrid approaches applied to highly variable chemical systems coupled with state-of-the-art quantum-chemical methods for elucidating complex physicochemical systems, scientifically relevant to modern society. The implementation will be carried out in Python using the Scikit-learn library for modeling and validation of machine learning algorithms, with support from the ORCA software for quantum data generation, if necessary.

 

COMMITTEE MEMBERS:
Presidente - 3131055 - GLADSON DE SOUZA MACHADO
Externo ao Programa - 1808284 - MARCIO SOARES PEREIRA - UFRRJExterno à Instituição - SILMAR ANDRADE DO MONTE - UFPB