Dados do Trabalho

Título

QUANTUM COMPUTING AS A TOOL TO IDENTIFY GYPSUM CONTENT IN PERUVIAN SOILS

Resumo

Gypsum-containing soils have a negative impact on agricultural production, so determining the amount of gypsum is necessary to mitigate these effects. Unfortunately, few farmers request their analysis and it is not possible to include it as a routine analysis due to its high cost. An alternative is to produce models that determine gypsum content using data from legacy data. New quantum computing algorithms are an opportunity to create models with better performance and lower computational costs. We aimed at using quantum neural networks (QNN) to model equivalent gypsum of Peruvian soils and compare it with a multi layer perceptron neural network (MLP-NN). Two thousand and hundred samples with equivalent gypsum contents between 0 and 3 % were obtained from legacy data of different regions of Peru. The IBM Quantum Lab Service and qiskit machine-learning were used to model QNN and keras library for MLP-NN. Principal component analysis was applied to reduce the variables to 3, train (80 %) and test (20 %) data split and normalized root mean square error were used to validate model performance. The results showed that there were no improvements in prediction error and time performance were not superior in quantum computing. The main drawback was the QNN's optimization, which highlights the challenge it will be for new pedometric scientists to train in quantum computing. The developed models can help regional labs in Peru determine the equivalent gypsum content and advise farmers to conduct a chemical analysis for more precise information.

Palavras-chave

saline soils; quantum machine learning; pedometrics.

Instituição financiadora

LASPAF of the Universidad Nacional Agraria La Molina

Agradecimentos

We acknowledge the use of IBM Quantum services, sample access by the soil laboratory (LASPAF) of the Universidad Nacional Agraria La Molina, collaboration of the engineering doctoral program at the Pontificia Universidad Católica del Perú and the Programa de Pós Graduação em Ecossistemas Agrícolas e Naturais at the Universidade Federal de Santa Catarina.