Speaker
Koji Hashimoto
(Kyoto University)
Description
Machine learning technologies has gained a great advance to affect various fields of research in physics, and nonperturbative QCD is not an exception. Here in this talk I will rephrase the AdS/CFT correspondence by a deep learning architecture, and demonstrate the emergence of the gravity geometry by using QCD data, thus establishing a possible duality between QCD and a gravitational theory. This solves the inverse problem of the AdS/CFT, in other words, it is a QFT-driven holographic modeling. The lattice data of the chiral condensate is used to train a neural network to make a bulk gravity model, and the model can predict Wilson loop expectation values, to be well compared with the lattice Wilson loop results.
Author
Koji Hashimoto
(Kyoto University)
