Supervised learning of few dirty bosons with variable particle number

Abstract

We investigate the supervised machine learning of few interacting bosons in optical speckle disorder via artificial neural networks. The learning curve shows an approximately universal power-law scaling for different particle numbers and for different interaction strengths. We introduce a network architecture that can be trained and tested on heterogeneous datasets including different particle numbers. This network provides accurate predictions for all system sizes included in the training set and, by design, is suitable to attempt extrapolations to (computationally challenging) larger sizes. Notably, a novel transfer-learning strategy is implemented, whereby the learning of the larger systems is substantially accelerated and made consistently accurate by including in the training set many small-size instances.

arXiv commentSubmission to SciPost

References in corpus (28)

• Adam: A Method for Stochastic Optimization
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Quantum Machine Learning
• Machine learning and the physical sciences
• Anderson localization of a non-interacting Bose-Einstein condensate
• Quantum machine learning in feature Hilbert spaces
• Direct observation of Anderson localization of matter-waves in a controlled disorder
• Machine Learning Unifies the Modelling of Materials and Molecules
• By-passing the Kohn-Sham equations with machine learning
• Machine learning prediction errors better than DFT accuracy
• Finding Density Functionals with Machine Learning
• Quantum machine learning: a classical perspective
• Deterministic Preparation of a Tunable Few-Fermion System
• How to represent crystal structures for machine learning: towards fast prediction of electronic properties
• From Few to Many: Observing the Formation of a Fermi Sea One Atom at a Time
• Understanding molecular representations in machine learning: The role of uniqueness and target similarity
• Ultracold Bose Gases in 1D Disorder: From Lifshits Glass to Bose-Einstein Condensate
• Machine Learning for Quantum Matter
• Deep Learning and Density Functional Theory
• Weakly interacting Bose gas in a random environment
• Deep learning the Hohenberg-Kohn maps of Density Functional Theory
• Collective dynamics and expansion of a Bose-Einstein condensate in a random potential
• Extensive deep neural networks for transferring small scale learning to large scale systems
• Quantum machine learning and quantum biomimetics: A perspective
• Artificial neural networks for density-functional optimizations in fermionic systems
• Scalable neural networks for the efficient learning of disordered quantum systems
• Numerically exact approach to few-body problems far from a perturbative regime
• Few-boson localization in a continuum with speckle disorder

Cited by in corpus (8)

• Supervised learning of random quantum circuits via scalable neural networks
• Machine learning for structure-property relationships: Scalability and limitations
• Quantum Reservoir Computing for Speckle-Disorder Potentials
• Deep learning density functionals for gradient descent optimization
• Machine learning one-dimensional spinless trapped fermionic systems with neural-network quantum states
• Deep learning nonlocal and scalable energy functionals for quantum Ising models
• Synergy between noisy quantum computers and scalable classical deep learning
• Challenges and opportunities in the supervised learning of quantum circuit outputs