Search for: All records

Generating 3D graphs of symmetry-group equivariance is of intriguing potential in broad applications from machine vision to molecular discovery. Emerging approaches adopt diffusion generative models (DGMs) with proper re-engineering to capture 3D graph distributions. In this paper, we raise an orthogonal and fundamental question of in what (latent) space we should diffuse 3D graphs. ❶ We motivate the study with theoretical analysis showing that the performance bound of 3D graph diffusion can be improved in a latent space versus the original space, provided that the latent space is of (i) low dimensionality yet (ii) high quality (i.e., low reconstruction error) and DGMs have (iii) symmetry preservation as an inductive bias. ❷ Guided by the theoretical guidelines, we propose to perform 3D graph diffusion in a low-dimensional latent space, which is learned through cascaded 2D–3D graph autoencoders for low-error reconstruction and symmetry-group invariance. The overall pipeline is dubbed latent 3D graph diffusion. ❸ Motivated by applications in molecular discovery, we further extend latent 3D graph diffusion to conditional generation given SE(3)-invariant attributes or equivariant 3D objects. ❹ We also demonstrate empirically that out-of-distribution conditional generation can be further improved by regularizing the latent space via graph self-supervised learning. We validate through comprehensive experiments that our method generates 3D molecules of higher validity / drug-likeliness and comparable or better conformations / energetics, while being an order of magnitude faster in training. Codes are released at https://github.com/Shen-Lab/LDM-3DG. 

Thermoelastic loss is an important energy dissipation mechanisms in resonant systems. A careful analysis of the thermoelastic loss is critical to the design of low-noise devices for high-precision applications, such as the mirrors used for gravitational-wave (GW) detectors. In this paper, we present analytical solutions to the thermoelastic loss due to thermoelasticity between different materials that are in contact. We find expressions for the thermoelastic loss of multimaterial coatings of finite substrates, and analyze its dependencies on material properties, mirror design and operating experimental conditions. Our results show that lower operating mirror temperature, thinner layers and higher number of interfaces in the coating, and the choice of the first layer of the coating that minimizes the thermal expansion mismatch with the substrate are strategies that reduce the thermoelastic loss and, therefore, diminish the thermal noise that limits the resolution in sensing applications. The results presented in this paper are relevant for the development of low-noise GW detectors and for other experiments sensitive to energy dissipation mechanisms when different materials are in contact.

 

This paper provides a comprehensive overview of how fitting of baryon acoustic oscillations (BAO) is carried out within the upcoming Dark Energy Spectroscopic Instrument’s (DESI) 2024 results using its DR1 data set, and the associated systematic error budget from theory and modelling of the BAO. We derive new results showing how non-linearities in the clustering of galaxies can cause potential biases in measurements of the isotropic ($\alpha _{\mathrm{iso}}$) and anisotropic ($\alpha _{\mathrm{ap}}$) BAO distance scales, and how these can be effectively removed with an appropriate choice of reconstruction algorithm. We then demonstrate how theory leads to a clear choice for how to model the BAO and develop, implement, and validate a new model for the remaining smooth-broad-band (i.e. without BAO) component of the galaxy clustering. Finally, we explore the impact of all remaining modelling choices on the BAO constraints from DESI using a suite of high-precision simulations, arriving at a set of best practices for DESI BAO fits, and an associated theory and modelling systematic error. Overall, our results demonstrate the remarkable robustness of the BAO to all our modelling choices and motivate a combined theory and modelling systematic error contribution to the post-reconstruction DESI BAO measurements of no more than 0.1 per cent (0.2 per cent) for its isotropic (anisotropic) distance measurements. We expect the theory and best practices laid out to here to be applicable to other BAO experiments in the era of DESI and beyond.

 

In this paper, we propose a Flexible processing-in-DRAM framework named FlexiDRAM that supports the efficient implementation of complex bulk bitwise operations. This framework is developed on top of a new reconfigurable in-DRAM accelerator that leverages the analog operation of DRAM sub-arrays and elevates it to implement XOR2-MAJ3 operations between operands stored in the same bit-line. FlexiDRAM first generates an efficient XOR-MAJ representation of the desired logic and then appropriately allocates DRAM rows to the operands to execute any in-DRAM computation. We develop ISA and software support required to compute in-DRAM operation. FlexiDRAM transforms current memory architecture to a massively parallel computational unit and can be leveraged to significantly reduce the latency and energy consumption of complex workloads. Our extensive circuit-to-architecture simulation results show that averaged across two well-known deep learning workloads, FlexiDRAM achieves ∼15× energy-saving and 13× speedup over the GPU outperforming recent processing-in-DRAM platforms.