The abundant demand for deep learning compute resources has created a renaissance in low-precision hardware. Going forward, it will be essential for simulation software to run on this new generation of machines without sacrificing scientific fidelity. In this paper, we examine the precision requirements of a representative kernel from quantum chemistry calculations: the calculation of the single-particle density matrix from a given mean-field Hamiltonian (i.e., Hartree-Fock or density functional theory) represented in an LCAO basis. We find that double precision affords an unnecessarily high level of precision, leading to optimization opportunities. We show how an approximation built from an error-free matrix multiplication transformation can be used to potentially accelerate this kernel on future hardware. Our results provide a roadmap for adapting quantum chemistry software for the next generation of high-performance computing platforms.

Download full-text PDF

Source
http://dx.doi.org/10.1021/acs.jctc.4c00938DOI Listing

Publication Analysis

Top Keywords

quantum chemistry
12
precision requirements
8
chemistry calculations
8
software generation
8
reducing numerical
4
precision
4
numerical precision
4
requirements quantum
4
calculations abundant
4
abundant demand
4

Similar Publications

In this study, a simple and efficient method for synthesizing nitrogen-doped carbon quantum dots (N-CQDs) has been developed through a one-step hydrothermal process using hedyotis diffusa willd. The morphology, chemical composition, and optical properties of the resulting N-CQDs were thoroughly characterized. The synthesized N-CQDs exhibited a spherical shape with an average particle size of 4.

View Article and Find Full Text PDF

In this study, a sensitive and selective spectrofluorimetric method was developed for the determination of the antidiabetic drug nateglinide based on its reaction with the xanthene dye acid red 87 (AR87). A fluorescence quenching process was observed for the AR87 at 545 nm upon the addition of nateglinide, which was exploited for the quantitative analysis. The reaction mechanism was investigated using quantum mechanical calculations suggesting a transfer between the electron-rich AR87 and the electron-deficient nateglinide.

View Article and Find Full Text PDF

Unprecedented carbonic anhydrase inhibition mechanism: Targeting histidine 64 side chain through a halogen bond.

Arch Pharm (Weinheim)

January 2025

Section of Pharmaceutical and Nutraceutical Sciences, Department of Neuroscience, Psychology, Drug Research and Child Health (NEUROFARBA), University of Florence, Sesto Fiorentino, Firenze, Italy.

2,2'-Thio-bis(4,6-dichlorophenol), namely bithionol, is a small molecule endowed with a multifaceted bioactivity. Its peculiar polychlorinated phenolic structure makes it a suitable candidate to explore its potentialities in establishing interaction patterns with enzymes of MedChem interest, such as the human carbonic anhydrase (hCA) metalloenzymes. Herein, bithionol was tested on a panel of specific hCAs through the stopped-flow technique, showing a promising micromolar inhibitory activity for the hCA II isoform.

View Article and Find Full Text PDF

Coherent manipulation of photochemical spin-triplet formation in quantum dot-molecule hybrids.

Nat Mater

January 2025

State Key Laboratory of Molecular Reaction Dynamics, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian, China.

The interconversion between singlet and triplet spin states of photogenerated radical pairs is a genuine quantum process, which can be harnessed to coherently manipulate the recombination products through a magnetic field. This control is central to such diverse fields as molecular optoelectronics, quantum sensing, quantum biology and spin chemistry, but its effect is typically fairly weak in pure molecular systems. Here we introduce hybrid radical pairs constructed from semiconductor quantum dots and organic molecules.

View Article and Find Full Text PDF

Want AI Summaries of new PubMed Abstracts delivered to your In-box?

Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!