The transferability limits of static benchmarks

Weymuth, Thomas; Reiher, Markus

doi:10.1039/d2cp01725c

Cited by 11 publications

(10 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here, we partition the reference simulation tasks into 𝑟 groups of similar tasks, each with their model errors described by a different Gaussian distribution. Such grouping is appropriate when different groups of tasks are observed to have different error statistics for models under consideration 12 . The small-𝜖 limit of the success measure generalizes from Eq.…”

Section: A Maximum Likelihood Estimationmentioning

confidence: 99%

Model selection in atomistic simulation

Moussa¹

2023

Preprint

View full text Add to dashboard Cite

There are many atomistic simulation methods with very different costs, accuracies, transferabilities, and numbers of empirical parameters. I show how statistical model selection can compare these methods fairly, even when they are very different. These comparisons are also useful for developing new methods that balance cost and accuracy. As an example, I build a semiempirical model for hydrogen clusters.

show abstract

Section: A Maximum Likelihood Estimationmentioning

confidence: 99%

Model selection in atomistic simulation

Moussa¹

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…It will now be important to substantially and dynamically 59 extend the HyDRA database in different directions such as larger downshifts, other compound classes, dihydrates, charged systems or spectral ranges, before another blind challenge might follow. Such a second generation blind challenge may decide more clearly between purely harmonic, anharmonically corrected, fully anharmonic and learning strategies.…”

Section: Future Extensionsmentioning

confidence: 99%

The first HyDRA challenge for computational vibrational spectroscopy

Fischer

Bödecker

Schweer

et al. 2023

Preprint

View full text Add to dashboard Cite

Vibrational spectroscopy in supersonic jet expansions is a powerful tool to assess molecular aggregates in close to ideal conditions for the benchmarking of quantum chemical approaches. The low temperatures achieved as well as the absence of environment effects allow for a direct comparison between computed and experimental spectra. This provides potential benchmarking data which can be revisited to hone different computational techniques, and it allows for the critical analysis of procedures under the setting of a blind challenge. In the latter case, the final result is unknown to modellers, providing an unbiased testing opportunity for quantum chemical models. In this work, we present the spectroscopic and computational results for the first HyDRA blind challenge. The latter deals with the prediction of water donor stretching vibrations in monohydrates of organic molecules. This edition features a test set of 10 systems. Experimental water donor OH vibrational wavenumbers for the vacuum-isolated monohydrates of formaldehyde, tetrahydrofuran, pyridine, tetrahydrothiophene, trifluoroethanol, methyl lactate, dimethylimidazolidinone, cyclooctanone, trifluoroacetophenone and 1-phenylcyclohexane-cis-1,2-diol are provided. The results of the challenge show promising predictive properties in both purely quantum mechanical approaches as well as regression and other machine learning strategies.

show abstract

“…Most of them are far from reaching chemical accuracy but flourish because their computational cost matches our limited resources. Various characteristics of the methods -their past evolution, [1] future development [2] and the ensuing chemical insights, [3] the uncertainty of a quantum chemical result [4] and challenges of benchmarking [5,6] -are a matter of scientific analysis and discussion. Typically, studies make accent on differences among the methods.…”

Section: Introductionmentioning

confidence: 99%

Statistical Equivalence of Quantum Chemical Methods for Energy Distribution in Water Clusters

Tokmachev

2023

ChemPhysChem

View full text Add to dashboard Cite

Methods of quantum chemistry are instrumental in understanding molecular structures and properties. However, the results demonstrate significant variability, which is difficult to predict and rationalize. The fundamental question is whether some molecular systems exhibit properties invariant with respect to the computational method. The idea explored here is that collective properties of statistical ensembles should be more robust than characteristics of individual molecules and their arbitrary sets. This effect is demonstrated for the complete set of hydrogen-bond topologies of the dodecahedral water cluster (H 2 O) 20 . Non-Gaussian energy distributions produced by various methods have the same functional form despite strong differences in mean values and standard deviations. The conclusion is tested on methods of different complexity and origin employing a number of criteria. A linear mapping between the energies produced by different methods is discussed. The significance of the results is in establishing a collective equivalence property of quantum chemical methods.

show abstract

The transferability limits of static benchmarks

Abstract: Every practical method to solve the Schr\"odinger equation for interacting many-particle systems introduces approximations. Such methods are therefore plagued by systematic errors. For computational chemistry, it is decisive to quantify...

Cited by 11 publications

References 32 publications

Model selection in atomistic simulation

Model selection in atomistic simulation

The first HyDRA challenge for computational vibrational spectroscopy

Statistical Equivalence of Quantum Chemical Methods for Energy Distribution in Water Clusters

Contact Info

Product

Resources

About