Kevin Rossi

Dr. Kevin Rossi


Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
16 February 2021 in The Journal of Chemical Physics
Verified
Journal Article
Closed access
PDF file
GIGiulio Imbalzano
Giulio Imbalzano
YZYongbin Zhuang
Yongbin Zhuang
VKVenkat Kapil
Venkat Kapil
KRKevin Rossi
Kevin Rossi
EEEdgar A. Engel
Edgar A. Engel
MCMichele Ceriotti
Michele Ceriotti
Machine-learning models have emerged as a very effective strategy to sidestep time-consuming electronic-structure calculations, enabling accurate simulations of greater size, time scale, and complexity. Given the interpolative nature of these models, the reliability of predictions depends on the position in phase space, and it is crucial to obtain an estimate of the error that derives from the finite number of reference structures included during model training. When using a machine-learning potential to sample a finite-temperature ensemble, the uncertainty on individual configurations translates into an error on thermodynamic averages and leads to a loss of accuracy when the simulation enters a previously unexplored region. Here, we discuss how uncertainty quantification can be used, together with a baseline energy model, or a more robust but less accurate interatomic potential, to obtain more resilient simulations and to support active-learning strategies. Furthermore, we introduce an on-the-fly reweighing scheme that makes it possible to estimate the uncertainty in thermodynamic averages extracted from long trajectories. We present examples covering different types of structural and thermodynamic properties and systems as diverse as water and liquid gallium.

17 December 2020 in Nanoscale
Verified
Journal Article
Open access
PDF file
LDLaia Delgado-Callico
Laia Delgado-Callico
KRKevin Rossi
Kevin Rossi
RPRaphael Pinto-Miles
Raphael Pinto-Miles
PSPascal Salzbrenner
Pascal Salzbrenner
FBFrancesca Baletto
Francesca Baletto
Predicting when phase changes occur in nanoparticles is fundamental for designing the next generation of devices suitable for catalysis, biomedicine, optics, chemical sensing and electronic circuits. The estimate of the temperature at which metallic nanoparticles become liquid is, however, a challenge and a standard definition is still missing. We discover a universal feature in the distribution of the atomic-pair distances that distinguishes the melting transition of monometallic nanoparticles. We analyse the solid–liquid change of several late-transition metals nanoparticles, i.e. Ni, Cu, Pd, Ag, Au and Pt, through classical molecular dynamics. We consider various initial shapes from 146 to 976 atoms, corresponding to the 1.5–4.1 nm size range, placing the nanoparticles in either a vacuum or embedded in a homogeneous environment, simulated by an implicit force-field. Regardless of the material, its initial shape, size and environment, the second peak in the pair-distance distribution function, expected at the bulk lattice distance, disappears when the nanoparticle melts. As the pair-distance distribution is a measurable quantity, the proposed criterion holds for both numerical and experimental investigations. For a more straightforward calculus of the melting temperature, we demonstrate that the cross-entropy between a reference solid pair-distance distribution function and the one of nanoparticles at increasing temperatures present a quasi-first order transition at the phase-change temperature.

21 April 2022 in Physical Review B
Verified
Journal Article
Closed access
PDF file
CZClaudio Zeni
Claudio Zeni
AAAndrea Anelli
Andrea Anelli
AGAldo Glielmo
Aldo Glielmo
KRKevin Rossi
Kevin Rossi
We show that, contrary to popular assumptions, predictions from machine learning potentials built upon high-dimensional atom-density representations almost exclusively occur in regions of the representation space which lie outside the convex hull defined by the training set points. We then propose a perspective to rationalize the domain of robust extrapolation and accurate prediction of atomistic machine learning potentials in terms of the probability density induced by training points in the representation space.

11 June 2021 in The Journal of Chemical Physics
Verified
Journal Article
Closed access
PDF file
CZClaudio Zeni
Claudio Zeni
KRKevin Rossi
Kevin Rossi
AGAldo Glielmo
Aldo Glielmo
SGStefano de Gironcoli
Stefano de Gironcoli
We probe the accuracy of linear ridge regression employing a three-body local density representation derived from the atomic cluster expansion. We benchmark the accuracy of this framework in the prediction of formation energies and atomic forces in molecules and solids. We find that such a simple regression framework performs on par with state-of-the-art machine learning methods which are, in most cases, more complex and more computationally demanding. Subsequently, we look for ways to sparsify the descriptor and further improve the computational efficiency of the method. To this aim, we use both principal component analysis and least absolute shrinkage operator regression for energy fitting on six single-element datasets. Both methods highlight the possibility of constructing a descriptor that is four times smaller than the original with a similar or even improved accuracy. Furthermore, we find that the reduced descriptors share a sizable fraction of their features across the six independent datasets, hinting at the possibility of designing material-agnostic, optimally compressed, and accurate descriptors.

20 November 2018 in Physical Chemistry Chemical Physics
Verified
Journal Article
Closed access
PDF file
KRKevin Rossi
Kevin Rossi
GAGian Giacomo Asara
Gian Giacomo Asara
FBFrancesca Baletto
Francesca Baletto
Because size and shape can affect the chemo-physical properties of nanoparticles, we extend the use of geometrical descriptors to sequence a genome of monometallic nanoparticles. Selecting the generalised coordination number as a descriptor, the derived geometrical genome distinguishes, catalogues, and counts the variety of adsorption sites available on each isomer with a diameter up to 10 nm, therefore it depends on the nanoparticle size and shape. This procedure allows us to elucidate the effects of morphological diversity within a sample and those of thermally activated structural rearrangements among isomers on nanocatalyst activity. By screening the geometrical genome of archetypal shapes, we forecast Pt stellated twinned nanoparticles, elongated along their five-fold axis and with their shortest diameter of ∼2 nm, as optimal candidates for the electro-reduction of molecular oxygen at room temperature, in agreement with available experimental data.

9 June 2025 in Digital Discovery
Verified
Journal Article
Open access
PDF file
FGFederico Grasselli
Federico Grasselli
SCSanggyu Chong
Sanggyu Chong
VKVenkat Kapil
Venkat Kapil
SBSilvia Bonfanti
Silvia Bonfanti
KRKevin Rossi
Kevin Rossi
Error and uncertainty quantification is a central paradigm of science. In this perspective we discuss these themes in the context of machine learning predictions for atomistic modeling.