Tsveta Miteva

Bayesian joint neural architecture search and hyperparameter optimisation automatically builds optimal deep learning models, markedly improving pigment classification accuracy from hyperspectral imaging of medieval illumination mock-ups.

Understanding the interaction between metal ions and their aqueous environment is fundamental in many areas of chemistry, biology, and environmental science. In this study, we investigate the electronic structure of hydrated calcium ions, focusing on how water molecules influence the behavior of the metal ion. We employed advanced X-ray techniques, including X-ray absorption, photoelectron, and Auger spectroscopies, combined with high-level quantum chemical calculations. Our analysis reveals that, alongside normal Auger decay, distinct ultrafast charge transfer processes occur between the calcium ion and surrounding water molecules, underscoring the complex nature of metal-solvent interactions. Two primary mechanisms were identified. The first one involves electron transfer from water to the calcium ion. The second mechanism depends on the photon energy and is tentatively attributed to the decay of photoelectron satellites, the capture of free solvated electrons or electrons from a Cl$^-$ ion in the second solvation shell. Additionally, we observed significant shifts in electron energies due to post-collision interactions and interpreted the Ca 1s-1 photoelectron satellites mainly as originating from inelastic photoelectron scattering (IPES). These findings provide deeper insights into the electronic properties of hydrated metal ions, with potential implications for fields such as catalysis and biochemistry, where metal ions play a crucial role.

We experimentally and theoretically investigate interatomic Coulombic decay in small helium clusters. The electronic and nuclear dynamics for these systems are studied in detail using the energy distributions of the ionic fragments.

Hyperspectral imaging (HSI) has become widely used in cultural heritage (CH). This very efficient method for artwork analysis is connected with the generation of large amounts of spectral data. The effective processing of such heavy spectral datasets remains an active research area. Along with the firmly established statistical and multivariate analysis methods, neural networks (NNs) represent a promising alternative in the field of CH. Over the last five years, the application of NNs for pigment identification and classification based on HSI datasets has drastically expanded due to the flexibility of the types of data they can process, and their superior ability to extract structures contained in the raw spectral data. This review provides an exhaustive analysis of the literature related to NNs applied for HSI data in the CH field. We outline the existing data processing workflows and propose a comprehensive comparison of the applications and limitations of the various input dataset preparation methods and NN architectures. By leveraging NN strategies in CH, the paper contributes to a wider and more systematic application of this novel data analysis method.

We investigate interatomic Coulombic decay in NeKr dimers after neon inner-valence photoionization [Ne⁺(2s⁻¹)] using a synchrotron light source. We measure with high energy resolution the two singly charged ions of the Coulomb-exploding dimer dication and the photoelectron in coincidence. By carefully tracing the post-collision interaction between the photoelectron and the emitted ICD electron we are able to probe the temporal evolution of the state as it decays. Although the ionizing light pulses are 80 picoseconds long, we determine the lifetime of the intermediate dimer cation state and visualize the contraction of the nuclear structure on the femtosecond time scale.

Hyperspectral reflectance imaging in the short-wave infrared range (SWIR, “extended NIR”, ca. 1000 to 2500 nm) has proven to provide enhanced characterization of paint materials. However, the interpretation of the results remains challenging due to the intrinsic complexity of the SWIR spectra, presenting both broad and narrow absorption features with possible overlaps. To cope with the high dimensionality and spectral complexity of such datasets acquired in the SWIR domain, one data treatment approach is tested, inspired by innovative development in the cultural heritage field: the use of a pigment spectral database (extracted from model and historical samples) combined with a deep neural network (DNN). This approach allows for multi-label pigment classification within each pixel of the data cube. Conventional Spectral Angle Mapping and DNN results obtained on both pigment reference samples and a Buddhist painting (thangka) are discussed.