arXiv subjects

Mikhail Petrov

Publications and source records attributed to Mikhail Petrov.

17 recordsLinked to original sources

arXiv2026

Non-cancellative varieties, maximal tori, and the Makar-Limanov invariant

In this paper, we construct a wide class of new counterexamples to the generalized Zariski cancellation problem. The cylinders over these counterexamples have infinitely many non-conjugate maximal tori in their regular automorphism groups. We provide an example of a variety with maximal tori of different dimensions in its automorphism group. Additionally, we prove that a cylinder over a non-rigid trinomial variety without a line factor is generically flexible, and hence, has a trivial Makar-Limanov invariant.

math.AG↗

arXiv2026

Wide-spectrum security of quantum key distribution

Implementations of quantum key distribution (QKD) need vulnerability assessment against loopholes in their optical scheme. Most of the optical attacks involve injecting or receiving extraneous light via the communication channel. An eavesdropper can choose her attack wavelengths arbitrarily within the quantum channel passband to maximise the attack performance, exploiting spectral transparency windows of system components. Here we propose a wide-spectrum security evaluation methodology to achieve full optical spectrum safety for QKD systems. This technique requires transmittance characterisation in a wide spectral band with a high sensitivity. We report a testbench that characterises insertion loss of fiber-optic components in a wide spectral range of 400 to 2300 nm and up to 70 dB dynamic range. To illustrate practical application of the proposed methodology, we give a full Trojan-horse attack analysis for some typical QKD system configurations and discuss briefly induced-photorefraction and detector-backflash attacks. Our methodology can be used for certification of QKD systems.

quant-ph↗

arXiv2026

Reference-Beam Attacks against Twin-Field Quantum Key Distribution using Optical Injection Locking

Twin-Field Quantum Key Distribution (TF-QKD) has become a leading protocol to bring quantum communications to the national scale. The protocol requires the establishment of a shared phase and frequency reference between distant parties, which is commonly achieved by using an external reference laser in an Optical Injection Locking (OIL) architecture. In this work, we analyze the side channels in OIL-based TF-QKD that may arise from adversarial manipulation of the various degrees of freedom of this untrusted reference beam. We experimentally demonstrate two realistic attack scenarios: fast intensity modulation of the reference laser, and additional signals embedded in the reference light exploiting wavelengths undetectable by conventional monitoring techniques. These attacks can allow a potential eavesdropper to deterministically increase the mean photon number of the sources, or circumvent the decoy-state technique, respectively. To counter these vulnerabilities, we propose practical and highly effective countermeasures that reinforce the security of TF-QKD systems without significant additional complexity or performance degradation.

quant-ph↗

arXiv2026

Learning Hyperparameters via a Data-Emphasized Variational Objective

When training large models on limited data, avoiding overfitting is paramount. Common grid search or smarter search methods rely on expensive separate runs for each candidate hyperparameter, while carving out a validation set that reduces available training data. In this paper, we study gradient-based learning of hyperparameters via the evidence lower bound (ELBO) objective from Bayesian variational methods. This avoids the need for any validation set. We focus on scenarios where the model is over-parameterized for flexibility and the approximate posterior is chosen to be Gaussian with isotropic covariance for tractability, even though it cannot match the true posterior. In such scenarios, we find the ELBO prioritizes posteriors that match the prior, leading to severe underfitting. Instead, we recommend a data-emphasized ELBO that upweights the likelihood but not the prior. In Bayesian transfer learning of image and text classifiers, our method reduces the 88+ hour grid search of past work to under 3 hours while delivering comparable accuracy. We further demonstrate how our approach enables efficient yet accurate approximations of Gaussian processes with learnable lengthscale kernels.

cs.LG↗

arXiv2025

Order-disorder duality of high entropy alloys extends non-linear optics

Order versus disorder in the structure of materials plays a key role in the theoretical prediction of their properties. However, this structural description appears to be ineffective for new families of materials such as high entropy alloys (HEAs), which combine crystallographic order with chemical disorder. Here, we demonstrate for five-element HEAs as pure solid solutions that the chemical disorder of the elements decorating their cubic structure underlies the generation of second optical harmonics, overcoming the theoretical limit imposed on centrosymmetric crystals. Moreover, we discover that this disorder, inherent to HEAs, sets a threshold for non-linear light emission from the 4th to the 26th order. As a consequence of the 0.5 eV broadening of the energy levels of the five elements of the HEA, the emission spectrum covers broad visible (400-650 nm) and infrared (800-1600 nm) ranges. In addition to the challenge of theoretically predicting non-linear effects in unconventional materials, the duality of structural order and chemical disorder in HEAs offers the opportunity to design sustainable alternatives to urgently needed optical materials.

cond-mat.mtrl-sci↗

arXiv2025

Mic-hackathon 2024: Hackathon on Machine Learning for Electron and Scanning Probe Microscopy

Microscopy is a primary source of information on materials structure and functionality at nanometer and atomic scales. The data generated is often well-structured, enriched with metadata and sample histories, though not always consistent in detail or format. The adoption of Data Management Plans (DMPs) by major funding agencies promotes preservation and access. However, deriving insights remains difficult due to the lack of standardized code ecosystems, benchmarks, and integration strategies. As a result, data usage is inefficient and analysis time is extensive. In addition to post-acquisition analysis, new APIs from major microscope manufacturers enable real-time, ML-based analytics for automated decision-making and ML-agent-controlled microscope operation. Yet, a gap remains between the ML and microscopy communities, limiting the impact of these methods on physics, materials discovery, and optimization. Hackathons help bridge this divide by fostering collaboration between ML researchers and microscopy experts. They encourage the development of novel solutions that apply ML to microscopy, while preparing a future workforce for instrumentation, materials science, and applied ML. This hackathon produced benchmark datasets and digital twins of microscopes to support community growth and standardized workflows. All related code is available at GitHub: https://github.com/KalininGroup/Mic-hackathon-2024-codes-publication/tree/1.0.0.1

cond-mat.mtrl-sci↗

arXiv2025

Real-space superconducting properties in the atomically-thin limit: Ab initio approach and its application to Josephson junctions

Real-space superconducting properties are increasingly important to characterize low-dimensional, layered, and nanostructured materials. Here, we present a method to extract the real-space superconducting order parameter from the superconducting gap spectrum obtained via anisotropic Migdal-Eliashberg calculations, using the Bloch wave functions of the Fermi states. We apply this approach to a selection of atomically thin material systems. Our analysis of gallenene, a monolayer of gallium atoms, shows that its planar and buckled phases exhibit distinct superconducting order parameter behaviors, shaped by their structural and electronic properties. Furthermore, we demonstrate that our real-space approach is exceptionally suited to identify and characterize Josephson junctions made from van der Waals materials. Our examination of a bilayer of NbSe$_2$ reveals that the van der Waals gap acts as an intrinsic weak link between the superconducting NbSe$_2$ layers. Therefore, a bilayer of NbSe$_2$ represents one of the thinnest and most tunable Josephson junction architectures, with potential applications in quantum devices. Our findings underscore the utility of transformation into real-space in understanding superconducting properties through ab initio calculations.

cond-mat.supr-con↗

arXiv2025

Learning the Regularization Strength for Deep Fine-Tuning via a Data-Emphasized Variational Objective

A number of popular transfer learning methods rely on grid search to select regularization hyperparameters that control over-fitting. This grid search requirement has several key disadvantages: the search is computationally expensive, requires carving out a validation set that reduces the size of available data for model training, and requires practitioners to specify candidate values. In this paper, we propose an alternative to grid search: directly learning regularization hyperparameters on the full training set via model selection techniques based on the evidence lower bound ("ELBo") objective from variational methods. For deep neural networks with millions of parameters, we specifically recommend a modified ELBo that upweights the influence of the data likelihood relative to the prior while remaining a valid bound on the evidence for Bayesian model selection. Our proposed technique overcomes all three disadvantages of grid search. We demonstrate effectiveness on image classification tasks on several datasets, yielding heldout accuracy comparable to existing approaches with far less compute time.

cs.LG↗

arXiv2024

Preparing a commercial quantum key distribution system for certification against implementation loopholes

A commercial quantum key distribution (QKD) system needs to be formally certified to enable its wide deployment. The certification should include the system's robustness against known implementation loopholes and attacks that exploit them. Here we ready a fiber-optic QKD system for this procedure. The system has a prepare-and-measure scheme with decoy-state BB84 protocol, polarisation encoding, qubit source rate of 312.5 MHz, and is manufactured by QRate. We detail its hardware and post-processing. We analyse the hardware for known implementation loopholes, search for possible new loopholes, and discuss countermeasures. We then amend the system design to address the highest-risk loopholes identified. We also work out technical requirements on the certification lab and outline its possible structure.

quant-ph↗

arXiv2024

Comment on "Coexistence of superconductivity and topological aspects in beryllenes", Materials Today Physics 38, 101257 (2023)

In a recent publication by Li $\textit{et al.}$, two phases of beryllene - $α$ and $β$ - were predicted to be single-gap superconductors with critical temperatures of 9.9 K and 12.6 K respectively. Moreover, the $α$-beryllene was shown to host type-I Dirac fermions with the existence of nontrivial edge states. We observe significantly weaker superconducting properties of both beryllene configurations. We argue that the superconducting gap evolution with temperature, as shown in Figure 5 (b and d) of Li $\textit{et al.}$, exhibits clearly unphysical trends with increasing temperature, leading to significantly overestimated values of the critical temperature and erroneous conclusions concerning the two-gap superconducting nature of $β$-beryllene. On a positive note, we report the value of the gap in the Dirac cone of the topological states of interest that exceeds the temperature range of superconductivity in $α$-beryllene, supporting the coexistence of topological features and superconductivity in this material.

cond-mat.supr-con↗

arXiv2024

Transfer Learning with Informative Priors: Simple Baselines Better than Previously Reported

We pursue transfer learning to improve classifier accuracy on a target task with few labeled examples available for training. Recent work suggests that using a source task to learn a prior distribution over neural net weights, not just an initialization, can boost target task performance. In this study, we carefully compare transfer learning with and without source task informed priors across 5 datasets. We find that standard transfer learning informed by an initialization only performs far better than reported in previous comparisons. The relative gains of methods using informative priors over standard transfer learning vary in magnitude across datasets. For the scenario of 5-300 examples per class, we find negative or negligible gains on 2 datasets, modest gains (between 1.5-3 points of accuracy) on 2 other datasets, and substantial gains (>8 points) on one dataset. Among methods using informative priors, we find that an isotropic covariance appears competitive with learned low-rank covariance matrix while being substantially simpler to understand and tune. Further analysis suggests that the mechanistic justification for informed priors -- hypothesized improved alignment between train and test loss landscapes -- is not consistently supported due to high variability in empirical landscapes. We release code to allow independent reproduction of all experiments.

cs.LG↗

arXiv2022

Stochastic circular persistent currents of exciton polaritons

We keep track of the orbital degree of freedom of an exciton polariton condensate, confined in an optical trap, and reveal the stochastic switching of persistent annular polariton currents in the pulse-periodic excitation regime.In an elliptic trap, the low-lying in energy polariton current states are inherent in a two-petalled density distribution and swirling phase. In the stochastic regime, the averaged over multiple excitation pulses density distribution gets homogenised in the azimuthal direction, while the weighted phase extracted from interference experiments experiences two compensating each other jumps, when varying around the center of the trap. Breaking the reciprocity of the system with a supplemental control optical pulse makes it possible to switch the system from the stochastic regime to the deterministic regime of an arbitrary polariton circulation.

cond-mat.mes-hall↗

arXiv2022

Independent quality assessment of a commercial quantum random number generator

We reverse-engineer, test and analyse hardware and firmware of the commercial quantum-optical random number generator Quantis from ID Quantique. We show that > 99% of its output data originates in physically random processes: random timing of photon absorption in a semiconductor material, and random growth of avalanche owing to impact ionisation. Under a strong assumption that these processes correspond to a measurement of an initially pure state of the components, our analysis implies the unpredictability of the generated randomness. We have also found minor non-random contributions from imperfections in detector electronics and an internal processing algorithm, specific to this particular device. Our work shows that the design quality of a commercial quantum-optical randomness source can be verified without cooperation of the manufacturer and without access to the engineering documentation.

quant-ph↗

arXiv2021

High-temperature multigap superconductivity in two-dimensional metal-borides

Using first-principles calculations in combination with the Eliashberg formalism, we systematically investigated phonon-mediated superconductivity in two-dimensional (2D) metal-boride crystals, consisting of a boron honeycomb network doped by diverse metal elements. Such 2D metal-boride compounds, named MBenes, are chemically exfoliable from single-crystalline layered ternary borides (MAB phases). First, we identified the MBene layers with potential for superconductivity via isotropic Eliashberg calculations, considering a wide range of metal elements, with focus on alkaline earth and transition metals. Subsequently, we performed a detailed analysis of the prominent superconducting MBenes by solving the anisotropic Eliashberg equations. The obtained high critical temperatures (up to 72 K), as well as the rich multigap superconducting behavior, recommend these crystals for further use in multifunctional 2D heterostructures and superconducting device applications.

cond-mat.supr-con↗

arXiv2021

Superconductivity and topological behavior in gallenene

Among the large variety of two-dimensional (2D) materials discovered to date, elemental monolayers that host superconductivity are very rare. Using ab initio calculations we show that recently synthesized gallium monolayers, coined gallenene, are intrinsically superconducting through electron-phonon coupling. We reveal that Ga-100 gallenene, a planar monolayer isostructural with graphene, is the structurally simplest 2D superconductor to date, furthermore hosting topological edge states due to its honeycomb structure. Our anisotropic Eliashberg calculations show distinctly three-gap superconductivity in Ga-100, in contrast to the alternative buckled Ga-010 gallenene which presents a single anisotropic superconducting gap. Strikingly, the critical temperature ($T_c$) of gallenene is in the range of $7-10$ K, exceeding the $T_c$ of bulk gallium from which it is exfoliated. Finally we explore chemical functionalization of gallenene with hydrogen, and report induced multigap superconductivity with an enhanced $T_c$ in the resulting gallenane compound.

cond-mat.supr-con↗

arXiv2019

Hydrogen-induced high-temperature superconductivity in two-dimensional materials: Example of hydrogenated monolayer MgB$_2$

Hydrogen-based compounds under ultra-high pressure, such as the polyhydrides H$_3$S and LaH$_{10}$, superconduct through the conventional electron-phonon coupling mechanism to attain the record critical temperatures known to date. We demonstrate here that the intrinsic advantages of hydrogen for phonon-mediated superconductivity can be exploited in a completely different system, namely two-dimensional (2D) materials. We find that hydrogen adatoms can strongly enhance superconductivity in 2D materials due to flatband states originating from atomic-like hydrogen orbitals, with a resulting high density of states, and due to the emergence of high-frequency hydrogen-related phonon modes that boost the electron-phonon coupling. As a concrete example, we investigate the effect of hydrogen adatoms on the superconducting properties of monolayer MgB$_2$, by solving the fully anisotropic Eliashberg equations, in conjunction with a first-principles description of the electronic and vibrational states, and the coupling between them. We show that hydrogenation leads to a high critical temperature of 67 K, which can be boosted to over 100 K by biaxial tensile strain.

cond-mat.supr-con↗

arXiv2015

Towards all-dielectric metamaterials and nanophotonics

We review a new, rapidly developing field of all-dielectric nanophotonics which allows to control both magnetic and electric response of structured matter by engineering the Mie resonances in high-index dielectric nanoparticles. We discuss optical properties of such dielectric nanoparticles, methods of their fabrication, and also recent advances in all-dielectric metadevices including couple-resonator dielectric waveguides, nanoantennas, and metasurfaces.

physics.optics↗