תקצירי ההרצאות

Prof. Volker Gebhardt

Title: Cataloguing Unlabeled Lattices – algebra, combinatorics and computing on a date

Abstract: Generating catalogues of examples that are in some sense complete has proved to be an important step towards understanding mathematical concepts, and it often is one of the key steps towards a successful classification and thus a complete theory. One type of structure I am particularly interested in are so-called combinatorial lattices; these are partially ordered sets where each pair of elements admits a unique least common upper bound and a unique greatest common lower bound. Enumerating all combinatorial lattices of a given size is a hard problem: even the number of lattices on n elements was previously only known for n≤19 (OEIS: A006966).

In this talk I will focus on explaining how to combine group theory, combinatorics and some clever ideas from computing to enumerate (i.e. list systematically, catalogue) unlabelled lattices more effectively. The basic idea is to structure the search space in a way that maximises symmetries and thus allows to reduce the computational complexity of the problem.

The main ideas are relatively elementary, and I will explain all required concepts, so the talk should be accessible to a non-specialist audience.

Dr. Avner Segal, Department of Mathematics

Title: Combinatorial Patterns in Intertwining Operator Singularities

Abstract: In 1997, S. Zampera demonstrated that the principal series representations of groups of type \(G_{2}\) admit a striking decomposition into a direct sum of two sub-representations. This decomposition was realized via the eigenspaces of an operator defined by the residue of an intertwining operator along a singular hyperplane.

In this talk, I will present a generalization of this phenomenon to the broader setting of quasi-split Lie groups, based on joint work with T. Nam and L. Silberman. We establish a set of geometric conditions within the group's root data that guarantee the existence of such decompositions. The primary focus of the lecture will be the combinatorial problem of cataloguing these configurations: using labeled Dynkin diagrams, we produce an extensive - and potentially exhaustive - catalogue of these occurrences.

These results have since been used by the author for determining the degenerate residual spectrum of groups of type \(D_{4}\) and \(E_{n}\). While the focus remains on the combinatorial-geometric classification, I will briefly discuss the representation-theoretic context of this work and, if time permits, discuss further generalizations.

Dr. Jeremy Miller, Department of Computer Science

Title: Feynman graphs and solving the Bethe-Salpeter equation to calculate the Higgs-Boson mass

Abstract: Feynman graphs offer an intuitive understanding of how fundamental particles interact in quantum theory.
There exist different classes of particles in the standard model (SM) that interact due to four fundamental forces:
the electromagnetic force, the weak force, the strong force and the gravitational force.
The origin of the masses of the SM particles is not fully understood.
In 1963 Higgs proved that the existence of an extra new particle, which later became known as the Higgs Boson, guarantees that all particles in the SM possess mass.
Such a particle was later detected at the Large Hadron Collider in 2008.
However, the structure of the Higgs Boson and how it interacts with SM particles, and with itself,
remain unknown. We propose a new model in which the Higgs Boson is composed of new class of fundamental scalar particles.
Even more, the fundamental scalars carry the interaction between fermions and gravity and cancel the Weyl anomaly.
 

Dr. Dean N. Riley, AZ, USA

Title: Optical and infrared spectroscopy mineral identification and volume estimates applied to forward modeling of cross-property rock physics models for some New Mexico granites.

Abstract: Forward modeling and inversion of geophysical, geochemical, geomechanically and geological data in rock mechanics, rock physics, and mineral exploration and ore deposit characterization are unconstrained problems which are challenging for machine learning (ML) and artificial (AI) algorithms. Precise mineralogy identification and volume estimates are critical for the development of integrated geophysical and geological multi-modal ML/AI using cross-property rock physics models to effectively constrain geophysical and geological inversions for multi-sensor fusion for mineral exploration, mineral processing, metallurgical (re)processing and mine planning information. Reliable mineralogy identification with accurate volume estimates have traditionally been the limiting factor in this process. Development of these models has been challenging mostly from the lack of synchronized data collection of physical property measurements and optical and infrared spectroscopy. Rapid, accurate, and nondestructive characterization of core, thin section billets, and hand samples using optical and infrared spectroscopy can identify mineralogy and provide reasonable volume estimates of mineralogy thus improving rock physics and mineralogy cross-property relationships to constrain the geophysical and geological inversions in a unified framework. A proposed unified framework will be discussed using results from some New Mexico granites focused on the development of cross-property rock physics models based on mineralogy from optical and infrared spectroscopy to predict elastic properties (P- and S-wave acoustic velocities, or Vp and Vs), unconfined compressive strength (UCS), and other physical and mechanical properties for improvement of geophysical and geological inversions.
 

Dr. Edward C. Wellman, AZ, USA

Title: SWIR Spectroscopy for Geotechnical Characterization of an Altered Granite

Abstract: Reflectance, emissivity, absorption, and transmission are fundamental physical properties of minerals. This study investigates whether Shortwave Infrared (SWIR) hyperspectral data can be used to quantify geotechnical properties for physical rock description, mineralogical interpretation, and estimation of weathering, alteration, and rock strength. Hyperspectral imaging provides a non-destructive means of linking mineralogical features to mechanical behavior, supporting objective rock mass characterization. Core samples from an altered granite were scanned in the laboratory before destructive testing. Visible to Near Infrared (VNIR) and SWIR hyperspectral images were acquired at a spatial resolution of approximately 0.2 mm and a spectral resolution of 5.6 nm. The dataset of 40 samples, although limited, provided the basis for testing data preprocessing, feature extraction, and mineral classification methods applicable to geotechnical contexts. The primary hypothesis is that SWIR hyperspectral imaging, combined with spatial segmentation and spectral feature extraction, can yield reliable rock descriptions that capture geotechnical features, including rock strength, alteration intensity, and discontinuities. Spectral Angle Mapper (SAM) classification and continuum-removal processing were applied to identify alteration minerals and map their spatial distribution. The spectral results were compared with conventional laboratory measurements, including uniaxial compressive strength (UCS), sonic velocity, density, and Leeb hardness, to assess correlations between spectral features and mechanical properties.
 

Prof. Vladimir Frid, Department of Civil Engineering

Title: Microwave Magic on Chert Gravel: From 20 Million Tons of Quarry Waste to Construction Material

Abstract: Deep in the Negev Desert, the sand quarries that supply Israel's entire construction industry hide a stubborn secret: roughly 40% of everything excavated is not sand at all. It is chert gravel - dark, flint-hard nodules of ancient microcrystalline quartz, impossibly abrasive, and utterly useless by conventional standards. Every crusher that meets it is destroyed. Every fragment it produces comes out flat and flaky, failing industry specifications. The result is a slow-motion catastrophe: some 20 million tons of chert piling up across former quarry fields, with nowhere to go and no known way to tame it. Until, perhaps, the kitchen microwave changed everything. This breakthrough offers a promising solution that could transform a major waste problem into valuable resources, inspiring confidence in the future of construction materials.

This talk presents a four-year research program that asked a deceptively simple question: What happens when you put chert gravel in a microwave oven? The answer, it turns out, involves resonant physics hidden within the rock's grain structure. Quartz crystals roughly 5 micrometers across vibrate at a natural frequency of approximately 2.3 GHz - almost exactly the frequency used in microwave ovens - and this near-perfect match triggers a piezoelectric coupling that conventional thermal models entirely miss. The result is a strength reduction of four to five times in under two and a half minutes, achieved without a single hammer blow. This innovative approach opens new avenues for resource recovery, sparking curiosity about future applications.

Three published studies (Minerals 2023, Resources 2024, Clean Technologies 2025) and an ongoing semi-field campaign at 2–4 kW systematically map how exposure duration, sample mass, surface moisture, and cooling method interact to produce well-graded construction aggregates that meet standard USCS classification. A practical recipe emerges: the right combination of microwave dose and cooling method allows the engineer to dial in the target aggregate class (well-graded gravel for road base and concrete, well-graded sand for fine-aggregate applications) with the desired outcome achieved in over half of all trials. At the semi-field scale, surface temperatures reach 280–360 °C in minutes, reproducing the laboratory window and pointing toward industrial adoption at throughputs already demonstrated by pilot systems elsewhere. This detailed understanding reassures industry professionals of the process's viability at scale.

The key takeaway is that a material traditionally considered waste, such as chert gravel, can be repurposed into valuable construction resources by tuning microwave frequency, highlighting a sustainable approach to waste management.

Prof. Dmitry Baimel, Department of Electrical and Electronics

Title: Harnessing Physics-Informed AI for Predictive Maintenance in Next-Gen Electric Vehicles

Abstract: As Electric Vehicles (EVs) dominate the automotive landscape, the shift from preventive to predictive maintenance (PdM) has become an operational necessity. Unlike internal combustion engines, EV degradation—ranging from battery chemical aging to power electronics fatigue—follows complex, non-linear trajectories that traditional models fail to capture.

This session explores a multi-layered framework for EV fault detection and health monitoring. We begin by examining the Cyber-Physical architecture, where high-frequency sensor data from the CAN-Bus and Automotive Ethernet are processed at the Edge. We will delve into the integration of Digital Twins with Deep Learning architectures, specifically focusing on Long Short-Term Memory (LSTM) networks for State-of-Health (SOH) estimation and Convolutional Neural Networks (CNN) for motor current signature analysis (MCSA).
 

Prof. Shlomo Greenberg, Head, Department of Computer Science

Drones Detection using Deep Neural Network with RF and Acoustic Features

Abstract: The use of drones has recently gained popularity in a diverse range of applications, such as aerial photography, agriculture, search and rescue operations, the entertainment industry, and more. However, misuse of drone technology can potentially lead to military threats, terrorist acts, as well as privacy and safety breaches. This emphasizes the need for effective and fast remote detection of potentially threatening drones. In this study, we propose a novel approach for automatic drone detection utilizing the usage of both radio frequency communication signals and acoustic signals derived from UAV rotor sounds. In particular, we propose the use of classical and deep machine-learning techniques and the fusion of RF and acoustic features for efficient and accurate drone classification. Distinct types of ML-based classifiers have been examined, including CNN- and RNN-based networks and the classical SVM method. The proposed approach has been evaluated with both frequency and audio features using common drone datasets, demonstrating better accuracy than existing state-of-the-art methods, especially in low SNR scenarios. The results presented in this paper show a classification accuracy of approximately 91% at an SNR ratio of −10 dB using the LSTM network and fused features.

Dr. Aviad Elyashar, Department of Computer Science

Title: Assessment and manipulation of latent constructs in pre-trained language models using psychometric scales

Abstract: Human-like personality traits have recently been discovered in large language models, raising the hypothesis that their (known and as yet undiscovered) biases conform to human

latent psychological constructs. While large conversational models may be tricked into answering psychometric questionnaires, the latent psychological constructs of thousands of simpler transformers, trained for other tasks, cannot be assessed because appropriate psychometric methods are currently lacking. Here, we show how standard psychological questionnaires can be reformulated into natural language inference prompts, and we provide a code library to support the psychometric assessment of arbitrary models. We demonstrate, using a sample of 88 publicly available models, the existence of human-like mental health-related constructs—including anxiety, depression, and Sense of Coherence—which conform with standard theories in human psychology and show similar correlations and mitigation strategies. The ability to interpret and improve the performance of language models using psychological tools can accelerate the development of more explainable, controllable, and trustworthy models.