Search

Large Language Models (LLMs) have shown remarkable ability to generalize effectively across numerous industry domains while executing a range of tasks. Many of these competencies are obtained from the data utilized during the pre-training phase of the Language Models (LMs). However, these models exhibit limitations when tasked with performing in specialized or low-resource industry domains. More recent approaches use LLMs for generating domain-specific synthetic data but most often they lack in truthfulness and complexity. Alternatively, in cases where domain data is available like healthcare and finance most of the LMs are proprietary necessitating the need for a scalable method to curate real world industry specific pre-training data. In this work, we propose an automated and scalable framework - DoPAMine:Domain-specific Pre-training Adaptation from seed-guided data Mining, to mine domain specific training data from a large data corpus for domain adaptation of a LM. The framework leverages the parametric knowledge of a LLM to generate diverse and representative seed data tailored to a specific domain which is then used to mine real world data from a large data corpus like Common Crawl. We evaluated our framework's performance in the continual pre-training (CPT) setting by training two domain specific 7B parameter LMs in healthcare and finance with data mined via DoPAMine. Our experiments show that DoPAMine boosts the performance of pre-trained LLMs on average by 4.9% and 5.1% in zero-shot and 5-shot settings respectively on healthcare tasks from MMLU, MedQA, MedMCQA and PubMedQA datasets, and 2.9% and 6.7% for zero-shot and 5-shot settings respectively on finance tasks from FiQA-SA, FPB and Headlines datasets when compared to the baseline.

Background: FLASH radiotherapy is a treatment regime that delivers therapeutic dose to tumors at an ultra-high dose rate while maintaining adequate normal tissue sparing. However, a comprehensive understanding of the underlying mechanisms, potential late toxicities, and optimal fractionation schemes is important for successful clinical translation. This has necessitated extensive pre-clinical investigations, leading several research institutions to initiate dedicated FLASH research programs. Purpose: This work describes a workflow for establishing an easily accessible electron FLASH (eFLASH) platform. The platform incorporates simplified pulse control, optimized dose rate delivery, and validated Monte Carlo (MC) dose engine for accurate in vivo dosimetry dedicated to FLASH pre-clinical studies. Methods: Adjustment of the automatic frequency control (AFC) module allowed us to optimize the LINAC pulse form to achieve a uniform dose rate. A MC model for the 6 MeV FLASH beam was commissioned to ensure accurate dose calculation necessary for reproducible in vivo studies. Results: Optimizing the AFC module enabled the generation of a uniform pulse form, ensuring consistent dose per pulse and a uniform dose rate throughout FLASH irradiation. The MC model closely agreed with film measurements. MC dose calculations indicated that 6 MeV FLASH is adequate to achieve a uniform dose distribution for mouse whole brain irradiation but may not be optimal for the spinal cord study. Conclusions: We present a novel workflow for establishing a LINAC-based eFLASH research platform, incorporating techniques for optimized dose rate delivery, a simplified pulse control system, and validated MC engine. This work provides researchers with valuable new approaches to facilitate the development of robust and accessible LINAC-based system for FLASH studies., Comment: 29 pages, 6 figures

The Simons Observatory (SO), due to start full science operations in early 2025, aims to set tight constraints on inflationary physics by inferring the tensor-to-scalar ratio $r$ from measurements of CMB polarization $B$-modes. Its nominal design targets a precision $\sigma(r=0) \leq 0.003$ without delensing. Achieving this goal and further reducing uncertainties requires the mitigation of other sources of large-scale $B$-modes such as Galactic foregrounds and weak gravitational lensing. We present an analysis pipeline aiming to estimate $r$ by including delensing within a cross-spectral likelihood, and demonstrate it on SO-like simulations. Lensing $B$-modes are synthesised using internal CMB lensing reconstructions as well as Planck-like CIB maps and LSST-like galaxy density maps. This $B$-mode template is then introduced into SO's power-spectrum-based foreground-cleaning algorithm by extending the likelihood function to include all auto- and cross-spectra between the lensing template and the SAT $B$-modes. Within this framework, we demonstrate the equivalence of map-based and cross-spectral delensing and use it to motivate an optimized pixel-weighting scheme for power spectrum estimation. We start by validating our pipeline in the simplistic case of uniform foreground spectral energy distributions (SEDs). In the absence of primordial $B$-modes, $\sigma(r)$ decreases by 37% as a result of delensing. Tensor modes at the level of $r=0.01$ are successfully detected by our pipeline. Even with more realistic foreground models including spatial variations in the dust and synchrotron spectral properties, we obtain unbiased estimates of $r$ by employing the moment-expansion method. In this case, delensing-related improvements range between 27% and 31%. These results constitute the first realistic assessment of the delensing performance at SO's nominal sensitivity level. (Abridged), Comment: Updated version accepted by Physical Review D

Understanding the dynamics of dissipative quantum systems, particularly beyond the weak coupling approximation, is central to various quantum applications. While numerically exact methods provide accurate solutions, they often lack the analytical insight provided by theoretical approaches. In this study, we employ the recently-developed method dubbed the effective Hamiltonian theory to understand the dynamics of system-bath configurations without resorting to a perturbative description of the system-bath coupling energy. Through a combination of mapping steps and truncation, the effective Hamiltonian theory offers both analytical insights into signatures of strong couplings in open quantum systems and a straightforward path for numerical simulations. To validate the accuracy of the method, we apply it to two canonical models: a single spin immersed in a bosonic bath and two noninteracting spins in a common bath. In both cases, we study the transient regime and the steady state limit at nonzero temperature, and spanning system-bath interactions from the weak to the strong regime. By comparing the results of the effective Hamiltonian theory with numerically exact simulations, we show that although the former overlooks non-Markovian features in the transient equilibration dynamics, it correctly captures non-perturbative bath-generated couplings between otherwise non-interacting spins as observed in their synchronization dynamics and correlations. Altogether, the effective Hamiltonian theory offers a powerful approach to understanding strong coupling dynamics and thermodynamics, capturing the signatures of such interactions in both relaxation dynamics and in the steady state limit., Comment: Journal version

The Crab Nebula, also known as Tau A, is a polarized astronomical source at millimeter wavelengths. It has been used as a stable light source for polarization angle calibration in millimeter-wave astronomy. However, it is known that its intensity and polarization vary as a function of time at a variety of wavelengths. Thus, it is of interest to verify the stability of the millimeter-wave polarization. If detected, polarization variability may be used to better understand the dynamics of Tau~A, and for understanding the validity of Tau~A as a calibrator. One intriguing application of such observation is to use it for the search of axion-like particles (ALPs). Ultralight ALPs couple to photons through a Chern-Simons term, and induce a temporal oscillation in the polarization angle of linearly polarized sources. After assessing a number of systematic errors and testing for internal consistency, we evaluate the variability of the polarization angle of the Crab Nebula using 2015 and 2016 observations with the 150 GHz POLARBEAR instrument. We place a median 95% upper bound of polarization oscillation amplitude $A < 0.065^\circ$ over the oscillation frequencies from $0.75~\mathrm{year}^{-1}$ to $0.66~\mathrm{hour}^{-1}$. Assuming that no sources other than ALP are causing Tau A's polarization angle variation, that the ALP constitutes all the dark matter, and that the ALP field is a stochastic Gaussian field, this bound translates into a median 95% upper bound of ALP-photon coupling $g_{a\gamma\gamma} < 2.16\times10^{-12}\,\mathrm{GeV}^{-1}\times(m_a/10^{-21} \mathrm{eV})$ in the mass range from $9.9\times10^{-23} \mathrm{eV}$ to $7.7\times10^{-19} \mathrm{eV}$. This demonstrates that this type of analysis using bright polarized sources is as competitive as those using the polarization of cosmic microwave background in constraining ALPs., Comment: 24 pages, 19 figures, 5 tables

We propose a framework for robust evaluation of reasoning capabilities of language models, using functional variants of benchmarks. Models that solve a reasoning test should exhibit no difference in performance over the static version of a problem compared to a snapshot of the functional variant. We have rewritten the relevant fragment of the MATH benchmark into its functional variant MATH(), with functionalization of other benchmarks to follow. When evaluating current state-of-the-art models over snapshots of MATH(), we find a reasoning gap -- the percentage difference between the static and functional accuracies. We find reasoning gaps from 58.35% to 80.31% among the state-of-the-art closed and open weights models that perform well on static benchmarks, with the caveat that the gaps are likely to be smaller with more sophisticated prompting strategies. Here we show that models which anecdotally have good reasoning performance over real-world tasks, have quantifiable lower gaps, motivating the open problem of building "gap 0" models. Code for evaluation and new evaluation datasets, three MATH() snapshots, are publicly available at https://github.com/consequentai/fneval/., Comment: 37 pages, 10 figures

We investigate connectivity properties of the SARS-CoV2 spike protein-human ACE2-receptor complex employing a protein side chain-based network method that allows us to span a range from atomic to global protein scales. We analyze network topology in terms of clusters and cliques obtained from averaging over snapshots of MD simulations (from D.E. Shaw Research). We demonstrate that SARS-CoV2 forms a more dominant, robust connection with the ACE2-receptor as compared to the less virulent SARS-CoV1. Globally, this stronger connectivity is reflected by our percolation analysis where the interface cluster for the SARS-CoV2-ACE2 complex persists when restricted to stronger and stronger bonds, as compared to the SARSCoV1- ACE2 complex. At the atomic level, interface clique structure reflects a stronger connectivity in the former complex. We pinpoint key functional residues in SARSCoV2 that play important roles in establishing this higher connectivity. Thus, our studies provide an objective method to map spatial connectivity of atomic level non-covalent interactions to global connectivity between any two amino acids in the complex. We also analyze specific snapshots of the MD simulations to highlight prominent variations in network topology that explore diverse conformational landscapes. Finally, we demonstrate that a majority of mutations that occur in the SARSCoV2 spike protein in variants of concern/interest (including the currently circulating JN.1) have been observed at the interface with the ACE2 receptor. Our analyses highlight the importance of interface interactions and provide a rationale for designing receptor-like peptides and proteins to combat immunity-escaping variants., Comment: 18 pages, 14 figures, 1 Table (Ancillary file)

Dissipative processes can drive different magnetic orders in quantum spin chains. Using a non-perturbative analytic mapping framework, we systematically show how to structure different magnetic orders in spin systems by controlling the locality of the attached baths. Our mapping approach reveals analytically the impact of spin-bath couplings, leading to the suppression of spin splittings, bath-dressing and mixing of spin-spin interactions, and emergence of non-local ferromagnetic interactions between spins coupled to the same bath, which become long-ranged for a global bath. Our general mapping method can be readily applied to a variety of spin models: We demonstrate (i) a bath-induced transition from antiferromangnetic (AFM) to ferromagnetic ordering in a Heisenberg spin chain, (ii) AFM to extended Neel phase ordering within a transverse-field Ising chain with pairwise couplings to baths, and (iii) a quantum phase transition in the fully-connected Ising model. Our method is non-perturbative in the system-bath coupling. It holds for a variety of non-Markovian baths and it can be readily applied towards studying bath-engineered phases in frustrated or topological materials., Comment: (4.5 pages + 4 figures) main text (13 pages + 3 figures) supplemental material

Blockchain applications are witnessing rapid evolution, necessitating the integration of upgradeable smart contracts. Software patterns have been proposed to summarize upgradeable smart contract best practices. However, research is missing on the comparison of these upgradeable smart contract patterns, especially regarding gas costs related to deployment and execution. This study aims to provide an in-depth analysis of gas costs associated with two prevalent upgradeable smart contract patterns: the Proxy and diamond patterns. The Proxy pattern utilizes a Proxy pointing to a logic contract, while the diamond pattern enables a Proxy to point to multiple logic contracts. We conduct a comparative analysis of gas costs for both patterns in contrast to a traditional non-upgradeable smart contract. We derive from this analysis a theoretical contribution in the form of two consolidated blockchain patterns and a corresponding decision model. By so doing we hope to contribute to the broader understanding of upgradeable smart contract patterns.

We introduce prethermal temperature probes for sensitive, fast and robust temperature estimation. While equilibrium thermal probes with a manifold of quasidegenerate excited states have been previously recognized for their maximal sensitivity, they suffer from long thermalization timescales. When considering time as a critical resource in thermometry, it becomes evident that these equilibrium probes fall short of ideal performance. Here, we propose a different paradigm for thermometry, where setups originally suggested for optimal equilibrium thermometry should instead be employed as prethermal probes, by making use of their long-lived quasiequilibrium state. This transient state emerges from the buildup of quantum coherences among quasidegenerate levels. For a class of physically-motivated initial conditions, we find that energy measurements of the prethermal state exhibit a similar sensitivity as the equilibrium state. However, they offer the distinct benefit of orders of magnitude reduction in the time required for the estimation protocol. Upon introducing a figure-of-merit that accounts for the estimation protocol time, prethermal probes surpass the corresponding equilibrium probes in terms of effective thermal sensitivity, opening avenues for rapid thermometry by harnessing the long-lived prethermal states.

Accurate and precise measurements of spectroradiometric temperature are crucial for many high pressure experiments that use diamond anvil cells or shock waves. In experiments with sub-millisecond timescales, specialized detectors such as streak cameras or photomultiplier tubes are required to measure temperature. High accuracy and precision are difficult to attain, especially at temperatures below 3000 K. Here we present a new spectroradiometry system based on multianode photomultiplier tube technology and passive readout circuitry that yields a 0.24 $\mu$s rise-time for each channel. Temperature is measured using five color spectroradiometry. During high pressure pulsed Joule heating experiments in a diamond anvil cell, we document measurement precision to be $\pm 30$ K at temperatures as low as 2000 K during single-shot heating experiments with $0.6$ $\mu$s time-resolution. Ambient pressure melting tests using pulsed Joule heating indicate that the accuracy is $\pm 80$ K in the temperature range 1800-2700 K., Comment: 17 pages, 18 figures

Visualization onboarding supports users in reading, interpreting, and extracting information from visual data representations. General-purpose onboarding tools and libraries are applicable for explaining a wide range of graphical user interfaces but cannot handle specific visualization requirements. This paper describes a first step towards developing an onboarding library called VisAhoi, which is easy to integrate, extend, semi-automate, reuse, and customize. VisAhoi supports the creation of onboarding elements for different visualization types and datasets. We demonstrate how to extract and describe onboarding instructions using three well-known high-level descriptive visualization grammars - Vega-Lite, Plotly.js, and ECharts. We show the applicability of our library by performing two usage scenarios that describe the integration of VisAhoi into a VA tool for the analysis of high-throughput screening (HTS) data and, second, into a Flourish template to provide an authoring tool for data journalists for a treemap visualization. We provide a supplementary website that demonstrates the applicability of VisAhoi to various visualizations, including a bar chart, a horizon graph, a change matrix or heatmap, a scatterplot, and a treemap visualization.

We extend and benchmark the recently-developed Effective-Hamiltonian (EFFH) method [PRX Quantum $\bf{4}$, 020307 (2023)] as an approximation to the equilibrium state ("mean-force Gibbs state") of a quantum system at strong coupling to a thermal bath. The EFFH method is an approximate framework. Through a combination of the reaction-coordinate mapping, a polaron transformation and a controlled truncation, it imprints the system-bath coupling parameters into the system's Hamiltonian. First, we develop a $\textit{variational}$ EFFH technique. In this method, system's parameters are renormalized by both the system-bath coupling parameters (as in the original EFFH approach) and the bath's temperature. Second, adopting the generalized spin-boson model, we benchmark the equilibrium state from the EFFH treatment against numerically-exact simulations and demonstrate a good agreement for both polarization and coherences using the Brownian spectral function. Third, we contrast the (normal and variational) EFFH approach with the familiar (normal and variational) polaron treatment. We show that the two methods predict a similar structure for the equilibrium state, albeit the EFFH approach offers the advantage of simpler calculations and closed-form analytical results. Altogether, we argue that for temperatures comparable to the system's frequencies, the EFFH methodology provides a good approximation for the mean-force Gibbs state in the full range of system-bath coupling, from ultraweak to ultrastrong., Comment: 21 pages, 5 figures