Research Software Engineers (RSEs) have become indispensable to computational research and scholarship. The fast rise of RSEs in higher education and the trend of universities to be slow creating or adopting models for new technology roles means a lack of structured career pathways that recognize technical mastery, scholarly impact, and leadership growth. In response to an immense demand for RSEs at Princeton University, and dedicated funding to grow the RSE group at least two-fold, Princeton was forced to strategize how to cohesively define job descriptions to match the rapid hiring of RSE positions but with enough flexibility to recognize the unique nature of each individual position. This case study describes our design and implementation of a comprehensive RSE career ladder spanning Associate through Principal levels, with parallel team-lead and managerial tracks. We outline the guiding principles, competency framework, Human Resources (HR) alignment, and implementation process, including engagement with external consultants and mapping to a standard job leveling framework utilizing market benchmarks. We share early lessons learned and outcomes including improved hiring efficie
As Princeton WordNet continues to gain significance as a semantic lexicon in Natural Language Processing, the need for its localization and for ensuring the quality of this process has become increasingly critical. Existing efforts remain limited in both scale and rigor, and there is a notable absence of studies addressing the accuracy of localization or its alignment with the cultural context of Arabic. This paper proposes a structured framework for the localization of Princeton WordNet, detailing the stages and procedures required to achieve high-quality results without compromising cultural authenticity. We further present our experience in applying this framework, reporting outcomes from the localization of 10,000 synsets.
These are lectures notes prepared for a series of seminars I am invited to give at Princeton Philosophy Department in November 2024. They cover the conceptual structure of quantum gravity, the relational interpretation of quantum mechanics, the structure of time, its orientation and the openness of the future, the physical underpinning of information and meaning, and some general considerations on the fact that concepts evolve, on perspectivalism and anti-foundationalism.
The Princeton Research Software Engineering Group has grown rapidly since its inception in late 2016. The group, housed in the central Research Computing Department, comprised of professional Research Software Engineers (RSEs), works directly with researchers to create high quality research software to enable new scientific advances. As the group has matured so has the need for formalizing operational details and procedures. The RSE group uses an RSE partnership model, where Research Software Engineers work long-term with a designated academic department, institute, center, consortium, or individual principal investigator (PI). This article describes the operation of the central Princeton RSE group including funding, partner & project selection, and best practices for defining expectations for a successful partnership with researchers.
Wordnets are rich lexico-semantic resources. Linked wordnets are extensions of wordnets, which link similar concepts in wordnets of different languages. Such resources are extremely useful in many Natural Language Processing (NLP) applications, primarily those based on knowledge-based approaches. In such approaches, these resources are considered as gold standard/oracle. Thus, it is crucial that these resources hold correct information. Thereby, they are created by human experts. However, human experts in multiple languages are hard to come by. Thus, the community would benefit from sharing of such manually created resources. In this paper, we release mappings of 18 Indian language wordnets linked with Princeton WordNet. We believe that availability of such resources will have a direct impact on the progress in NLP for these languages.
The PTOLEMY experiment (Princeton Tritium Observatory for Light, Early-Universe, Massive-Neutrino Yield) aims to achieve the sensitivity required to detect the relic neutrino background through a combination of a large area surface-deposition tritium target, MAC-E filter methods, cryogenic calorimetry, and RF tracking and time-of-flight systems. A small-scale prototype is in operation at the Princeton Plasma Physics Laboratory with the goal of validating the technologies that would enable the design of a 100 gram PTOLEMY. With precision calorimetry in the prototype setup, the limitations from quantum mechanical and Doppler broadening of the tritium target for different substrates will be measured, including graphene substrates. Beyond relic neutrino physics, sterile neutrinos contributing to the dark matter in the universe are allowed by current constraints on partial contributions to the number of active neutrino species in thermal equilibrium in the early universe. The current PTOLEMY prototype is expected to have unique sensitivity in the search for sterile neutrinos with electron-flavor content for masses of 0.1--1keV, where less stringent, 10eV, energy resolution is required.
The Princeton Variability Survey (PVS) is a robotic survey which makes use of readily available, ``off-the-shelf'' type hardware products, in conjunction with a powerful set of commercial software products, in order to monitor and discover variable objects in the night sky. The main goal of the PVS has been to devise an automated telescope and data reduction system, requiring only moderate technical and financial resources to assemble, which may be easily replicated by the dedicated amateur, a student group, or a professional and used to study and discover a variety of variable objects, such as stars. This paper describes the hardware and software components of the PVS device as well as observational results from the initial season of the PVS, including the discovery of a new bright variable star.
We investigate numerically the Princeton magneto-rotational instability (MRI) experiment and the effect of conducting axial boundaries or endcaps. MRI is identified and found to reach a much higher saturation than for insulating endcaps. This is probably due to stronger driving of the base flow by the magnetically rather than viscously coupled boundaries. Although the computations are necessarily limited to lower Reynolds numbers ($\Re$) than their experimental counterparts, it appears that the saturation level becomes independent of $\Re$ when $\Re$ is sufficiently large, whereas it has been found previously to decrease roughly as $\Re^{-1/4}$ with insulating endcaps. The much higher saturation levels will allow for the first positive detection of MRI beyond its theoretical and numerical predictions.
This report provides a summary and bibliography for the research activities at the Princeton University Observatory and Department of Astrophysical Sciences during the period July 1, 1997 to June 30, 1998.
Core-polarization (CP) and Brown-Rho (BR) scaling were among Gerry's most favorite topics. In this contribution, we will discuss some of the early history as well as more recent work associated with these two fascinating phenomena. I (TTSK) will begin with some recollections of Princeton, where I met Gerry for the first time in 1964 and worked as his postdoc. Core polarization was in fact the first topic he assigned to me. JWH started working with Gerry at Stony Brook from 2003 and was Gerry's last student in nuclear physics. We three had teamed up well, working closely on both CP and BR scaling, particularly on the latter's connection to the anomalously-long beta-decay lifetime of carbon-14. We shall here briefly review these topics, including a recently developed new Brown-Rho scaling based on a Skyrmion half-Skyrmion two-phase model.
Submitted by the authors for the June 27-29 Princeton Conference. Questions should be directed to: qian@cfd.princeton.edu
In preparation for an experimental study of magnetorotational instability (MRI) in liquid metal, we present non-ideal axisymmetric magnetohydrodynamic simulations of the nonlinear evolution of MRI in the experimental geometry. The simulations adopt fully insulating boundary conditions. No-slip conditions are imposed at the cylinders. A clear linear phase is observed with reduced linear growth rate. MRI results in an inflowing "jet" near the midplane and enhances the angular momentum transport at saturation.
暂无摘要，请点击原文查看详情
In core-collapse supernovae (CCSNe), deleptonization normally favors $ν_e$ over $\barν_e$ emission. However, lepton-number emission self-sustained asymmetry (LESA) can make the energy-integrated emitted lepton-number flux negative along some directions. We derive a simple diagnostic for this transition and test it in 33 independent 3D CCSN simulations: 25 Princeton/Fornax models ($8.1$--$100\,M_\odot$) and 8 Garching models, including non-, slow-, and fast-rotating $15\,M_\odot$ cases. Of 23 non-black-hole-forming Princeton models, 22 cross the threshold, with median onset $t_c=225\,\mathrm{ms}$, IQR $162$--$264\,\mathrm{ms}$, and cross-model scatter $\mathrm{CV}=18.6\%$. Full-sky flux-sign searches show that the threshold identifies the anti-LESA-pole transition, distinguishing the global LESA-driven crossing from early localized turbulent crossings. The fast-rotating Garching $15\,M_\odot$ model, where rapid rotation suppresses the LESA dipole, is correctly classified as a non-crosser without using any rotation parameter. Both black-hole-forming Princeton models cross near $250\,\mathrm{ms}$ post-bounce and remain above threshold for $1807$ and $2463\,\mathrm{ms}$ before collapse
Existing benchmarks for evaluating long-context language models (LCLMs) primarily focus on long-context recall, requiring models to produce short responses based on a few critical snippets while processing thousands of irrelevant tokens. We introduce LongProc (Long Procedural Generation), a new benchmark that requires both the integration of highly dispersed information and long-form generation. LongProc consists of six diverse procedural generation tasks, such as extracting structured information from HTML pages into a TSV format and executing complex search procedures to create travel plans. These tasks challenge LCLMs by testing their ability to follow detailed procedural instructions, synthesize and reason over dispersed information, and generate structured, long-form outputs (up to 8K tokens). Furthermore, as these tasks adhere to deterministic procedures and yield structured outputs, they enable reliable rule-based evaluation. We evaluated 23 LCLMs, including instruction-tuned models and recent reasoning models, on LongProc at three difficulty levels, with the maximum number of output tokens set at 500, 2K, and 8K. Notably, while all tested models claim a context window size
Language agents increasingly require persistent worlds in which they can act, remember, and learn. Existing approaches sit at two extremes: conventional web frameworks provide reliable but fixed contexts backed by databases, while fully generative world models aim for unlimited environments at the expense of controllability and practical engineering. In this work, we introduce the Web World Model (WWM), a middle ground where world state and ``physics'' are implemented in ordinary web code to ensure logical consistency, while large language models generate context, narratives, and high-level decisions on top of this structured latent state. We build a suite of WWMs on a realistic web stack, including an infinite travel atlas grounded in real geography, fictional galaxy explorers, web-scale encyclopedic and narrative worlds, and simulation- and game-like environments. Across these systems, we identify practical design principles for WWMs: separating code-defined rules from model-driven imagination, representing latent state as typed web interfaces, and utilizing deterministic generation to achieve unlimited but structured exploration. Our results suggest that web stacks themselves ca
In 2003, Simon Baron-Cohen, a world expert on autism, diagnosed Einstein posthumously with Asperger's syndrome. I think we cannot diagnose a dead person. Historians of science have fiercely objected to this trend of diagnosing deceased scientists by reconstructing from scant evidence, calling these diagnoses myths. Despite the historians' efforts at demolishing myths, Einstein has been constantly diagnosed with Asperger's syndrome. I will stick my neck out and suggest in this paper that although historians' critique of Baron-Cohen and others includes debunking myths, it piggybacks on another myth that uses the following metaphors: a dull and socially adept Einstein who worked at Zurich, Prague, Berlin, and Princeton, an industrious scientist who earned his living through his work as a professor at the university; he had a special gift of friendship and collegiality, and he was deeply embedded in the academic community. These explanations do not make sense from the perspective of Einstein sitting in his office at Princeton, let alone Einstein sitting in the patent office. This perhaps explains the tendency of people to find counterclaims and myths more persuasive than historians' ex
These lecture notes were written for the course 18.657, High Dimensional Statistics at MIT. They build on a set of notes that was prepared at Princeton University in 2013-14 that was modified (and hopefully improved) over the years.
We have developed a web-based pedagogical proof assistant, the Proof Tree Builder, that lets you apply rules upwards from the initial goal in sequent calculus and Hoare logic for a simple imperative language. We equipped our tool with a basic proof automation feature and Z3 integration. Students in the automated reasoning course at Princeton University used our tool and found it intuitive. The Proof Tree Builder is available online at https://proof-tree-builder.github.io.