Search

A realisation of a graph in the plane as a bar-joint framework is rigid if there are finitely many other realisations, up to isometries, with the same edge lengths. Each of these finitely-many realisations can be seen as a solution to a system of quadratic equations prescribing the distances between pairs of points. For generic realisations, the size of the solution set depends only on the underlying graph so long as we allow for complex solutions. We provide a characterisation of the realisation number - that is the cardinality of this complex solution set - of a minimally rigid graph. Our characterisation uses tropical geometry to express the realisation number as an intersection of Bergman fans of the graphic matroid. As a consequence, we derive a combinatorial upper bound on the realisation number involving the Tutte polynomial. Moreover, we provide computational evidence that our upper bound is usually an improvement on the mixed volume bound., Comment: 33 pages, 10 figures

The Milky Way's disk-halo interface mediates energy and mass exchange between the interstellar thin disk and the halo. In the first detailed study of the Perseus arm's disk-halo interface, we combine HST/STIS and COS absorption spectra toward 6 stars and 23 AGNs projected behind a narrow section (95 degree < l <145 degree, -46 degree < b <0 degree), providing a unique dataset that bridges the disk and its extended vertical structure in these directions. We detect S II, Si IV, and C IV absorption, along with HI 21 cm emission, within -70 pc to -3.3 kpc from the mid-plane. The arm's southern vertical structure exhibits complexity beyond simple exponential scaling: HI and S II column densities sharply decline with height up to 1.5 kpc before flattening, while high ion (Si IV and C IV) column densities remain relatively constant. In this region, where warm neutral medium (WNM) dominates, S II and the high ions show similar kinematics, and we find a remarkably uniform CIV/SiIV ratio ( = 2.5 pm 0.5) within -0.9 to -3.25 kpc. Both the kinematic correspondence and high-ion ratio are consistent with the high ions probing turbulent mixing layers at the interfaces between warm/cool and hot gas phases. AGN sightlines reveal minimal circumgalactic medium (CGM) contribution in the low-velocity gas at |v_{LSR}|< 100 km/s, suggesting the observed properties may be attributed to previous fountain activity.

We systematically probe different parameterizations of the attractive nuclear force based on real gas models to construct the nuclear matter equation of state. In each of the cases, the repulsion between nucleons is treated in the framework of excluded volume and interaction parameters are fitted to the empirical properties of the nuclear ground state. We calculate the critical temperature $T_c$ and critical particle number density $n_c$, and find that they are strongly correlated. Both are also correlated with the incompressibility $K_0$ in the nuclear ground state. We also include a quarkyonic matter phase in the quasiparticle description and investigate the relationship between $K_0$, transition density to the quarkyonic phase $n_{tr}$, and corresponding peak in the speed of sound $v_{s, {\rm max}}^2$. At each density, the quark fraction is found by minimizing the energy density. We find that both $n_{tr}$ and $v_{s, {\rm max}}^2$ are negatively correlated with $K_0$, $n_c$, and $T_c$., Comment: 8 pages, 4 figures, 3 tables

We introduce QualityFlow, a dynamic agentic workflow for program synthesis. Given the English description of a programming problem and a set of unit tests, the model's goal is to synthesize the correct program that solves the problem and passes the tests. QualityFlow consists of multiple large language model (LLM) agents that resemble a software development team, including code generation, testing, and self-debugging. Existing program synthesis methods face three major limitations: assumption of visible unit test conformity, bottleneck of synthesized test quality, and deviation of self-debugging trajectory. To address them, we propose the LLM Quality Checker, which explicitly "imagines" whether the synthesized programs' execution would conform to the unit tests. The Quality Checks dynamically control the workflow, including actions to submit the final answer, clarify the problem statement, and revert previous workflow steps. As a result, our Quality Checker can precisely accept any correct program, mitigate faulty synthesized tests, and prevent potential workflow deviation. The success of the Quality Checker further enables Diversified Prompting, which encourages variations in LLM responses to maximize the possibility that a correct program appears and passes the quality check. In experiments, QualityFlow establishes the state-of-the-art results on four program synthesis benchmarks: MBPP, HumanEval, and the stricter evaluations of both MBPP and HumanEval from EvalPlus. Our systematic analysis shows that the dynamic workflow controlled by LLM quality checks can outperform static workflows and single-attempt zero-shot synthesis. The Quality Checker is the center of our investigation, and we dissect its individual performance and integrated impact on the workflow accuracy, as well as other ablations experiments to justify our workflow design.

This paper introduces a context-aware model for robust counterspeech generation, which achieved significant success in the MCG-COLING-2025 shared task. Our approach particularly excelled in low-resource language settings. By leveraging a simulated annealing algorithm fine-tuned on multilingual datasets, the model generates factually accurate responses to hate speech. We demonstrate state-of-the-art performance across four languages (Basque, English, Italian, and Spanish), with our system ranking first for Basque, second for Italian, and third for both English and Spanish. Notably, our model swept all three top positions for Basque, highlighting its effectiveness in low-resource scenarios. Evaluation of the shared task employs both traditional metrics (BLEU, ROUGE, BERTScore, Novelty) and JudgeLM based on LLM. We present a detailed analysis of our results, including an empirical evaluation of the model performance and comprehensive score distributions across evaluation metrics. This work contributes to the growing body of research on multilingual counterspeech generation, offering insights into developing robust models that can adapt to diverse linguistic and cultural contexts in the fight against online hate speech., Comment: to be published in MCG-COLING's 2025 conference proceedings

Despite the global prevalence of Modern Standard Chinese language, counterspeech (CS) resources for Chinese remain virtually nonexistent. To address this gap in East Asian counterspeech research we introduce the a corpus of Modern Standard Mandarin counterspeech that focuses on combating hate speech in Mainland China. This paper proposes a novel approach of generating CS by using an LLM-as-a-Judge, simulated annealing, LLMs zero-shot CN generation and a round-robin algorithm. This is followed by manual verification for quality and contextual relevance. This paper details the methodology for creating effective counterspeech in Chinese and other non-Eurocentric languages, including unique cultural patterns of which groups are maligned and linguistic patterns in what kinds of discourse markers are programmatically marked as hate speech (HS). Analysis of the generated corpora, we provide strong evidence for the lack of open-source, properly labeled Chinese hate speech data and the limitations of using an LLM-as-Judge to score possible answers in Chinese. Moreover, the present corpus serves as the first East Asian language based CS corpus and provides an essential resource for future research on counterspeech generation and evaluation., Comment: to be published in MCG-COLING 2025's conference proceedings

NREL's computational sciences center hosts the largest high performance computing (HPC) capabilities dedicated to sustainability research while functioning as a living laboratory for sustainable computing. NREL's HPC capabilities support the research needs of the Department of Energy's Office of Energy Efficiency and Renewable Energy (EERE). In ten years of operation, HPC use in EERE-sponsored sustainability research has grown by a factor of 30. This paper analyzes this research portfolio, providing examples of individual use cases. The paper documents NREL's history of operating one of the world's most sustainable data centers while examining pathways to improving sustainability beyond reduction of PUE. This paper concludes by examining the unique opportunities created for sustainable computing research created by combining an HPC system dedicated to sustainability research and a research program in sustainable computing., Comment: Invited submission to IEEE Computing in Science and Engineering

Concerns about the environmental footprint of machine learning are increasing. While studies of energy use and emissions of ML models are a growing subfield, most ML researchers and developers still do not incorporate energy measurement as part of their work practices. While measuring energy is a crucial step towards reducing carbon footprint, it is also not straightforward. This paper introduces the main considerations necessary for making sound use of energy measurement tools and interpreting energy estimates, including the use of at-the-wall versus on-device measurements, sampling strategies and best practices, common sources of error, and proxy measures. It also contains practical tips and real-world scenarios that illustrate how these considerations come into play. It concludes with a call to action for improving the state of the art of measurement methods and standards for facilitating robust comparisons between diverse hardware and software environments., Comment: Released as NREL Tech Report

Ionizable lipids are essential in developing lipid nanoparticles (LNPs) for effective messenger RNA (mRNA) delivery. While traditional methods for designing new ionizable lipids are typically time-consuming, deep generative models have emerged as a powerful solution, significantly accelerating the molecular discovery process. However, a practical challenge arises as the molecular structures generated can often be difficult or infeasible to synthesize. This project explores Monte Carlo tree search (MCTS)-based generative models for synthesizable ionizable lipids. Leveraging a synthetically accessible lipid building block dataset and two specialized predictors to guide the search through chemical space, we introduce a policy network guided MCTS generative model capable of producing new ionizable lipids with available synthesis pathways.

Lipid nanoparticles (LNPs) are vital in modern biomedicine, enabling the effective delivery of mRNA for vaccines and therapies by protecting it from rapid degradation. Among the components of LNPs, ionizable lipids play a key role in RNA protection and facilitate its delivery into the cytoplasm. However, designing ionizable lipids is complex. Deep generative models can accelerate this process and explore a larger candidate space compared to traditional methods. Due to the structural differences between lipids and small molecules, existing generative models used for small molecule generation are unsuitable for lipid generation. To address this, we developed a deep generative model specifically tailored for the discovery of ionizable lipids. Our model generates novel ionizable lipid structures and provides synthesis paths using synthetically accessible building blocks, addressing synthesizability. This advancement holds promise for streamlining the development of lipid-based delivery systems, potentially accelerating the deployment of new therapeutic agents, including mRNA vaccines and gene therapies., Comment: NeurIPS 2024 Workshop on AI for New Drug Modalities

The intracluster medium (ICM) in the far outskirts (r $>$ 2-3 R$_{200}$) of galaxy clusters interfaces with the intergalactic medium (IGM) and is theorized to comprise diffuse, multiphase gas. This medium may hold vital clues to clusters' thermodynamic evolution and far-reaching impacts on infalling, future cluster galaxies. The diffuse outskirts of clusters are well-suited for quasar absorption line observations, capable of detecting gas to extremely low column densities. We analyze 18 QSO spectra observed with the Cosmic Origins Spectrograph aboard the Hubble Space Telescope whose lines of sight trace the gaseous environments of 26 galaxy clusters from within R$_{200}$ to 6 R$_{200}$ in projection. We measure the dN/dz and covering fraction of H I and O VI associated with the foreground clusters as a function of normalized impact parameter. We find the dN/dz for H I is consistent with the IGM field value for all impact parameter bins, with an intriguing slight elevation between 2 and 3 R$_{200}$. The dN/dz for O VI is also consistent with the field value (within 3$\sigma$) for all impact parameter bins, with potential elevations in dN/dz both within 1-2 R$_{200}$ and beyond 4 R$_{200}$ at $>2\sigma$. We propose physical scenarios that may give rise to these tentative excesses, such as a buildup of neutral gas at the outer accretion shock front and a signature of the warm-hot IGM. We do not find a systematic excess of potentially associated galaxies near the sightlines where O VI is detected; thus, the detected O VI does not have a clear circumgalactic origin., Comment: Submitted to ApJ, 33 pages, 11 primary figures, 2 tables, 25 supplementary figures

We extend the recently developed quantum van der Waals quarkyonic matter to non-zero isospin asymmetries by utilizing the two-component van der Waals equation with a generalized excluded volume prescription. The isospin dependence of van der Waals interaction parameters is determined by constraints on the symmetry energy, slope of the symmetry energy, and nuclear ground state properties. We find that the speed of sound has a peak for all values of the asymmetry parameter, signifying a transition to quarkyonic matter. The quarkyonic matter onset density is found to have a mild dependence on isospin asymmetry, with specific details influenced by the isospin dependence of the repulsive interactions. We also incorporate leptonic degrees of freedom and explore the neutron star matter equation of state, calculating mass-radius relations and tidal properties of neutron stars. We find that quarkyonic matter supports heavy neutron stars with a maximum mass of at least 2.6 solar masses. We observe quantitatively different behavior for the excluded volume cases of isospin-blind ($b_{n}=b_{pn}$) and isospin-dependent ($b_{n} \neq b_{pn}$) repulsion, the latter being preferred by observational constraints., Comment: 10 pages, 4 figures

Recent studies have shown that many nonconvex machine learning problems meet a so-called generalized-smooth condition that extends beyond traditional smooth nonconvex optimization. However, the existing algorithms designed for generalized-smooth nonconvex optimization encounter significant limitations in both their design and convergence analysis. In this work, we first study deterministic generalized-smooth nonconvex optimization and analyze the convergence of normalized gradient descent under the generalized Polyak-Lojasiewicz condition. Our results provide a comprehensive understanding of the interplay between gradient normalization and function geometry. Then, for stochastic generalized-smooth nonconvex optimization, we propose an independently-normalized stochastic gradient descent algorithm, which leverages independent sampling, gradient normalization and clipping to achieve an $\mathcal{O}(\epsilon^{-4})$ sample complexity under relaxed assumptions. Experiments demonstrate the fast convergence of our algorithm., Comment: 3 figures, 30 pages

Batched Bayesian optimization (BO) can accelerate molecular design by efficiently identifying top-performing compounds from a large chemical library. Existing acquisition strategies for batch design in BO aim to balance exploration and exploitation. This often involves optimizing non-additive batch acquisition functions, necessitating approximation via myopic construction and/or diversity heuristics. In this work, we propose an acquisition strategy for discrete optimization that is motivated by pure exploitation, qPO (multipoint Probability of Optimality). qPO maximizes the probability that the batch includes the true optimum, which is expressible as the sum over individual acquisition scores and thereby circumvents the combinatorial challenge of optimizing a batch acquisition function. We differentiate the proposed strategy from parallel Thompson sampling and discuss how it implicitly captures diversity. Finally, we apply our method to the model-guided exploration of large chemical libraries and provide empirical evidence that it performs better than or on par with state-of-the-art methods in batched Bayesian optimization.

Making successful use of cloud computing requires nuanced approaches to both system design and deployment methodology, involving reasoning about the elasticity, cost, and security models of cloud services. Building cloud-native applications without a firm understanding of the fundamentals of cloud engineering can leave students susceptible to cost and security pitfalls. Yet, cloud computing is not commonly taught at the undergraduate level. To address this gap, we designed an undergraduate-level course that frames cloud infrastructure deployment as a software engineering practice. Our course featured a number of hands-on assignments that gave students experience with modern, best-practice concepts and tools including infrastructure-as-code (IaC). We describe the design of the course, our experience teaching its initial offering, and provide our reflections on what worked well and potential areas for improvement. Our course material is available at https://infracourse.cloud., Comment: To appear in SIGCSE TS 2025

Bayesian optimization (BO) is a principled approach to molecular design tasks. In this paper we explain three pitfalls of BO which can cause poor empirical performance: an incorrect prior width, over-smoothing, and inadequate acquisition function maximization. We show that with these issues addressed, even a basic BO setup is able to achieve the highest overall performance on the PMO benchmark for molecule design (Gao et al 2022). These results suggest that BO may benefit from more attention in the machine learning for molecules community., Comment: 8 pages, 4 figures. ICML 2024 AI for science workshop (https://openreview.net/forum?id=V4aG4wsoIt). Code at: https://github.com/AustinT/basic-mol-bo-workshop2024

Large language models (LLMs) have shown impressive effectiveness in various software engineering tasks, including automated program repair (APR). In this study, we take a deep dive into automated bug fixing utilizing LLMs. In contrast to many deep learning-based APR methods that assume known bug locations, rely on line-level localization tools, or address bug prediction and fixing in one step, our approach uniquely employs LLMs to predict bug location at the token level and subsequently utilizes them for bug fixing. This methodological separation of bug localization and fixing using different LLMs enables effective integration of diverse contextual information and improved incorporation of inductive biases. We introduce Toggle: Token-Granulated Bug Localization and Repair, a comprehensive program repair framework that integrates a bug localization model, an adjustment unit, and a bug-fixing model. Toggle takes a buggy function as input and generates a complete corrected function. We investigate various styles of prompting to the bug fixing model to identify the most effective prompts that better utilize the inductive bias and significantly outperform others. Toggle achieves the new state-of-the-art (SOTA) performance on the CodeXGLUE code refinement benchmark, and exhibits better and comparable performance on several other widely-used APR datasets, including Defects4J.

This paper introduces a nonconvex approach for sparse signal recovery, proposing a novel model termed the $\tau_2$-model, which utilizes the squared $\ell_1/\ell_2$ norms for this purpose. Our model offers an advancement over the $\ell_0$ norm, which is often computationally intractable and less effective in practical scenarios. Grounded in the concept of effective sparsity, our approach robustly measures the number of significant coordinates in a signal, making it a powerful alternative for sparse signal estimation. The $\tau_2$-model is particularly advantageous due to its computational efficiency and practical applicability. We detail two accompanying algorithms based on Dinkelbach's procedure and a difference of convex functions strategy. The first algorithm views the model as a linear-constrained quadratic programming problem in noiseless scenarios and as a quadratic-constrained quadratic programming problem in noisy scenarios. The second algorithm, capable of handling both noiseless and noisy cases, is based on the alternating direction linearized proximal method of multipliers. We also explore the model's properties, including the existence of solutions under certain conditions, and discuss the convergence properties of the algorithms. Numerical experiments with various sensing matrices validate the effectiveness of our proposed model.

Addressing the so-called ``Red-AI'' trend of rising energy consumption by large-scale neural networks, this study investigates the actual energy consumption, as measured by node-level watt-meters, of training various fully connected neural network architectures. We introduce the BUTTER-E dataset, an augmentation to the BUTTER Empirical Deep Learning dataset, containing energy consumption and performance data from 63,527 individual experimental runs spanning 30,582 distinct configurations: 13 datasets, 20 sizes (number of trainable parameters), 8 network ``shapes'', and 14 depths on both CPU and GPU hardware collected using node-level watt-meters. This dataset reveals the complex relationship between dataset size, network structure, and energy use, and highlights the impact of cache effects. We propose a straightforward and effective energy model that accounts for network size, computing, and memory hierarchy. Our analysis also uncovers a surprising, hardware-mediated non-linear relationship between energy efficiency and network design, challenging the assumption that reducing the number of parameters or FLOPs is the best way to achieve greater energy efficiency. Highlighting the need for cache-considerate algorithm development, we suggest a combined approach to energy efficient network, algorithm, and hardware design. This work contributes to the fields of sustainable computing and Green AI, offering practical guidance for creating more energy-efficient neural networks and promoting sustainable AI., Comment: 25 pages, 8 figures, for associated dataset see https://data.openei.org/submissions/5991

The Software Supply Chain (SSC) has captured considerable attention from attackers seeking to infiltrate systems and undermine organizations. There is evidence indicating that adversaries utilize Social Engineering (SocE) techniques specifically aimed at software developers. That is, they interact with developers at critical steps in the Software Development Life Cycle (SDLC), such as accessing Github repositories, incorporating code dependencies, and obtaining approval for Pull Requests (PR) to introduce malicious code. This paper aims to comprehensively explore the existing and emerging SocE tactics employed by adversaries to trick Software Engineers (SWEs) into delivering malicious software. By analyzing a diverse range of resources, which encompass established academic literature and real-world incidents, the paper systematically presents an overview of these manipulative strategies within the realm of the SSC. Such insights prove highly beneficial for threat modeling and security gap analysis., Comment: 7 pages, 2 figures

While stellar processes are believed to be the main source of feedback in dwarf galaxies, the accumulating discoveries of AGN in dwarf galaxies over recent years arouse the interest to also consider AGN feedback in them. Fast, AGN-driven outflows, a major mechanism of AGN feedback, have indeed been discovered in dwarf galaxies and may be powerful enough to provide feedback to their dwarf hosts. In this paper, we search for outflows traced by the blueshifted ultraviolet absorption features in three dwarf galaxies with AGN from the sample examined in our previous ground-based study. We confirm outflows traced by blueshifted absorption features in two objects and tentatively detect an outflow in the third object. In one object where the outflow is clearly detected in multiple species, photoionization modeling suggests that this outflow is located $\sim$0.5 kpc from the AGN, implying a galactic-scale impact. This outflow is much faster and possesses higher kinetic energy outflow rate than starburst-driven outflows in sources with similar star formation rates, and is likely energetic enough to provide negative feedback to its host galaxy as predicted by simulations. Much broader ($\sim$4000 km s$^{-1}$) absorption features are also discovered in this object which may have the same origin as that of broad absorption lines in quasars. Additionally, strong He II $\lambda$1640 emission is detected in both objects where the transition falls in the wavelength coverage, and is consistent with an AGN origin. In one of these two objects, blueshifted He II emission line is clearly detected, likely tracing a highly-ionized AGN wind., Comment: 22 pages, 14 figures, 2 tables, accepted for publication in ApJ

This abstract explores classroom Human-Robot Interaction (HRI) scenarios with an emphasis on the adaptation of human-inspired social gaze models in robot cognitive architecture to facilitate a more seamless social interaction. First, we detail the HRI scenarios explored by us in our studies followed by a description of the social gaze model utilized for our research. We highlight the advantages of utilizing such an attentional model in classroom HRI scenarios. We also detail the intended goals of our upcoming study involving this social gaze model., Comment: In WTF Workshop Proceedings (arXiv:2401.04108) held in conjunction with the ACM conference on Conversational User Interfaces (CUI), 19 - 21/07 2023, in Eindhoven, The Netherlands

We perform the first measurement of the thermal and ionization state of the intergalactic medium (IGM) across 0.9 < z < 1.5 using 301 \lya absorption lines fitted from 12 HST STIS quasar spectra, with a total pathlength of \Delta z=2.1. We employ the machine-learning-based inference method that uses joint b-N distributions obtained from \lyaf decomposition. Our results show that the HI photoionization rates, \Gamma, are in good agreement with the recent UV background synthesis models, with \log (\Gamma/s^{-1})={-11.79}^{0.18}_{-0.15}, -11.98}^{0.09}_{-0.09}, and {-12.32}^{0.10}_{-0.12} at z=1.4, 1.2, and 1 respectively. We obtain the IGM temperature at the mean density, T_0, and the adiabatic index, \gamma, as [\log (T_0/K), \gamma]= [{4.13}^{+0.12}_{-0.10}, {1.34}^{+0.10}_{-0.15}], [{3.79}^{+0.11}_{-0.11}, {1.70}^{+0.09}_{-0.09}] and [{4.12}^{+0.15}_{-0.25}, {1.34}^{+0.21}_{-0.26}] at z=1.4, 1.2 and 1 respectively. Our measurements of T_0 at z=1.4 and 1.2 are consistent with the expected trend from z<3 temperature measurements as well as theoretical expectations that, in the absence of any non-standard heating, the IGM should cool down after HeII reionization. Whereas, our T_0 measurements at z=1 show unexpectedly high IGM temperature. However, because of the relatively large uncertainty in these measurements of the order of \Delta T_0~5000 K, mostly emanating from the limited redshift path length of available data in these bins, we can not definitively conclude whether the IGM cools down at z<1.5. Lastly, we generate a mock dataset to test the constraining power of future measurement with larger datasets. The results demonstrate that, with redshift pathlength \Delta z \sim 2 for each redshift bin, three times the current dataset, we can constrain the T_0 of IGM within 1500K. Such precision would be sufficient to conclusively constrain the history of IGM thermal evolution at z < 1.5.

As is well known, both sampling from the posterior and computing the mean of the posterior in Gaussian process regression reduces to solving a large linear system of equations. We study the use of stochastic gradient descent for solving this linear system, and show that when \emph{done right} -- by which we mean using specific insights from the optimisation and kernel communities -- stochastic gradient descent is highly effective. To that end, we introduce a particularly simple \emph{stochastic dual descent} algorithm, explain its design in an intuitive manner and illustrate the design choices through a series of ablation studies. Further experiments demonstrate that our new method is highly competitive. In particular, our evaluations on the UCI regression tasks and on Bayesian optimisation set our approach apart from preconditioned conjugate gradients and variational Gaussian process approximations. Moreover, our method places Gaussian process regression on par with state-of-the-art graph neural networks for molecular binding affinity prediction.

Automated Synthesis Planning has recently re-emerged as a research area at the intersection of chemistry and machine learning. Despite the appearance of steady progress, we argue that imperfect benchmarks and inconsistent comparisons mask systematic shortcomings of existing techniques, and unnecessarily hamper progress. To remedy this, we present a synthesis planning library with an extensive benchmarking framework, called syntheseus, which promotes best practice by default, enabling consistent meaningful evaluation of single-step models and multi-step planning algorithms. We demonstrate the capabilities of syntheseus by re-evaluating several previous retrosynthesis algorithms, and find that the ranking of state-of-the-art models changes in controlled evaluation experiments. We end with guidance for future works in this area, and call the community to engage in the discussion on how to improve benchmarks for synthesis planning., Comment: Accepted for publication in Faraday Discussions

Retrosynthesis is the task of planning a series of chemical reactions to create a desired molecule from simpler, buyable molecules. While previous works have proposed algorithms to find optimal solutions for a range of metrics (e.g. shortest, lowest-cost), these works generally overlook the fact that we have imperfect knowledge of the space of possible reactions, meaning plans created by algorithms may not work in a laboratory. In this paper we propose a novel formulation of retrosynthesis in terms of stochastic processes to account for this uncertainty. We then propose a novel greedy algorithm called retro-fallback which maximizes the probability that at least one synthesis plan can be executed in the lab. Using in-silico benchmarks we demonstrate that retro-fallback generally produces better sets of synthesis plans than the popular MCTS and retro* algorithms., Comment: ICLR 2024 camera ready version (https://openreview.net/forum?id=dl0u4ODCuW). 58 pages total. Code available at: https://github.com/AustinT/retro-fallback-iclr24. This version has 1) updated writing 2) updated figures 3) additional experimental results 4) more complete explanation of AND/OR graphs in the appendices 5) correct typos + error in fig G.5 caption

Generating molecules, both in a directed and undirected fashion, is a huge part of the drug discovery pipeline. Genetic algorithms (GAs) generate molecules by randomly modifying known molecules. In this paper we show that GAs are very strong algorithms for such tasks, outperforming many complicated machine learning methods: a result which many researchers may find surprising. We therefore propose insisting during peer review that new algorithms must have some clear advantage over GAs, which we call the GA criterion. Ultimately our work suggests that a lot of research in molecule generation should be re-assessed., Comment: Currently under review. Code will be made available at a later date