arXiv subjects

David Broman

Publications and source records attributed to David Broman.

At least 19 recordsLinked to original sources

arXiv2026

Reactive Graphs for Efficient Markov Chain Monte Carlo Inference in Probabilistic Programming Languages

An important aspect of making inference based on a probabilistic program practical is efficiency; faster evaluation enables more work per unit of time, which can be translated into more precision. Inference via Markov chain Monte Carlo has a property that can be favorably exploited for efficiency: most proposed samples are computed as minor variations of previous samples, i.e., a clever implementation can skip computations pertaining to what is unchanged. This paper provides an approach for automatically translating a probabilistic program to a dynamic graph, reminiscent of functional reactive programming, that explicitly represents data dependencies, enabling proposals to only recompute the parts of the graph that depend on redrawn random variables. The graph-building interface follows familiar functional programming interfaces, which also connect to their expressiveness in terms of probabilistic programming: models using the applicative functor portion express Bayesian networks, while those using monads represent universal probabilistic programming languages.

cs.PL↗

arXiv2026

Advantage-Guided Diffusion for Model-Based Reinforcement Learning

Model-based reinforcement learning (MBRL) with autoregressive world models suffers from compounding errors, whereas diffusion world models mitigate this by generating trajectory segments jointly. However, existing diffusion guides are either policy-only, discarding value information, or reward-based, which becomes myopic when the diffusion horizon is short. We introduce Advantage-Guided Diffusion for MBRL (AGD-MBRL), which steers the reverse diffusion process using the agent's advantage estimates so that sampling concentrates on trajectories expected to yield higher long-term return beyond the generated window. We develop two guides: (i) Sigmoid Advantage Guidance (SAG) and (ii) Exponential Advantage Guidance (EAG). We prove that a diffusion model guided through SAG or EAG allows us to perform reweighted sampling of trajectories with weights increasing in state-action advantage-implying policy improvement under standard assumptions. Additionally, we show that the trajectories generated from AGD-MBRL follow an improved policy (that is, with higher value) compared to an unguided diffusion model. AGD integrates seamlessly with PolyGRAD-style architectures by guiding the state components while leaving action generation policy-conditioned, and requires no change to the diffusion training objective. On MuJoCo control tasks (HalfCheetah, Hopper, Walker2D and Reacher), AGD-MBRL improves sample efficiency and final return over PolyGRAD, an online Diffuser-style reward guide, and model-free baselines (PPO/TRPO), in some cases by a margin of 2x. These results show that advantage-aware guidance is a simple, effective remedy for short-horizon myopia in diffusion-model MBRL.

cs.AI↗

arXiv2026

Learning to Rank the Initial Branching Order of SAT Solvers

Finding good branching orders is key to solving SAT problems efficiently, but finding such branching orders is a difficult problem. Using a learning based approach to predict a good branching order before solving, therefore, has potential. In this paper, we investigate predicting branching orders using graph neural networks as a preprocessing step to conflict-driven clause learning (CDCL) SAT solvers. We show that there are significant gains to be made in existing CDCL SAT solvers by providing a good initial branching. Further, we provide three labeling methods to find such initial branching orders in a tractable way. Finally, we train a graph neural network to predict these branching orders and show through our evaluations that a GNN-initialized ordering yields significant speedups on random 3-CNF and pseudo-industrial benchmarks, with generalization capabilities to instances much larger than the training set. However, we also find that the predictions fail at speeding up more difficult and industrial instances. We attribute this to the solver's dynamic heuristics, which rapidly overwrite the provided initialization, and to the complexity of these instances, making GNN prediction hard.

cs.AI↗

arXiv2025

DiagramIR: An Automatic Pipeline for Educational Math Diagram Evaluation

Large Language Models (LLMs) are increasingly being adopted as tools for learning; however, most tools remain text-only, limiting their usefulness for domains where visualizations are essential, such as mathematics. Recent work shows that LLMs are capable of generating code that compiles to educational figures, but a major bottleneck remains: scalable evaluation of these diagrams. We address this by proposing DiagramIR: an automatic and scalable evaluation pipeline for geometric figures. Our method relies on intermediate representations (IRs) of LaTeX TikZ code. We compare our pipeline to other evaluation baselines such as LLM-as-a-Judge, showing that our approach has higher agreement with human raters. This evaluation approach also enables smaller models like GPT-4.1-Mini to perform comparably to larger models such as GPT-5 at a 10x lower inference cost, which is important for deploying accessible and scalable education technologies.

cs.AI↗

arXiv2025

A Matter of Interest: Understanding Interestingness of Math Problems in Humans and Language Models

The evolution of mathematics is shaped importantly by interestingness: researchers choose which problems to pursue, and students choose which problems to engage with, based on expectations of interest and challenge. As AI systems, particularly large language models (LLMs) that operate flexibly over natural language and formal mathematics, are increasingly used in mathematics research and education, it becomes crucial to characterize how closely their judgments align with people from different mathematical backgrounds. We study whether LLMs align with human interestingness judgments by comparing LLM ratings with those of two populations, crowdsourced participants with college math experience and International Math Olympiad competitors. Although many LLMs broadly agree with human notions of interestingness, they largely fail to match the distribution of human judgments. They also weakly align with why humans find problems interesting, with low correlation to human-selected rationales. Finally, we evaluate LLMs' ability to generate interesting problems and find that, after filtering for validity, LLMs are able to generate engaging problems. We conclude with takeaways, including the need for multi-LLM human-AI collaborative systems, that highlight both the promise and current limits of LLMs as partners in mathematical reasoning.

cs.AI↗

arXiv2025

Fast Collection Operations from Indexed Stream Fusion

We present a system of efficient methods for traversing and combining associative collection data structures. A distinguishing feature of the system is that, like traditional sequential iterator libraries, it does not require specialized compiler infrastructure or staged compilation for efficiency and composability. By using a representation based on indexed streams, the library can express complex joins over input collections while using no intermediate allocations. We implement the library for the Lean, Morphic, and Rust programming languages and provide a mechanized proof of functional correctness in Lean.

cs.PL↗

arXiv2025

Adaptive Reinforcement Learning for Unobservable Random Delays

In standard reinforcement learning (RL) settings, the interaction between the agent and the environment is typically modeled as a Markov decision process (MDP), which assumes that the agent observes the system state instantaneously, selects an action without delay, and executes it immediately. In real-world dynamic environments, such as cyber-physical systems, this assumption often breaks down due to delays in the interaction between the agent and the system. These delays can vary stochastically over time and are typically unobservable when deciding on an action. Existing methods deal with this uncertainty conservatively by assuming a known fixed upper bound on the delay, even if the delay is often much lower. In this work, we introduce the interaction layer, a general framework that enables agents to adaptively handle unobservable and time-varying delays. Specifically, the agent generates a matrix of possible future actions, anticipating a horizon of potential delays, to handle both unpredictable delays and lost action packets sent over networks. Building on this framework, we develop a model-based algorithm, Actor-Critic with Delay Adaptation (ACDA), which dynamically adjusts to delay patterns. Our method significantly outperforms state-of-the-art approaches across a wide range of locomotion benchmark environments, including real-world measured delays.

cs.LG↗

arXiv2025

CoreDPPL: Towards a Sound Composition of Differentiation, ODE Solving, and Probabilistic Programming

In recent years, there has been extensive research on how to extend general-purpose programming language semantics with domain-specific modeling constructs. Two areas of particular interest are (i) universal probabilistic programming where Bayesian probabilistic models are encoded as programs, and (ii) differentiable programming where differentiation operators are first class or differential equations are part of the language semantics. These kinds of languages and their language constructs are usually studied separately or composed in restrictive ways. In this paper, we study and formalize the combination of probabilistic programming constructs, first-class differentiation, and ordinary differential equations in a higher-order setting. We propose formal semantics for a core of such differentiable probabilistic programming language (DPPL), where the type system tracks random computations and rejects unsafe compositions during type checking. The semantics and its type system are formalized, mechanized, and proven sound in Agda with respect to abstract language constructs.

cs.PL↗

arXiv2024

Repr Types: One Abstraction to Rule Them All

The choice of how to represent an abstract type can have a major impact on the performance of a program, yet mainstream compilers cannot perform optimizations at such a high level. When dealing with optimizations of data type representations, an important feature is having extensible representation-flexible data types; the ability for a programmer to add new abstract types and operations, as well as concrete implementations of these, without modifying the compiler or a previously defined library. Many research projects support high-level optimizations through static analysis, instrumentation, or benchmarking, but they are all restricted in at least one aspect of extensibility. This paper presents a new approach to representation-flexible data types without such restrictions and which still finds efficient optimizations. Our approach centers around a single built-in type $\texttt{repr}$ and function overloading with cost annotations for operation implementations. We evaluate our approach (i) by defining a universal collection type as a library, a single type for all conventional collections, and (ii) by designing and implementing a representation-flexible graph library. Programs using $\texttt{repr}$ types are typically faster than programs with idiomatic representation choices -- sometimes dramatically so -- as long as the compiler finds good implementations for all operations. Our compiler performs the analysis efficiently by finding optimized solutions quickly and by reusing previous results to avoid recomputations.

cs.PL↗

arXiv2024

Learning Formal Mathematics From Intrinsic Motivation

How did humanity coax mathematics from the aether? We explore the Platonic view that mathematics can be discovered from its axioms - a game of conjecture and proof. We describe Minimo (Mathematics from Intrinsic Motivation): an agent that jointly learns to pose challenging problems for itself (conjecturing) and solve them (theorem proving). Given a mathematical domain axiomatized in dependent type theory, we first combine methods for constrained decoding and type-directed synthesis to sample valid conjectures from a language model. Our method guarantees well-formed conjectures by construction, even as we start with a randomly initialized model. We use the same model to represent a policy and value function for guiding proof search. Our agent targets generating hard but provable conjectures - a moving target, since its own theorem proving ability also improves as it trains. We propose novel methods for hindsight relabeling on proof search trees to significantly improve the agent's sample efficiency in both tasks. Experiments on 3 axiomatic domains (propositional logic, arithmetic and group theory) demonstrate that our agent can bootstrap from only the axioms, self-improving in generating true and challenging conjectures and in finding proofs.

cs.AI↗

arXiv2024

Automatic Tracing in Task-Based Runtime Systems

Implicitly parallel task-based runtime systems often perform dynamic analysis to discover dependencies in and extract parallelism from sequential programs. Dependence analysis becomes expensive as task granularity drops below a threshold. Tracing techniques have been developed where programmers annotate repeated program fragments (traces) issued by the application, and the runtime system memoizes the dependence analysis for those fragments, greatly reducing overhead when the fragments are executed again. However, manual trace annotation can be brittle and not easily applicable to complex programs built through the composition of independent components. We introduce Apophenia, a system that automatically traces the dependence analysis of task-based runtime systems, removing the burden of manual annotations from programmers and enabling new and complex programs to be traced. Apophenia identifies traces dynamically through a series of dynamic string analyses, which find repeated program fragments in the stream of tasks issued to the runtime system. We show that Apophenia is able to come between 0.92x--1.03x the performance of manually traced programs, and is able to effectively trace previously untraced programs to yield speedups of between 0.91x--2.82x on the Perlmutter and Eos supercomputers.

cs.DC↗

arXiv2024

Optimizing Instructions and Demonstrations for Multi-Stage Language Model Programs

Language Model Programs, i.e. sophisticated pipelines of modular language model (LM) calls, are increasingly advancing NLP tasks, but they require crafting prompts that are jointly effective for all modules. We study prompt optimization for LM programs, i.e. how to update these prompts to maximize a downstream metric without access to module-level labels or gradients. To make this tractable, we factorize our problem into optimizing the free-form instructions and few-shot demonstrations of every module and introduce several strategies to craft task-grounded instructions and navigate credit assignment across modules. Our strategies include (i) program- and data-aware techniques for proposing effective instructions, (ii) a stochastic mini-batch evaluation function for learning a surrogate model of our objective, and (iii) a meta-optimization procedure in which we refine how LMs construct proposals over time. Using these insights we develop MIPRO, a novel algorithm for optimizing LM programs. MIPRO outperforms baseline optimizers on five of seven diverse multi-stage LM programs using a best-in-class open-source model (Llama-3-8B), by as high as 13% accuracy. We have released our new optimizers and benchmark in DSPy at http://dspy.ai

cs.CL↗

arXiv2024

Scorch: A Library for Sparse Deep Learning

The rapid growth in the size of deep learning models strains the capabilities of traditional dense computation paradigms. Leveraging sparse computation has become increasingly popular for training and deploying large-scale models, but existing deep learning frameworks lack extensive support for sparse operations. To bridge this gap, we introduce Scorch, a library that seamlessly integrates efficient sparse tensor computation into the PyTorch ecosystem, with an initial focus on inference workloads on CPUs. Scorch provides a flexible and intuitive interface for sparse tensors, supporting diverse sparse data structures. Scorch introduces a compiler stack that automates key optimizations, including automatic loop ordering, tiling, and format inference. Combined with a runtime that adapts its execution to both dense and sparse data, Scorch delivers substantial speedups over hand-written PyTorch Sparse (torch.sparse) operations without sacrificing usability. More importantly, Scorch enables efficient computation of complex sparse operations that lack hand-optimized PyTorch implementations. This flexibility is crucial for exploring novel sparse architectures. We demonstrate Scorch's ease of use and performance gains on diverse deep learning models across multiple domains. With only minimal code changes, Scorch achieves 1.05-5.78x speedups over PyTorch Sparse on end-to-end tasks. Scorch's seamless integration and performance gains make it a valuable addition to the PyTorch ecosystem. We believe Scorch will enable wider exploration of sparsity as a tool for scaling deep learning and inform the development of other sparse libraries.

cs.LG↗

arXiv2023

Real-Time Probabilistic Programming

Complex cyber-physical systems interact in real-time and must consider both timing and uncertainty. Developing software for such systems is expensive and difficult, especially when modeling, inference, and real-time behavior must be developed from scratch. Recently, a new kind of language has emerged -- called probabilistic programming languages (PPLs) -- that simplify modeling and inference by separating the concerns between probabilistic modeling and inference algorithm implementation. However, these languages have primarily been designed for offline problems, not online real-time systems. In this paper, we combine PPLs and real-time programming primitives by introducing the concept of real-time probabilistic programming languages (RTPPL). We develop an RTPPL called ProbTime and demonstrate its usability on an automotive testbed performing indoor positioning and braking. Moreover, we study fundamental properties and design alternatives for runtime behavior, including a new fairness-guided approach that automatically optimizes the accuracy of a ProbTime system under schedulability constraints.

cs.PL↗

arXiv2023

Exact Worst-Case Execution-Time Analysis for Implicit Model Predictive Control

We propose the first method that determines the exact worst-case execution time (WCET) for implicit linear model predictive control (MPC). Such WCET bounds are imperative when MPC is used in real time to control safety-critical systems. The proposed method applies when the quadratic programming solver in the MPC controller belongs to a family of well-established active-set solvers. For such solvers, we leverage a previously proposed complexity certification framework to generate a finite set of archetypal optimization problems; we prove that these archetypal problems form an execution-time equivalent cover of all possible problems; that is, that they capture the execution time for solving any possible optimization problem that can be encountered online. Hence, by solving just these archetypal problems on the hardware on which the MPC is to be deployed, and by recording the execution times, we obtain the exact WCET. In addition to providing formal proofs of the methods efficacy, we validate the method on an MPC example where an inverted pendulum on a cart is stabilized. The experiments highlight the following advantages compared with classical WCET methods: (i) in contrast to classical static methods, our method gives the exact WCET; (ii) in contrast to classical measurement-based methods, our method guarantees a correct WCET estimate and requires fewer measurements on the hardware.

eess.SY↗

arXiv2023

Suspension Analysis and Selective Continuation-Passing Style for Universal Probabilistic Programming Languages

Universal probabilistic programming languages (PPLs) make it relatively easy to encode and automatically solve statistical inference problems. To solve inference problems, PPL implementations often apply Monte Carlo inference algorithms that rely on execution suspension. State-of-the-art solutions enable execution suspension either through (i) continuation-passing style (CPS) transformations or (ii) efficient, but comparatively complex, low-level solutions that are often not available in high-level languages. CPS transformations introduce overhead due to unnecessary closure allocations -- a problem the PPL community has generally overlooked. To reduce overhead, we develop a new efficient selective CPS approach for PPLs. Specifically, we design a novel static suspension analysis technique that determines parts of programs that require suspension, given a particular inference algorithm. The analysis allows selectively CPS transforming the program only where necessary. We formally prove the correctness of the analysis and implement the analysis and transformation in the Miking CorePPL compiler. We evaluate the implementation for a large number of Monte Carlo inference algorithms on real-world models from phylogenetics, epidemiology, and topic modeling. The evaluation results demonstrate significant improvements across all models and inference algorithms.

cs.PL↗

arXiv2023

Automatic Alignment in Higher-Order Probabilistic Programming Languages

Probabilistic Programming Languages (PPLs) allow users to encode statistical inference problems and automatically apply an inference algorithm to solve them. Popular inference algorithms for PPLs, such as sequential Monte Carlo (SMC) and Markov chain Monte Carlo (MCMC), are built around checkpoints -- relevant events for the inference algorithm during the execution of a probabilistic program. Deciding the location of checkpoints is, in current PPLs, not done optimally. To solve this problem, we present a static analysis technique that automatically determines checkpoints in programs, relieving PPL users of this task. The analysis identifies a set of checkpoints that execute in the same order in every program run -- they are aligned. We formalize alignment, prove the correctness of the analysis, and implement the analysis as part of the higher-order functional PPL Miking CorePPL. By utilizing the alignment analysis, we design two novel inference algorithm variants: aligned SMC and aligned lightweight MCMC. We show, through real-world experiments, that they significantly improve inference execution time and accuracy compared to standard PPL versions of SMC and MCMC.

cs.PL↗

arXiv2022

Expression Acceleration: Seamless Parallelization of Typed High-Level Languages

Efficient parallelization of algorithms on general-purpose GPUs is essential in many areas today. However, it is a non-trivial task for software engineers to utilize GPUs to improve the performance of high-level programs in general. Although many domain-specific approaches are available for GPU acceleration, it is difficult to accelerate existing high-level programs without rewriting parts of the programs using low-level GPU code. We present a compiler implementation using an alternative approach called expression acceleration. This approach marks expressions for acceleration, and the compiler automatically infers which dependent code needs to be accelerated. We design and implement a compiler supporting expression acceleration for a statically typed functional language and evaluate its applicability and performance.

cs.PL↗