arXiv subjects

Pavel Panchekha

Publications and source records attributed to Pavel Panchekha.

At least 19 recordsLinked to original sources

arXiv2026

Accurate Residues for Floating-Point Debugging

Floating-point arithmetic is error-prone and unintuitive. Floating-point debuggers instrument programs to monitor floating-point arithmetic at run time and flag numerical issues. They estimate residues, i.e., the difference between actual floating-point and ideal real values, for every floating-point value in the program. Prior work explores various approaches for computing these residues accurately and efficiently. Unfortunately, the most efficient methods, based on "error-free transformations", have a high rate of false reports, while the most accurate methods, based on high-precision arithmetic, are very slow. This paper builds on error-free-transformations-based approaches and aims to improve their accuracy while preserving efficiency. To more accurately compute residues, this paper divides residue computation into two steps (rounding error computation and residue function evaluation) and shows how to perform each step accurately via careful improvements to the current state of the art. We evaluate on 44 large scientific computing workloads, focusing on the 14 benchmarks where prior tools produce false reports: our approach eliminates false reports on 10 benchmarks and substantially reduces them on the remaining 3 benchmarks. Moreover, complex numerical issues require additional care due to absorption, where two machine-precision residues cannot both be computed accurately in a single execution. This paper introduces residue override, which re-executes the program multiple times, computing different residues in different executions and assembling a final "patchwork" execution. We evaluate on 169 standard benchmarks drawn from numerical analysis papers and textbooks, requiring only 3.6 re-executions on average. Among 34 benchmarks with false reports in the initial run, residue override is triggered on 29 of them and reduces false reports on 25 of them, averaging 7.1 re-executions.

cs.MS↗

arXiv2026

Numerical Superoptimization for Library Learning

Numerical software depends on fast, accurate implementations of mathematical primitives like sin, exp, and log. Modern superoptimizers can optimize floating-point kernels against a given set of such primitives, but a more fundamental question remains open: which new primitives are worth implementing in the first place? We formulate this as numerical library learning: given a workload of floating-point kernels, identify the mathematical primitives whose expert implementations would most improve speed and accuracy. Our key insight is that numerical superoptimizers already have the machinery well-suited to this problem. Their search procedures happen to enumerate candidate primitives, their equivalence procedures can generalize and deduplicate candidates, and their cost models can estimate counterfactual utility: how much the workload would improve if a given primitive were available. We present GrowLibm, which repurposes the Herbie superoptimizer as a numerical library learner. GrowLibm mines candidate primitives from the superoptimizer's intermediate search results, ranks them by counterfactual utility, and prunes redundant candidates. Across three scientific applications (PROJ, CoolProp, and Basilisk), GrowLibm identifies compact, reusable primitives that can be implemented effectively using standard numerical techniques. When Herbie is extended with these expert implementations, kernel speed improves by up to 2.2x at fixed accuracy, and maximum achievable accuracy also improves, in one case from 56.0% to 93.5%. We also prototype an LLVM matcher that recognizes learned primitives in optimized IR, recovering 26 replacement sites across five PROJ projections and improving end-to-end application performance by up to 5%.

cs.PL↗

arXiv2026

Semantics for 2D Rasterization

Rasterization is the process of determining the color of every pixel drawn by an application. Powerful rasterization libraries like Skia, CoreGraphics, and Direct2D put exceptional effort into drawing, blending, and rendering efficiently. Yet applications are still hindered by the inefficient sequences of operations that they ask these libraries to perform. Even Google Chrome, a highly optimized program co-developed with the Skia rasterization library, still produces inefficient instruction sequences even on the top 100 most visited websites. The underlying reason for this inefficiency is that rasterization libraries have complex semantics and opaque and non-obvious execution models. To address this issue, we introduce $μ$Skia, a formal semantics for the Skia 2D graphics library, and mechanize this semantics in Lean. $μ$Skia covers language and graphics features like canvas state, the layer stack, blending, and color filters, and the semantics itself is split into three strata to separate concerns and enable extensibility. We then identify four patterns of sub-optimal Skia code produced by Google Chrome, and then write replacements for each pattern. $μ$Skia allows us to verify the replacements are correct, including identifying numerous tricky side conditions. We then develop a high-performance Skia optimizer that applies these patterns to speed up rasterization. On 99 Skia programs gathered from the top 100 websites, this optimizer yields a speedup of 18.7% over Skia's most modern GPU backend, while taking at most 32 $μ$s for optimization. The speedups persist across a variety of websites, Skia backends, and GPUs. To provide true, end-to-end verification, optimization traces produced by the optimizer are loaded back into the $μ$Skia semantics and translation validated in Lean.

cs.PL↗

arXiv2026

Incremental Live Programming via Shortcut Memoization

Live programming systems aim to quickly show programmers the dynamic impacts of program edits. To do so, they re-execute the program whenever it is edited, which poses a computational challenge when programs become large or complex. This has led to the need for incrementality in the implementation of live program interpreters. This paper introduces Chordata, an incremental program interpreter based on shortcut memoization, which learns repeated patterns of computation, called shortcuts, by observing executions of previous versions of a program. It can then apply these shortcuts when the same or a structurally similar program fragment is re-executed. This paper contributes a formal semantics of shortcut memoization for any language with a rewrite-based semantics, with mechanized proofs of key correctness properties. We then express a variant of the Hazel live programming system, expressed as a CEK machine, in Chordata, and develop a number of practical heuristics to learn high-value shortcuts. We evaluate the resulting system on editing traces of students solving simple programming problems. Chordata achieves a speedup of 13.03\times compared to baseline with a 19.97\times memory overhead. For smaller changes and for more complex programs, Chordata achieves even greater speedups. Furthermore, we show that Chordata is capable of providing a speedup even within a single execution, with a faster speedup on a larger input.

cs.PL↗

arXiv2025

Spineless Traversal for Layout Invalidation

Latency is a major concern for web rendering engines like those in Chrome, Safari, and Firefox. These engines reduce latency by using an incremental layout algorithm to redraw the page when the user interacts with it. In such an algorithm, elements that change frame-to-frame are marked dirty, and only those elements are processed to draw the next frame, dramatically reducing latency. However, the standard incremental layout algorithm must search the page for dirty elements, accessing auxiliary elements in the process. These auxiliary elements add cache misses and stalled cycles, and are responsible for a sizable fraction of all layout latency. We introduce a new, faster incremental layout algorithm called Spineless Traversal. Spineless Traversal uses a cache-friendlier priority queue algorithm that avoids accessing auxiliary nodes and thus reduces cache traffic and stalls. This leads to dramatic speedups on the most latency-critical interactions such as hovering, typing, and animation. Moreover, thanks to numerous low-level optimizations, Spineless Traversal is competitive across the whole spectrum of incremental layout workloads. Spineless Traversal is faster than the standard approach on 83.0% of 2216 benchmarks, with a mean speedup of 1.80x concentrated in the most latency-critical interactions.

cs.PL↗

arXiv2025

Fast Mixed-Precision Real Evaluation

Evaluating real-valued expressions to high precision is a key building block in computational mathematics, physics, and numerics. A typical implementation evaluates the whole expression in a uniform precision, doubling that precision until a sufficiently-accurate result is achieved. This is wasteful: usually only a few operations really need to be performed at high precision, and the bulk of the expression could be computed much faster. However, such non-uniform precision assignments have, to date, been impractical to compute. We propose a fast new algorithm for deriving such precision assignments. The algorithm leverages results computed at lower precisions to analytically determine a mixed-precision assignment that will result in a sufficiently-accurate result. Our implementation, Reval, achieves an average speed-up of 1.72x compared to the state-of-the-art Sollya tool, with the speed-up increasing to 5.21x on the most difficult input points. An examination of the precisions used with and without precision tuning shows that the speed-up results from assigning lower precisions for the majority of operations, though additional optimizations enabled by the non-uniform precision assignments also play a role.

cs.MS↗

arXiv2025

Fast Mixed-Precision Real Evaluation

math.NA↗

arXiv2025

Incremental Bidirectional Typing via Order Maintenance

Live programming environments provide various semantic services, including type checking and evaluation, continuously as the user is editing the program. The live paradigm promises to improve the developer experience, but liveness is an implementation challenge particularly when working with large programs. This paper specifies and efficiently implements a system the is able to incrementally update type information for a live program in response to fine-grained program edits. This information includes type error marks and information about the expected and actual type on every expression. The system is specified type-theoretically as a small-step dynamics that propagates updates through the marked and annotated program. Most updates flow according to a base bidirectional type system. Additional pointers are maintained to connect bound variables to their binding locations, with edits traversing these pointers directly. Order maintenance data structures are employed to efficiently maintain these pointers and to prioritize the order of update propagation. We prove this system is equivalent to naive re-analysis in the Agda theorem prover, along with other important metatheoretic properties. We then implement it efficiently in OCaml, detailing a number of impactful optimizations. We evaluate this implementation's performance with a large stress-test and find that it is able to achieve dramatic speed-ups of 275.96$\times$ compared to from-scratch reanalysis.

cs.PL↗

arXiv2025

Mixing Condition Numbers and Oracles for Accurate Floating-point Debugging

Recent advances have made numeric debugging tools much faster by using double-double oracles, and numeric analysis tools much more accurate by using condition numbers. But these techniques have downsides: double-double oracles have correlated error so miss floating-point errors while condition numbers cannot cleanly handle over- and under- flow. We combine both techniques to avoid these downsides. Our combination, EXPLANIFLOAT, computes condition numbers using double-double arithmetic, which avoids correlated errors. To handle over- and under- flow, it introduces a separate logarithmic oracle. As a result, EXPLANIFLOAT achieves a precision of 80.0% and a recall of 96.1% on a collection of 546 difficult numeric benchmarks: more accurate than double-double oracles yet dramatically faster than arbitrary-precision condition number computations.

math.NA↗

arXiv2024

Target-Aware Implementation of Real Expressions

New low-precision accelerators, vector instruction sets, and library functions make maximizing accuracy and performance of numerical code increasingly challenging. Two lines of work$\unicode{x2013}$traditional compilers and numerical compilers$\unicode{x2013}$attack this problem from opposite directions. Traditional compiler backends optimize for specific target environments but are limited in their ability to balance performance and accuracy. Numerical compilers trade off accuracy and performance, or even improve both, but ignore the target environment. We join aspects of both to produce Chassis, a target-aware numerical compiler. Chassis compiles mathematical expressions to operators from a target description, which lists the real expressions each operator approximates and estimates its cost and accuracy. Chassis then uses an iterative improvement loop to optimize for speed and accuracy. Specifically, a new instruction selection modulo equivalence algorithm efficiently searches for faster target-specific programs, while a new cost-opportunity heuristic supports iterative improvement. We demonstrate Chassis' capabilities on 9 different targets, including hardware ISAs, math libraries, and programming languages. Chassis finds better accuracy and performance trade-offs than both Clang (by 3.5x) or Herbie (by up to 2.0x) by leveraging low-precision accelerators, accuracy-optimized numerical helper functions, and library subcomponents.

cs.PL↗

arXiv2023

Implementation and Synthesis of Math Library Functions

Achieving speed and accuracy for math library functions like exp, sin, and log is difficult. This is because low-level implementation languages like C do not help math library developers catch mathematical errors, build implementations incrementally, or separate high-level and low-level decision making. This ultimately puts development of such functions out of reach for all but the most experienced experts. To address this, we introduce MegaLibm, a domain-specific language for implementing, testing, and tuning math library implementations. MegaLibm is safe, modular, and tunable. Implementations in MegaLibm can automatically detect mathematical mistakes like sign flips via semantic wellformedness checks, and components like range reductions can be implemented in a modular, composable way, simplifying implementations. Once the high-level algorithm is done, tuning parameters like working precisions and evaluation schemes can be adjusted through orthogonal tuning parameters to achieve the desired speed and accuracy. MegaLibm also enables math library developers to work interactively, compiling, testing, and tuning their implementations and invoking tools like Sollya and type-directed synthesis to complete components and synthesize entire implementations. MegaLibm can express 8 state-of-the-art math library implementations with comparable speed and accuracy to the original C code, and can synthesize 5 variations and 3 from-scratch implementations with minimal guidance.

cs.PL↗

arXiv2023

Odyssey: An Interactive Workbench for Expert-Driven Floating-Point Expression Rewriting

In recent years, researchers have proposed a number of automated tools to identify and improve floating-point rounding error in mathematical expressions. However, users struggle to effectively apply these tools. In this paper, we work with novices, experts, and tool developers to investigate user needs during the expression rewriting process. We find that users follow an iterative design process. They want to compare expressions on multiple input ranges, integrate and guide various rewriting tools and understand where errors come from. We organize this investigation's results into a three-stage workflow and implement that workflow in a new, extensible workbench dubbed Odyssey. Odyssey enables users to: (1) diagnose problems in an expression, (2) generate solutions automatically or by hand, and (3) tune their results. Odyssey tracks a working set of expressions and turns a state-of-the-art automated tool "inside out," giving the user access to internal heuristics, algorithms, and functionality. In a user study, Odyssey enabled five expert numerical analysts to solve challenging rewriting problems where state-of-the-art automated tools fail. In particular, the experts unanimously praised Odyssey's novel support for interactive range modification and local error visualization.

cs.HC↗

arXiv2022

Optimal Heap Limits for Reducing Browser Memory Use

Garbage-collected language runtimes carefully tune heap limits to reduce garbage collection time and memory usage. However, there's a trade-off: a lower heap limit reduces memory use but increases garbage collection time. Classic methods for setting heap limits include manually tuned heap limits and multiple-of-live-size rules of thumb, but it is not clear when one rule is better than another or how to compare them. We address this problem with a new framework where heap limits are set for multiple heaps at once. Our key insight is that every heap limit rule induces a particular allocation of memory across multiple processes, and this allocation can be sub-optimal. We use our framework to derive an optimal "square-root" heap limit rule, which minimizes total memory usage for any amount of total garbage collection time. Paradoxically, the square-root heap limit rule achieves this coordination without communication: it allocates memory optimally across multiple heaps without requiring any communication between those heaps. To demonstrate that this heap limit rule is effective, we prototype it for V8, the JavaScript runtime used in Google Chrome, Microsoft Edge, and other browsers, as well as in server-side frameworks like node.js and Deno. On real-world web pages, our prototype achieves reductions of approximately 16.0% of memory usage while keeping garbage collection time constant. On memory-intensive benchmarks, reductions of up to 30.0% of garbage collection time are possible with no change in total memory usage.

cs.PL↗

arXiv2022

Small Proofs from Congruence Closure

Satisfiability Modulo Theory (SMT) solvers and equality saturation engines must generate proof certificates from e-graph-based congruence closure procedures to enable verification and conflict clause generation. Smaller proof certificates speed up these activities. Though the problem of generating proofs of minimal size is known to be NP-complete, existing proof minimization algorithms for congruence closure generate unnecessarily large proofs and introduce asymptotic overhead over the core congruence closure procedure. In this paper, we introduce an O(n^5) time algorithm which generates optimal proofs under a new relaxed "proof tree size" metric that directly bounds proof size. We then relax this approach further to a practical O(n \log(n)) greedy algorithm which generates small proofs with no asymptotic overhead. We implemented our techniques in the egg equality saturation toolkit, yielding the first certifying equality saturation engine. We show that our greedy approach in egg quickly generates substantially smaller proofs than the state-of-the-art Z3 SMT solver on a corpus of 3760 benchmarks.

cs.PL↗

arXiv2022

Synthesizing Mathematical Identities with E-Graphs

Identities compactly describe properties of a mathematical expression and can be leveraged into faster and more accurate function implementations. However, identities must currently be discovered manually, which requires a lot of expertise. We propose a two-phase synthesis and deduplication pipeline that discovers these identities automatically. In the synthesis step, a set of rewrite rules is composed, using an e-graph, to discover candidate identities. However, most of these candidates are duplicates, which a secondary deduplication step discards using integer linear programming and another e-graph. Applied to a set of 61 benchmarks, the synthesis phase generates 7215 candidate identities which the deduplication phase then reduces down to 125 core identities.

cs.SC↗

arXiv2021

Faster Math Functions, Soundly

Standard library implementations of functions like sin and exp optimize for accuracy, not speed, because they are intended for general-purpose use. But applications tolerate inaccuracy from cancellation, rounding error, and singularities-sometimes even very high error-and many application could tolerate error in function implementations as well. This raises an intriguing possibility: speeding up numerical code by tuning standard function implementations. This paper thus introduces OpTuner, an automatic method for selecting the best implementation of mathematical functions at each use site. OpTuner assembles dozens of implementations for the standard mathematical functions from across the speed-accuracy spectrum. OpTuner then uses error Taylor series and integer linear programming to compute optimal assignments of function implementation to use site and presents the user with a speed-accuracy Pareto curve they can use to speed up their code. In a case study on the POV-Ray ray tracer, OpTuner speeds up a critical computation, leading to a whole program speedup of 9% with no change in the program output (whereas human efforts result in slower code and lower-quality output). On a broader study of 37 standard benchmarks, OpTuner matches 216 implementations to 89 use sites and demonstrates speed-ups of 107% for negligible decreases in accuracy and of up to 438% for error-tolerant applications.

cs.MS↗

arXiv2021

An Interval Arithmetic for Robust Error Estimation

Interval arithmetic is a simple way to compute a mathematical expression to an arbitrary accuracy, widely used for verifying floating-point computations. Yet this simplicity belies challenges. Some inputs violate preconditions or cause domain errors. Others cause the algorithm to enter an infinite loop and fail to compute a ground truth. Plus, finding valid inputs is itself a challenge when invalid and unsamplable points make up the vast majority of the input space. These issues can make interval arithmetic brittle and temperamental. This paper introduces three extensions to interval arithmetic to address these challenges. Error intervals express rich notions of input validity and indicate whether all or some points in an interval violate implicit or explicit preconditions. Movability flags detect futile recomputations and prevent timeouts by indicating whether a higher-precision recomputation will yield a more accurate result. Andinput search restricts sampling to valid, samplable points, so they are easier to find. We compare these extensions to the state-of-the-art technical computing software Mathematica, and demonstrate that our extensions are able to resolve 60.3% more challenging inputs, return 10.2x fewer completely indeterminate results, and avoid 64 cases of fatal error.

math.NA↗

arXiv2020

egg: Fast and Extensible Equality Saturation

An e-graph efficiently represents a congruence relation over many expressions. Although they were originally developed in the late 1970s for use in automated theorem provers, a more recent technique known as equality saturation repurposes e-graphs to implement state-of-the-art, rewrite-driven compiler optimizations and program synthesizers. However, e-graphs remain unspecialized for this newer use case. Equality saturation workloads exhibit distinct characteristics and often require ad-hoc e-graph extensions to incorporate transformations beyond purely syntactic rewrites. This work contributes two techniques that make e-graphs fast and extensible, specializing them to equality saturation. A new amortized invariant restoration technique called rebuilding takes advantage of equality saturation's distinct workload, providing asymptotic speedups over current techniques in practice. A general mechanism called e-class analyses integrates domain-specific analyses into the e-graph, reducing the need for ad hoc manipulation. We implemented these techniques in a new open-source library called egg. Our case studies on three previously published applications of equality saturation highlight how egg's performance and flexibility enable state-of-the-art results across diverse domains.

cs.PL↗