Source author record

Anindya Banerjee

Anindya Banerjee appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Logic in Computer Science Programming Languages hep-th Machine Learning Artificial Intelligence astro-ph.CO Software Engineering

Catalog footprint

What is connected

11works

7topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2022arXiv

A Relational Program Logic with Data Abstraction and Dynamic Framing

Dedicated to Tony Hoare. In a paper published in 1972 Hoare articulated the fundamental notions of hiding invariants and simulations. Hiding: invariants on encapsulated data representations need not be mentioned in specifications that comprise the API of a module. Simulation: correctness of a new data representation and implementation can be established by proving simulation between the old and new implementations using a coupling relation defined on the encapsulated state. These results were formalized semantically and for a simple model of state, though the paper claimed this could be extended to encompass dynamically allocated objects. In recent years, progress has been made towards formalizing the claim, for simulation, though mainly in semantic developments. In this article, hiding and simulation are combined with the idea in Hoare's 1969 paper: a logic of programs. For an object-based language with dynamic allocation, we introduce a relational Hoare logic with stateful frame conditions that formalizes encapsulation, hiding of invariants, and couplings that relate two implementations. Relations and other assertions are expressed in first-order logic. Specifications can express a wide range of relational properties such as conditional equivalence and noninterference with declassification. The proof rules facilitate relational reasoning by means of convenient alignments and are shown sound with respect to a conventional operational semantics. A derived proof rule for equivalence of linked programs directly embodies representation independence. Applicability to representative examples is demonstrated using an SMT-based implementation.

preprint2021arXiv

A Formal Proof of PAC Learnability for Decision Stumps

We present a formal proof in Lean of probably approximately correct (PAC) learnability of the concept class of decision stumps. This classic result in machine learning theory derives a bound on error probabilities for a simple type of classifier. Though such a proof appears simple on paper, analytic and measure-theoretic subtleties arise when carrying it out fully formally. Our proof is structured so as to separate reasoning about deterministic properties of a learning function from proofs of measurability and analysis of probabilities.

preprint2021arXiv

On Algebraic Abstractions for Concurrent Separation Logics

Concurrent separation logic is distinguished by transfer of state ownership upon parallel composition and framing. The algebraic structure that underpins ownership transfer is that of partial commutative monoids (PCMs). Extant research considers ownership transfer primarily from the logical perspective while comparatively less attention is drawn to the algebraic considerations. This paper provides an algebraic formalization of ownership transfer in concurrent separation logic by means of structure-preserving partial functions (i.e., morphisms) between PCMs, and an associated notion of separating relations. Morphisms of structures are a standard concept in algebra and category theory, but haven't seen ubiquitous use in separation logic before. Separating relations are binary relations that generalize disjointness and characterize the inputs on which morphisms preserve structure. The two abstractions facilitate verification by enabling concise ways of writing specs, by providing abstract views of threads' states that are preserved under ownership transfer, and by enabling user-level construction of new PCMs out of existing ones.

preprint2020arXiv

Hyperkähler Isometries Of K3 Surfaces

We consider symmetries of K3 manifolds. Holomorphic symplectic automorphisms of K3 surfaces have been classified, and observed to be subgroups of the Mathieu group $M_{23}$. More recently, automorphisms of K3 sigma models commuting with $SU(2)\times SU(2)$ $R$-symmetry have been classified by Gaberdiel, Hohenegger, and Volpato. These groups are all subgroups of the Conway group. We fill in a small gap in the literature and classify the possible hyperkähler isometry groups of K3 manifolds. There is an explicit list of $40$ possible groups, all of which are realized in the moduli space. The groups are all subgroups of $M_{23}$.

preprint2020arXiv

Verification of ML Systems via Reparameterization

As machine learning is increasingly used in essential systems, it is important to reduce or eliminate the incidence of serious bugs. A growing body of research has developed machine learning algorithms with formal guarantees about performance, robustness, or fairness. Yet, the analysis of these algorithms is often complex, and implementing such systems in practice introduces room for error. Proof assistants can be used to formally verify machine learning systems by constructing machine checked proofs of correctness that rule out such bugs. However, reasoning about probabilistic claims inside of a proof assistant remains challenging. We show how a probabilistic program can be automatically represented in a theorem prover using the concept of \emph{reparameterization}, and how some of the tedious proofs of measurability can be generated automatically from the probabilistic program. To demonstrate that this approach is broad enough to handle rather different types of machine learning systems, we verify both a classic result from statistical learning theory (PAC-learnability of decision stumps) and prove that the null model used in a Bayesian hypothesis test satisfies a fairness criterion called demographic parity.

preprint2016arXiv

Hoare-style Specifications as Correctness Conditions for Non-linearizable Concurrent Objects

Designing scalable concurrent objects, which can be efficiently used on multicore processors, often requires one to abandon standard specification techniques, such as linearizability, in favor of more relaxed consistency requirements. However, the variety of alternative correctness conditions makes it difficult to choose which one to employ in a particular case, and to compose them when using objects whose behaviors are specified via different criteria. The lack of syntactic verification methods for most of these criteria poses challenges in their systematic adoption and application. In this paper, we argue for using Hoare-style program logics as an alternative and uniform approach for specification and compositional formal verification of safety properties for concurrent objects and their client programs. Through a series of case studies, we demonstrate how an existing program logic for concurrency can be employed off-the-shelf to capture important state and history invariants, allowing one to explicitly quantify over interference of environment threads and provide intuitive and expressive Hoare-style specifications for several non-linearizable concurrent objects that were previously specified only via dedicated correctness criteria. We illustrate the adequacy of our specifications by verifying a number of concurrent client scenarios, that make use of the previously specified concurrent objects, capturing the essence of such correctness conditions as concurrency-aware linearizability, quiescent, and quantitative quiescent consistency. All examples described in this paper are verified mechanically in Coq.

preprint2016arXiv

Relational Logic with Framing and Hypotheses: Technical Report

Relational properties arise in many settings: relating two versions of a program that use different data representations, noninterference properties for security, etc. The main ingredient of relational verification, relating aligned pairs of intermediate steps, has been used in numerous guises, but existing relational program logics are narrow in scope. This paper introduces a logic based on novel syntax that weaves together product programs to express alignment of control flow points at which relational formulas are asserted. Correctness judgments feature hypotheses with relational specifications, discharged by a rule for the linking of procedure implementations. The logic supports reasoning about program-pairs containing both similar and dissimilar control and data structures. Reasoning about dynamically allocated objects is supported by a frame rule based on frame conditions amenable to SMT provers. We prove soundness and sketch how the logic can be used for data abstraction, loop optimizations, and secure information flow.

preprint2015arXiv

Constraining parameters of effective field theory of inflation from Planck data

The Cosmic Microwave Background can provide information regarding physics of the very early universe, more specifically, of the matter-radiation distribution of the inflationary era. Starting from the effective field theory of inflation, we use the Goldstone action to calculate the three point correlation function for the Goldstone field, whose results can be directly applied to the field describing the curvature perturbations around a de Sitter solution for the inflationary era. We then use the data from the recent Planck mission for the parameters $f_{NL}^{equil}$ and $f_{NL}^{orthog}$ which parametrize the size and shape of non-Gaussianities generated in single field models of inflation. Using these known values, we calculate the parameters relevant to our analysis, $f_{NL}^{\dotπ^3}$, $f_{NL}^{\dotπ(\partial _i π)^2}$ and the speed of sound $c_s$ which parametrize the non-Gaussianities arising from two different kinds of generalized interactions of the scalar field in question.

preprint2014arXiv

Specifying and Verifying Concurrent Algorithms with Histories and Subjectivity

We present a lightweight approach to Hoare-style specifications for fine-grained concurrency, based on a notion of time-stamped histories that abstractly capture atomic changes in the program state. Our key observation is that histories form a partial commutative monoid, a structure fundamental for representation of concurrent resources. This insight provides us with a unifying mechanism that allows us to treat histories just like heaps in separation logic. For example, both are subject to the same assertion logic and inference rules (e.g., the frame rule). Moreover, the notion of ownership transfer, which usually applies to heaps, has an equivalent in histories. It can be used to formally represent helping---an important design pattern for concurrent algorithms whereby one thread can execute code on behalf of another. Specifications in terms of histories naturally abstract granularity, in the sense that sophisticated fine-grained algorithms can be given the same specifications as their simplified coarse-grained counterparts, making them equally convenient for client-side reasoning. We illustrate our approach on a number of examples and validate all of them in Coq.

preprint2013arXiv

A Simple Semantics and Static Analysis for Stack Inspection

The Java virtual machine and the .NET common language runtime feature an access control mechanism specified operationally in terms of run-time stack inspection. We give a denotational semantics in "eager" form, and show that it is equivalent to the "lazy" semantics using stack inspection. We give a static analysis of safety, i.e., the absence of security errors, that is simpler than previous proposals. We identify several program transformations that can be used to remove run-time checks. We give complete, detailed proofs for safety of the analysis and for the transformations, exploiting compositionality of the eager semantics.