arXiv subjects

Andrew Morgan

Publications and source records attributed to Andrew Morgan.

13 recordsLinked to original sources

arXiv2026

EgoInfinity: A Web-Scale 4D Hand-Object Interaction Data Engine for Any-View Robot Retargeting and Video-to-Action Robot Learning

Internet videos constitute the largest reservoir of embodied human manipulation knowledge, yet converting arbitrary RGB footage into actionable robot training data remains a major bottleneck. Existing lab- or factory-collected datasets are narrow in scale and diversity, limiting open-world robot learning. Instead of proposing a static dataset, we introduce EgoInfinity, a universal 4D hand-object interaction data engine that enables web-scale data generation for robot retargeting and learning. EgoInfinity is a modular engine integrating perception, segmentation, reconstruction, interaction-aware refinement, and retargeting to automate this traditionally unscalable video-to-action problem without human-in-the-loop annotation. Its modular design lets the engine continuously benefit from advances in any incorporated component. With EgoInfinity, in-the-wild human manipulation videos are lifted into agent-agnostic, metric 4D hand-object representations, including hand trajectories, 6-DoF object poses, and contact-relevant states. Rather than naively connecting standalone components, EgoInfinity combines cross-module metric calibration with interaction-aware refinement to improve physical reliability, reducing drift and contact inconsistencies common in pure visual reconstruction. We further propose a novel motion retargeter that compiles the recovered 3D hand motions into executable joint trajectories for diverse robot morphologies, enabling video-to-action retargeting on any robot from arbitrary viewpoints and shot sizes (e.g., the human body is only partially visible). We validate EgoInfinity across perception fidelity, kinematic feasibility, contact consistency, cross-embodiment generalization, and real-robot skill acquisition (e.g., grasping, cutting, wiping, and pouring), demonstrating a scalable bridge from internet videos to executable robot behavior for open-world robot learning.

cs.RO↗

arXiv2026

Can we Watermark Low-Entropy LLM Outputs?

A recent and exciting thread of work focuses on developing methods for watermarking the output of large language models (LLMs). We focus on provably undetectable watermarking-that is, schemes that do not alter the output distribution of the LLM, yet enable embedding a watermark in the output that identifies the output as having been generated by the particular LLM. Furthermore, the watermark should be hard to remove by an adversary that may potentially edit, insert, or delete tokens from the watermarked output. Indeed, recent work (Christ et al. [COLT'24], Christ et al. [CRYPTO'24], Golowich et al. [NeuroIPS'24]) shows how to develop such schemes that are robust against a constant fraction of substitutions, or even against a constant fraction of arbitrary edits. These works, however, make strong assumptions on the entropy present in the output of the LLM. Most notably, they all require constant entropy rate-that is, a constant fraction of the tokens in a sufficiently long substring of the output need to have empirical entropy at least O(log |T|), where T is the alphabet of tokens, and Golowich et al. additionally require T to be larger than the security parameter. In this work, we consider whether we can also watermark the outputs of LLMs when the per-token entropy is just a constant, discarding the dependence on the alphabet size or security parameter. In this regime, we construct: - A watermarking scheme robust against random substitutions (assuming subexponential LPN, as in Christ et al. [CRYPTO'24]) - A watermarking scheme robust against random substitutions and random deletions, given either the additional heuristic assumption that the output of the LLM only introduces random errors (analogous to the assumption made by Christ et al. [CRYPTO'24]) or a construction of a pseudorandom error-correcting code robust to adversarial substitutions and random deletions.

cs.CR↗

arXiv2022

Lower Bound Techniques in the Comparison-Query Model and Inversion Minimization on Trees

Given a rooted tree and a ranking of its leaves, what is the minimum number of inversions of the leaves that can be attained by ordering the tree? This variation of the problem of counting inversions in arrays originated in mathematical psychology, with the evaluation of the Mann--Whitney statistic for detecting differences between distributions as a special case. We study the complexity of the problem in the comparison-query model, used for problems like sorting and selection. For many types of trees with $n$ leaves, we establish lower bounds close to the strongest known in the model, namely the lower bound of $\log_2(n!)$ for sorting $n$ items. We show: (a) $\log_2((\alpha(1-\alpha)n)!) - O(\log n)$ queries are needed whenever the tree has a subtree that contains a fraction $\alpha$ of the leaves. This implies a lower bound of $\log_2((\frac{k}{(k+1)^2}n)!) - O(\log n)$ for trees of degree $k$. (b) $\log_2(n!) - O(\log n)$ queries are needed in case the tree is binary. (c) $\log_2(n!) - O(k \log k)$ queries are needed for certain classes of trees of degree $k$, including perfect trees with even $k$. The lower bounds are obtained by developing two novel techniques for a generic problem $\Pi$ in the comparison-query model and applying them to inversion minimization on trees. Both techniques can be described in terms of the Cayley graph of the symmetric group with adjacent-rank transpositions as the generating set. Consider the subgraph consisting of the edges between vertices with the same value under $\Pi$. We show that the size of any decision tree for $\Pi$ must be at least: (i) the number of connected components of this subgraph, and (ii) the factorial of the average degree of the complementary subgraph, divided by $n$. Lower bounds on query complexity then follow by taking the base-2 logarithm.

cs.DS↗

arXiv2022

Polynomial Identity Testing via Evaluation of Rational Functions

$ \newcommand{\ie}{i.\,e.} $We introduce a hitting set generator for Polynomial Identity Testing based on evaluations of low-degree univariate rational functions at abscissas associated with the variables. We establish an equivalence up to rescaling with a generator introduced by Shpilka and Volkovich, which has a similar structure but uses multivariate polynomials. We initiate a systematic analytic study of the power of hitting set generators by characterizing their vanishing ideals, \ie, the sets of polynomials that they fail to hit. We provide two such characterizations for our generator. First, we develop a small collection of polynomials that jointly produce the vanishing ideal. As corollaries, we obtain tight bounds on the minimum degree, sparseness, and partition class size of set-multilinearity in the vanishing ideal. Second, inspired by a connection to alternating algebra, we develop a structured deterministic membership test for the multilinear part of the vanishing ideal. We present a derivation based on alternating algebra along with the required background, as well as one in terms of zero substitutions and partial derivatives, avoiding the need for alternating algebra. As evidence of the utility of our analytic approach, we rederive known derandomization results based on the generator by Shpilka and Volkovich and present a new application in derandomization / lower bounds for read-once oblivious algebraic branching programs.

cs.CC↗

arXiv2020

New aerodynamic lens injector for single particle diffractive imaging

An aerodynamic lens injector was developed specifically for the needs of single-particle diffractive imaging experiments at free-electron lasers. Its design allows for quick changes of injector geometries and focusing properties in order to optimize injection for specific individual samples. Here, we present results of its first use at the FLASH free-electron-laser facility. Recorded diffraction patterns of polystyrene spheres are modeled using Mie scattering, which allowed for the characterization of the particle beam under diffractive-imaging conditions and yield good agreement with particle-trajectory simulations.

physics.ins-det↗

arXiv2019

X-ray diffractive imaging of controlled gas-phase molecules: Toward imaging of dynamics in the molecular frame

We report experimental results on the diffractive imaging of three-dimensionally aligned 2,5-diiodothiophene molecules. The molecules were aligned by chirped near-infrared laser pulses, and their structure was probed at a photon energy of 9.5 keV ($\lambda\approx130 \text{pm}$) provided by the Linac Coherent Light Source. Diffracted photons were recorded on the CSPAD detector and a two-dimensional diffraction pattern of the equilibrium structure of 2,5-diiodothiophene was recorded. The retrieved distance between the two iodine atoms agrees with the quantum-chemically calculated molecular structure to within 5 %. The experimental approach allows for the imaging of intrinsic molecular dynamics in the molecular frame, albeit this requires more experimental data which should be readily available at upcoming high-repetition-rate facilities.

physics.atm-clus↗

arXiv2017

Paradoxes in Fair Computer-Aided Decision Making

Computer-aided decision making--where a human decision-maker is aided by a computational classifier in making a decision--is becoming increasingly prevalent. For instance, judges in at least nine states make use of algorithmic tools meant to determine "recidivism risk scores" for criminal defendants in sentencing, parole, or bail decisions. A subject of much recent debate is whether such algorithmic tools are "fair" in the sense that they do not discriminate against certain groups (e.g., races) of people. Our main result shows that for "non-trivial" computer-aided decision making, either the classifier must be discriminatory, or a rational decision-maker using the output of the classifier is forced to be discriminatory. We further provide a complete characterization of situations where fair computer-aided decision making is possible.

cs.LG↗

arXiv2017

Minimum Circuit Size, Graph Isomorphism, and Related Problems

We study the computational power of deciding whether a given truth-table can be described by a circuit of a given size (the Minimum Circuit Size Problem, or MCSP for short), and of the variant denoted as MKTP where circuit size is replaced by a polynomially-related Kolmogorov measure. All prior reductions from supposedly-intractable problems to MCSP / MKTP hinged on the power of MCSP / MKTP to distinguish random distributions from distributions produced by hardness-based pseudorandom generator constructions. We develop a fundamentally different approach inspired by the well-known interactive proof system for the complement of Graph Isomorphism (GI). It yields a randomized reduction with zero-sided error from GI to MKTP. We generalize the result and show that GI can be replaced by any isomorphism problem for which the underlying group satisfies some elementary properties. Instantiations include Linear Code Equivalence, Permutation Group Conjugacy, and Matrix Subspace Conjugacy. Along the way we develop encodings of isomorphism classes that are efficiently decodable and achieve compression that is at or near the information-theoretic optimum; those encodings may be of independent interest.

cs.CC↗

arXiv2017

Femtosecond X-ray Fourier holography imaging of free-flying nanoparticles

Ultrafast X-ray imaging provides high resolution information on individual fragile specimens such as aerosols, metastable particles, superfluid quantum systems and live biospecimen, which is inaccessible with conventional imaging techniques. Coherent X-ray diffractive imaging, however, suffers from intrinsic loss of phase, and therefore structure recovery is often complicated and not always uniquely-defined. Here, we introduce the method of in-flight holography, where we use nanoclusters as reference X-ray scatterers in order to encode relative phase information into diffraction patterns of a virus. The resulting hologram contains an unambiguous three-dimensional map of a virus and two nanoclusters with the highest lat- eral resolution so far achieved via single shot X-ray holography. Our approach unlocks the benefits of holography for ultrafast X-ray imaging of nanoscale, non-periodic systems and paves the way to direct observation of complex electron dynamics down to the attosecond time scale.

physics.atm-clus↗

arXiv2017

Continuous Diffraction of Molecules and Disordered Molecular Crystals

The diffraction pattern of a single non-periodic compact object, such as a molecule, is continuous and is proportional to the square modulus of the Fourier transform of that object. When arrayed in a crystal, the coherent sum of the continuous diffracted wave-fields from all objects gives rise to strong Bragg peaks that modulate the single-object transform. Wilson statistics describe the distribution of continuous diffraction intensities to the same extent that they apply to Bragg diffraction. The continuous diffraction obtained from translationally-disordered molecular crystals consists of the incoherent sum of the wave-fields from the individual rigid units (such as molecules) in the crystal, which is proportional to the incoherent sum of the diffraction from the rigid units in each of their crystallographic orientations. This sum over orientations modifies the statistics in a similar way that crystal twinning modifies the distribution of Bragg intensities. These statistics are applied to determine parameters of continuous diffraction such as its scaling, the beam coherence, and the number of independent wave-fields or object orientations contributing. Continuous diffraction is generally much weaker than Bragg diffraction and may be accompanied by a background that far exceeds the strength of the signal. Instead of just relying upon the smallest measured intensities to guide the subtraction of the background it is shown how all measured values can be utilised to estimate the background, noise, and signal, by employing a modified "noisy Wilson" distribution that explicitly includes the background. Parameters relating to the background and signal quantities can be estimated from the moments of the measured intensities. The analysis method is demonstrated on previously-published continuous diffraction data measured from imperfect crystals of photosystem II.

physics.optics↗

arXiv2017

Femtosecond x-ray diffraction from an aerosolized beam of protein nanocrystals

We demonstrate near-atomic-resolution Bragg diffraction from aerosolized single granulovirus crystals using an x-ray free-electron laser. The form of the aerosol injector is nearly identical to conventional liquid-microjet nozzles, but the x-ray-scattering background is reduced by several orders of magnitude by the use of helium carrier gas rather than liquid. This approach provides a route to study the weak diffuse or lattice-transform signal arising from small crystals. The high speed of the particles is particularly well suited to upcoming MHz-repetition-rate x-ray free-electron lasers.

physics.bio-ph↗

arXiv2015

Graph Isomorphism and Circuit Size

It is well-known [KST93] that the complexity of the Graph Automorphism problem is characterized by a special case of Graph Isomorphism, where the input graphs satisfy the "promise" of being rigid (that is, having no nontrivial automorphisms). In this brief note, we observe that the reduction of Graph Automorphism to the Rigid Graph Ismorphism problem can be accomplished even using Grollman and Selman's notion of a "smart reduction".

cs.CC↗

arXiv2015

Strongly aligned gas-phase molecules at Free-Electron Lasers

We demonstrate a novel experimental implementation to strongly align molecules at full repetition rates of free-electron lasers. We utilized the available in-house laser system at the coherent x-ray imaging beamline at the Linac Coherent Light Source. Chirped laser pulses, i. e., the direct output from the regenerative amplifier of the Ti:Sa chirped pulse amplification laser system, were used to strongly align 2,5-diiodothiophene molecules in a molecular beam. The alignment laser pulses had pulse energies of a few mJ and a pulse duration of 94 ps. A degree of alignment of $\left<\cos^2\!\theta_{2D}\right>$ = 0.85 was measured, limited by the intrinsic temperature of the molecular beam rather than by the available laser system. With the general availability of synchronized chirped-pulse-amplified near-infrared laser systems at short-wavelength laser facilities, our approach allows for the universal preparation of molecules tightly fixed in space for experiments with x-ray pulses.

physics.atm-clus↗