arXiv subjects

Adam Smith

Publications and source records attributed to Adam Smith.

At least 19 recordsLinked to original sources

arXiv2026

Cuscuton Fifth Force: Implications of Non-Dynamical Scalars for Structure Growth

The cuscuton is a scalar field that obeys a constraint rather than an evolution equation, so it modifies gravity without adding a propagating scalar degree of freedom. We study the cosmology of effective cuscuton theory when matter is minimally coupled in the Jordan frame, and therefore conformally coupled to the cuscuton in the Einstein-frame. We derive the background and linear perturbation equations and show that the algebraic cuscuton response mediates a fifth force even though the scalar itself does not propagate. The same conformal factor also changes Einstein-frame particle masses and the friction of matter perturbations, tying the strength and sign of the force to the background evolution rather than to an independent coupling. We study a power-law family of models which contain both tracking solutions and sign-switching solutions; in the latter the effective cuscuton density runs from negative during matter domination to a positive, cosmological-constant-like value today. Using a modified version of CLASS, including the induced particle-mass dependence of recombination, we illustrate the corresponding changes to CMB temperature and linear matter power spectra for benchmark models, without fitting to data. In the strongest benchmarks the fifth force changes the effective gravitational coupling on sub-horizon scales by about one percent, smaller than the accompanying changes to the matter power spectrum and the acoustic scale. For the sign-switching solutions this change is negative over most of cosmic history, so a non-propagating field can weaken rather than strengthen gravitational attraction.

astro-ph.CO↗

arXiv2026

A Simple Presentation of Robust Membership Inference Attacks

In a membership inference attack, an attacker is given the data of a particular individual, and tries to determine if that individual was included in a particular dataset, given some kind of statistics derived from the dataset or a model trained on the dataset. Roughly, a membership inference attack is robust if it succeeds even when the statistics are highly noisy, regardless of how that noise is introduced. We give a simple presentation of robust membership inference attacks for mean estimation, covering both multivariate Gaussian distributions and binary product distributions. We also give a summary of the literature on robust membership inference, covering history, extensions, and connections to other research areas.

cs.CR↗

arXiv2026

The Limits of Post-Selection Generalization

While statistics and machine learning offer numerous methods for ensuring generalization, these methods often fail in the presence of adaptivity---the common practice in which the choice of analysis depends on previous interactions with the same dataset. A recent line of work has introduced powerful, general purpose algorithms that ensure post hoc generalization (also called robust or post-selection generalization), which says that, given the output of the algorithm, it is hard to find any statistic for which the data differs significantly from the population it came from. In this work we show several limitations on the power of algorithms satisfying post hoc generalization. First, we show a tight lower bound on the error of any algorithm that satisfies post hoc generalization and answers adaptively chosen statistical queries, showing a strong barrier to progress in post selection data analysis. Second, we show that post hoc generalization is not closed under composition, despite many examples of such algorithms exhibiting strong composition properties.

cs.LG↗

arXiv2026

Private Evaluation of High-Dimensional Functions

We study differentially private black-box evaluation of arbitrary high-dimensional functions. Given a dataset $S$ and query access to a function $f$---whose sensitivity may be unknown or unbounded---the goal is to privately output an approximation to $f(S)$. The black-box evaluator can only query the value of $f$ over a down neighborhood of $S$ (i.e., those subsets of $S$ that are obtained by removing a bounded number of elements from $S$). We design generic black-box evaluators for both pure and approximate differential privacy through a reduction to high-dimensional private quasi-concave optimization (QCO). Our black-box evaluator enjoys a geometric notion of utility. Consider the vectors obtained by applying $f$ on sets in a down neighborhood of $S$. With high probability, the evaluator's output is guaranteed to be in the convex hull of these vectors. Our work generalizes the result of Linder et al. [LRSS25] for real-valued function evaluation to high dimensions. It also recovers the private center-point guarantees of Beimel et al. [BMNS19], with an explicit additive approximation to maximum Tukey depth. We also apply our evaluators to user-level quasi-concave and convex optimization, showing how they can improve upon the sample-and-aggregate baseline of Nissim et al. [NRS07] by replacing the multiplicative data overhead with an additive overhead. Finally, we show that our formulation of high dimensional black-box evaluation and high dimensional private QCO are equivalent up to a factor of the dimension: the hardest private evaluation instances already arise from the cases that $f$ is a (non-private) quasi-concave optimizer.

stat.ML↗

arXiv2026

Subgraph Counting with Local Node Differential Privacy

We study the problem of approximately counting subgraphs under local node differential privacy (LNDP), a recently introduced model that enforces strong privacy guarantees in analyses of distributed social networks. We design a noninteractive $(\varepsilon,0)$-LNDP algorithm that, given any public $k$-node template graph $H$, counts occurrences of $H$ in a sensitive undirected $n$-node graph with additive error $O(n^{k-\frac{1}{2}}/\varepsilon^{k-1})$, for any constant $k$. We complement this algorithm with a lower bound showing that every noninteractive $(\varepsilon,δ)$-LNDP algorithm for counting copies of any $k$-node template graph with maximum degree greater than $1$ must have error $Ω(n^{k - \frac{1}{2}})$. Thus, our algorithm achieves optimal error for counting constant-size subgraphs, up to $\mathrm{poly}(1/\varepsilon)$ factors. Our results substantially expand the types of network statistics known to be approximable under LNDP: prior work considered only functions of the degree distribution. In addition, our lower bound yields the first known $\mathrm{poly}(n)$ gap between central and local node differential privacy when accuracy is required for all input graphs. Previous separations either only involved $\mathrm{poly}(1/\varepsilon)$ factors or only applied to promise problems.

cs.DS↗

arXiv2026

The Sample Complexity of Membership Inference and Privacy Auditing

A membership-inference attack gets the output of a learning algorithm, and a target individual, and tries to determine whether this individual is a member of the training data or an independent sample from the same distribution. A successful membership-inference attack typically requires the attacker to have some knowledge about the distribution that the training data was sampled from, and this knowledge is often captured through a set of independent reference samples from that distribution. In this work we study how much information the attacker needs for membership inference by investigating the sample complexity-the minimum number of reference samples required-for a successful attack. We study this question in the fundamental setting of Gaussian mean estimation where the learning algorithm is given $n$ samples from a Gaussian distribution $\mathcal{N}(μ,Σ)$ in $d$ dimensions, and tries to estimate $\hatμ$ up to some error $\mathbb{E}[\|\hat μ- μ\|^2_Σ]\leq ρ^2 d$. Our result shows that for membership inference in this setting, $Ω(n + n^2 ρ^2)$ samples can be necessary to carry out any attack that competes with a fully informed attacker. Our result is the first to show that the attacker sometimes needs many more samples than the training algorithm uses to train the model. This result has significant implications for practice, as all attacks used in practice have a restricted form that uses $O(n)$ samples and cannot benefit from $ω(n)$ samples. Thus, these attacks may be underestimating the possibility of membership inference, and better attacks may be possible when information about the distribution is easy to obtain.

cs.LG↗

arXiv2026

Local Node Differential Privacy

We initiate an investigation of node differential privacy for graphs in the local model of private data analysis. In our model, dubbed LNDP*, each node sees its own edge list and releases the output of a local randomizer on this input. These outputs are aggregated by an untrusted server to obtain a final output. We develop a novel algorithmic framework for this setting that allows us to accurately answer arbitrary linear queries about the input graph's degree distribution. Our framework is based on a new object, called the blurry degree distribution, which closely approximates the degree distribution and has lower sensitivity. Instead of answering queries about the degree distribution directly, our algorithms answer queries about the blurry degree distribution. This framework yields accurate LNDP* algorithms for the edge count, PMF and CDF of the degree distribution, and other graph statistics. For some natural problems, our algorithms match the accuracy achievable with node privacy in the central model, where data are held and processed by a trusted server. We also prove lower bounds on the error required by LNDP* algorithms that imply the optimality of our framework for edge counting in sparse graphs and Erdos-Renyi parameter estimation. Our lower bounds apply even to interactive protocols with a constant number of rounds of interaction between the nodes and the server. Existing lower-bound techniques for related models either yield loose bounds or do not apply in our setting, as graph data results in inherently overlapping inputs to local randomizers. To prove our bounds, we develop a splicing argument that stitches together views from locally similar but globally different distributions on graphs to obtain hard instances. Finally, we prove structural results that reveal qualitative differences between local node privacy and the standard local model for tabular data.

cs.DS↗

arXiv2026

RSstitcher -- Merging 2D diffraction frames for Wide Range Reciprocal Space Maps with absorption correction and integration functions

Wide Range Reciprocal Space Mapping (WRRSM) is a technique that allows visualisation of the geometric relationships among multiple hkl spots in a whole reciprocal space map. However, commercial softwares for WRRSMs generation are associated with several issues or limitations, which are overcome by the currently reporting open-source python program RSstitcher (Reciprocal Space Stitcher). RSstitcher merges 2D scan frames formats supported by FabIO and enables WRRSM function on most laboratory X-ray diffractometers equipped with a goniometer cradle and a 2D detector of any sensor size. It is so far the only WRRSMs generation tool that applies diffraction intensity correction due to sample self-absorption, which enables quantitative analyses for WRRSMs including texture measurement and 1D data integration. The conversion equations used in the python program is explained geometrically, including novel WRRSM measurements in ω-ϕ compensated Side Inclination Grazing Incident Diffraction mode for thin film samples. The applications of RSstitcher for bulk and thin film samples are demonstrated using two common 2D X-ray diffraction systems.

physics.ins-det↗

arXiv2026

A Bridge Between Climate Science and Economics: OPTiMEM and the Heat Conjecture for Estimation of Social Cost of Greenhouse Gases

We present an entirely new physics founded approach to estimating the social cost of carbon (SCC). For this, we developed our Ocean-Heat-Content Physics and Time Macro Economic Model (OPTiMEM) to estimate future heat content. The heat conjecture assumes that weather damages curves are stochastically proportional to ocean heat increase. We model carbon combustion, validate to datasets for greenhouse gas (GHG), temperature, and ocean heat content (OHC). We show that the social cost of 4 GHGs: CO2, CH4, N2O and halogenated hydrocarbons, cannot be single values, but must be represented by a kind of economic phase space. We propose very long-term carbon bonds to implement real discounting. This obviates the Gordian knot of the descriptivist versus prescriptivist discount disagreement that is unsolvable. Implementing these bonds leads to a new monitoring metric: real-dollar spending and bond discount rates compared to SC-GHG cost with variation on the discount scale, where the discount has no relationship to the pure rate of time preference (PRTP). This heat conjecture is based on OPTiMEM. OPTiMEM initiates from a fossil fuel consumption function to produce CO2, with 18 scenarios implemented to provide the uncertainty range. We provide 1:N year loss risk models (1:10, 1:100, 1:1000) that government, engineers, and actuaries should find useful. A scenario implementing DICE family of models carbon and growth assumptions shows +18° C is breached by 2210 CE, and +110° C by 2300 CE -- both of which outcomes are obviously not compatible with the fairly rosy conclusions of DICE models. Concerns are raised about having enough low-cost fossil fuel for conversion to minimal CO$_2$ maximal energy return on energy invested (EROEI) power if nations wait too long, and low EROEI power is questioned because monetary value is dependent on energy.

econ.TH↗

arXiv2026

Social Cost of Greenhouse Gases -- OPTiMEM and the Heat Conjecture(s)

Despite well-meaning scenarios that propose global CO2 emissions will decline presented in every IPCC report since 1988, the trend of global CO2 increase continues without significant change. Even if any individual nation manages to flatten its emissions, what matters is the trajectory of the globe. Together the gulf between climate science and climate economics, plus the urgent need for alternative methods of estimation, provided the incentives for development of our Ocean-Heat-Content (OHC) Physics and Time Macro Economic Model (OPTiMEM) system. To link NOAA damages to climate required creating a carbon consumption model to drive a physics model of climate. How fast could carbon be burned and how much coal, oil and natural gas was reasonably available? A carbon model driving climate meant burning the carbon, and modelling how the earth heated up. We developed this using the most recent best greenhouse gas equations and production models for CO2, CH4, N2O, and halogenated gases. This developed an ocean heat content model for the globe. Each step is validated against Known carbon consumption, CO2, temperature, and ocean heat content. This allows a physics founded model of climate costs to be projected.

econ.TH↗

arXiv2026

A Unified Framework for Adversary-Aware Differential Privacy Bounds

Differential Privacy (DP) bounds the privacy leakage of a mechanism against worst-case membership inference, but the precise tradeoff between complex adversarial models and DP protections remains poorly understood. In this paper, we present a unified framework that generalizes the patchwork of existing bounds across membership inference, attribute inference, and data reconstruction attacks. Crucially, our framework is the first to evaluate attacks that target multiple individuals simultaneously and measure success beyond exact matches under a single cohesive bound. Our bounds capture this broad family of previously unexplored attack settings by relying solely on the privacy parameters and the adversary's baseline success rate (i.e. its prior without access to the mechanism's output). To illustrate this, we compare our high-probability guarantees to empirical attacks in two novel settings: extracting multiple non-uniform secrets (passwords and PII) from DP-finetuned language models, and reconstructing tabular data from noisy marginals. Ultimately, this framework provides a rigorous theoretical foundation to investigate the risk landscape of DP algorithms in new adversarial settings.

cs.CR↗

arXiv2026

Robust Statistical Estimators with Bounded Empirical Sensitivity

We introduce a new measure of robustness for statistical estimators, which we call \emph{empirical sensitivity}. An estimator $\hat θ$ has bounded empirical sensitivity if, with high probability over a dataset $X = (X_1, \dots, X_n) \sim \mathcal{D}^{\otimes n}$, for any dataset $Y$ obtained by modifying at most $ηn$ points in $X$, we have that $\hat θ(Y)$ is close to $\hat θ(X)$. We study bounds on this quantity for the prototypical problem of Gaussian mean estimation. We prove new lower bounds, showing that for any estimator $\hat μ$ which achieves an optimal $\ell_2$-error bound of $O\left(\sqrt{d/n}\right)$, the empirical sensitivity is at least $Ω\left(η+ \sqrt{ηd/n}\right)$. The two terms arise due to obstructions on the mean and variance (via an Efron-Stein argument) of such an estimator. We show that this bound is tight up to logarithmic factors, by employing recent results for robust empirical mean estimation.

math.ST↗

arXiv2026

Differentially Private Modeling of Disease Transmission within Human Contact Networks

Epidemiologic studies of infectious diseases often rely on models of contact networks to capture the complex interactions that govern disease spread, and ongoing projects aim to vastly increase the scale at which such data can be collected. However, contact networks may include sensitive information, such as sexual relationships or drug use behavior. Protecting individual privacy while maintaining the scientific usefulness of the data is crucial. We propose a privacy-preserving pipeline for disease spread simulation studies based on a sensitive network that integrates differential privacy (DP) with statistical network models such as stochastic block models (SBMs) and exponential random graph models (ERGMs). Our pipeline comprises three steps: (1) compute network summary statistics using \emph{node-level} DP (which corresponds to protecting individuals' contributions); (2) fit a statistical model, like an ERGM, using these summaries, which allows generating synthetic networks reflecting the structure of the original network; and (3) simulate disease spread on the synthetic networks using an agent-based model. We evaluate the effectiveness of our approach using a simple Susceptible-Infected-Susceptible (SIS) disease model under multiple configurations. We compare both numerical results, such as simulated disease incidence and prevalence, as well as qualitative conclusions such as intervention effect size, on networks generated with and without differential privacy constraints. Our experiments are based on egocentric sexual network data from the ARTNet study (a survey about HIV-related behaviors). Our results show that the noise added for privacy is small relative to other sources of error (sampling and model misspecification). This suggests that, in principle, curators of such sensitive data can provide valuable epidemiologic insights while protecting privacy.

cs.CR↗

arXiv2026

Ensemble of Small Classifiers For Imbalanced White Blood Cell Classification

Automating white blood cell classification for diagnosis of leukaemia is a promising alternative to time-consuming and resource-intensive examination of cells by expert pathologists. However, designing robust algorithms for classification of rare cell types remains challenging due to variations in staining, scanning and inter-patient heterogeneity. We propose a lightweight ensemble approach for classification of cells during Haematopoiesis, with a focus on the biology of Granulopoiesis, Monocytopoiesis and Lymphopoiesis. Through dataset expansion to alleviate some class imbalance, we demonstrate that a simple ensemble of lightweight pretrained SwinV2-Tiny, DinoBloom-Small and ConvNeXT-V2-Tiny models achieves excellent performance on this challenging dataset. We train 3 instantiations of each architecture in a stratified 3-fold cross-validation framework; for an input image, we forward-pass through all 9 models and aggregate through logit averaging. We further reason on the weaknesses of our model in confusing similar-looking myelocytes in granulopoiesis and lymphocytes in lymphopoiesis. Code: https://gitlab.com/siddharthsrivastava/wbc-bench-2026.

cs.CV↗

arXiv2026

Can QCD Axions Survive the Cosmological Constant Problem?

Mechanisms that dynamically relax the vacuum energy offer a concrete way to approach the cosmological constant problem, but because relaxation is not confined to the vacuum energy alone it can have consequences for the rest of low-energy physics. We explore this issue using the recently proposed 'yoga' relaxation models as an explicit framework and show how relaxation differentially suppresses 'slow' physics relative to a characteristic timescale set by the mass of the relaxon. It therefore need not alter e.g. Higgs & collider physics but can dramatically change how light scalar fields participate in cosmology. We revisit the QCD axion in this setting and show that the suppression of the axion's vacuum potential reshapes its behaviour on cosmological timescales while leaving fast, high-energy processes unaffected. The result is to alter the axion mass-coupling relation away from the standard QCD band, driving it into a regime already ruled out by observational constraints. In particular, suppression of the vacuum axion potential allows the QCD matter-induced potential to dominate even for matter densities relevant to cosmology and everyday matter, potentially driving the axion away from the CP-conserving minimum for QCD-motivated parameters. We conclude that conventional QCD axions are unlikely to remain viable in their standard form within vacuum-energy relaxation frameworks.

hep-ph↗

arXiv2026

Multi-Field Dilaton Screening Beyond the Thin-Shell Mechanism

We analyse screening in multi-field scalar-tensor theories, focusing on systems with a dilaton coupled to matter and an axion with a dilaton-dependent kinetic term, in the presence of both planetary and stellar density profiles. Using analytic arguments and fully coupled numerical solutions, we identify a regime in which full screening for a dark-energy-light, effectively unpinned string-dilaton, can occur without fine-tuning. The backreaction of the dilaton's partnered axion field can suppress the exterior scalar charge by selecting a minimum-energy configuration (the BBQ mechanism), yielding robust screening for generic axion gradients. In this regime screening is achieved by cancelling the dilaton's gradient rather than localising it. This reduces the exterior scalar charge and allows for gravity tests in the solar system to be passed. We then show that the more familiar thin-shell intuition need not apply in the multi-field setting. Axion surface gradients can drastically reshape the dilaton profile and drive a more localised transition without generically suppressing the fifth force. The exterior charge can remain essentially unchanged or even be enhanced as the shell is made thinner by a kinetically coupled field. Multi-field two-derivative dynamics therefore decouple localisation in thin shells from screening, evade single-field no-go arguments, and reopen viable parameter space for cosmologically light dilaton-like scalars with strong couplings to matter.

gr-qc↗

arXiv2026

Variational quantum eigensolver for chemical molecules

Solving interacting multi-particle systems is a central challenge in quantum chemistry and condensed matter physics. In this work, we investigate the computation of ground states and ground-state energies for the He-H+ and H2O molecules using quantum computing techniques. We employ the variational quantum eigensolver (VQE), implemented both on a quantum computer simulator and on an IBM quantum device. The resulting energies are benchmarked against exact ground-state energies obtained via classical methods. Simulations of the H2O molecule were performed on Nottingham's High Performance Computing (HPC) facilities.

quant-ph↗

arXiv2025

The Serendipitous Axiodilaton: A Self-Consistent Recombination-Era Solution to the Hubble Tension

Axio-dilaton cosmology provides a minimal benchmark model for both Dark Matter (DM) and Dark Energy (DE) that is well motivated by fundamental physics. The axion and dilaton arise as pseudo-Goldstone modes of symmetries that predict particle masses depend on the dilaton, and therefore to evolve cosmologically, leading to correlated modifications of recombination physics, the sound horizon, and late-time expansion and growth histories. We confront this model with Planck 2018 temperature, polarisation, and lensing data, SPT-3G high-$\ell$ measurements, DESI DR2 BAO, and Pantheon$+$ supernovae, assuming that the axion makes up all of the dark matter and that the dilaton plays the role of a dark energy field. We find that it fits the data somewhat better than $Λ$CDM cosmology, with the $χ^2$ lowered by $\simeq 7$ for three additional parameters, and significantly raises the inferred Hubble constant to $H_0 \simeq 69.2\,\mathrm{km\,s^{-1}\,Mpc^{-1}}$, reducing the Hubble tension to $\lesssim 3σ$ and thereby allowing a joint fit of CMB and SH0ES data. The model fits this enlarged data set as well as the $w_0w_a$ model with an electron mass modified by hand at recombination, though it does so with calculable dynamics. Axio-dilaton self-interactions robustly fake a phantom equation of state in DESI measurements. There is a sting: cosmology prefers dilaton-matter couplings $|\mathbf{g}|\sim 10^{-2}$-$10^{-1}$, which are large enough to have been detected in solar-system tests of General Relativity. These results show how axio-dilatons can provide a viable cosmology preferred by current data at surprisingly large couplings, within a framework that links dark energy, dark matter, and time-dependent particle masses in a coherent way. They suggest both new observable signals and new theoretical directions, aimed at resolving the apparent inconsistency with non-cosmological observations.

astro-ph.CO↗