arXiv subjects

Yash Kanoria

Publications and source records attributed to Yash Kanoria.

17 recordsLinked to original sources

arXiv2026

The Generalized Excursion Coupling as the Limit of Concave Optimal Transport

The Monge-Kantorovich problem with the Euclidean distance cost is degenerate, typically admitting infinitely many optimal plans. A unique optimal plan is selected by perturbing the distance to a strictly convex or increasing strictly concave cost and passing to the limit: the convex side gives the plan induced by the map which is monotone on each transport ray, while the concave side was understood only on the real line, where Juillet proved that power-cost optimizers converge to the so-called excursion coupling, which is induced by a map when the source is atomless. This paper extends the concave selection to $\mathbb{R}^n$ for mutually singular finite positive Borel measures $μ$ and $ν$ with equal total mass and finite first moments, assuming that $μ$ is absolutely continuous with respect to the Lebesgue measure: for a broad class of increasing strictly concave perturbations of the distance with a well-defined first-order profile, the corresponding optimal maps converge in $μ$-measure to the same intrinsic limit $t_\#$, independent of the perturbation family and its profile. The map $t_\#$ is obtained by disintegrating the transport along its maximal rays and applying the one-dimensional excursion coupling on each ray. The same limit is established for the power costs $\|x-y\|^{1-\varepsilon}$ under a finite $\|x\|\ln(1+\|x\|)$ moment condition, establishing a conjecture of Juillet.

math.OC↗

arXiv2026

Impact of Rankings and Personalized Recommendations in Marketplaces

Decision-making often requires individuals to navigate large sets of options with incomplete information and imperfectly formed preferences. Information provisioning tools, such as public rankings and personalized recommendations, have become central to guiding these choices, yet their welfare implications across different market environments remain poorly understood. This paper studies a stylized large-market model to quantify the aggregate value of these tools under uncapacitated supply and capacitated supply. Agent utility is a weighted combination of a common term (population-level quality) and an idiosyncratic term (individual-specific fit). Agents observe noisy signals of these components: public rankings improve estimates of common quality, while personalized recommendations additionally reveal idiosyncratic fit. In uncapacitated settings, both tools improve welfare through better selection. Their relative value is governed by preference heterogeneity: rankings are highly effective when preferences are relatively homogeneous, whereas the marginal value of personalization grows as heterogeneity increases. In stark contrast, under capacity constraints, a strict conservation logic applies: learning common quality primarily reshuffles assignments without increasing aggregate average agent welfare. However, revealing idiosyncratic fit generates substantial welfare gains by mitigating congestion on overdemanded items and unlocking match-specific value. We demonstrate that these core insights are robust across model extensions including alternative extreme-value tail distributions and heterogeneous supply-side priorities. Finally, introducing strategic supply-side pricing reveals a stark distributional dichotomy: while information expands total surplus, the newly generated value is ultimately extracted by the short side of the market.

econ.TH↗

arXiv2026

Policy Optimization in Hybrid Discrete-Continuous Action Spaces via Mixed Gradients

We study reinforcement learning in hybrid discrete-continuous action spaces, such as settings where the discrete component selects a regime (or index) and the continuous component optimizes within it -- a structure common in robotics, control, and operations problems. Standard model-free policy gradient methods rely on score-function (SF) estimators and suffer from severe credit-assignment issues in high-dimensional settings, leading to poor gradient quality. On the other hand, differentiable simulation largely sidesteps these issues by backpropagating through a simulator, but the presence of discrete actions or non-smooth dynamics yields biased or uninformative gradients. To address this, we propose Hybrid Policy Optimization (HPO), which backpropagates through the simulator wherever smoothness permits, using a mixed gradient estimator that combines pathwise and SF gradients while maintaining unbiasedness. We also show how problems with action discontinuities can be reformulated in hybrid form, further broadening its applicability. Empirically, HPO substantially outperforms PPO on inventory control and switched linear-quadratic regulator problems, with performance gaps increasing as the continuous action dimension grows. Finally, we characterize the structure of the mixed gradient, showing that its cross term -- which captures how continuous actions influence future discrete decisions -- becomes negligible near a discrete best response, thereby enabling approximate decentralized updates of the continuous and discrete components and reducing variance near optimality. All resources are available at github.com/MatiasAlvo/hybrid-rl.

cs.LG↗

arXiv2026

Right-Sizing Communication and Recommendation Set Size in AI-Assisted Search

We model the interaction between a user and an AI driven recommendation system. The user initiates the process by conveying preference information through a costly and noisy message. The AI assistant, acting as a Bayesian agent, interprets the user's message to form a posterior belief about their true preferences and make product recommendations. In particular, it determines how many recommendations to present so as to maximize the user's expected utility from their final choice, while accounting for the search cost induced by the size of the recommendation set. We use mutual information based cost functions to model the two distinct costs incurred by the user during the interaction: (i) a communication cost, which increases with the precision of their preference message, and (ii) a search cost, which increases with the size of the recommendation set provided by the AI assistant. We study products and preferences which live in d dimensional space, and ask how the user's expected payoff can be maximized. For large d, we characterize how optimal message precision and recommendation set size depend on the cost parameters, under two distinct distributions from which recommendations can be sampled from the product universe: (i) Bayes' posterior belief, and (ii) an optimized tilted distribution. Under the posterior sampling scheme (i), we identify a hybrid regime, in which an efficient interaction policy requires jointly optimizing the amount of information (in bits) conveyed by the user and the number of recommendations provided by the AI assistant. In the tilted sampling scheme (ii), our results show that the optimal interaction policy uses only one of communication and search, favoring whichever of them is less costly.

cs.AI↗

arXiv2025

What Is Your AI Agent Buying? Evaluation, Biases, Model Dependence, & Emerging Implications for Agentic E-Commerce

Online marketplaces will be transformed by autonomous AI agents acting on behalf of consumers. Rather than humans browsing and clicking, AI agents can parse webpages or leverage APIs to view, evaluate and choose products. We investigate the behavior of AI agents using ACES, a provider-agnostic framework for auditing agent decision-making. We reveal that agents can exhibit choice homogeneity, often concentrating demand on a few ``modal'' products while ignoring others entirely. Yet, these preferences are unstable: model updates can drastically reshuffle market shares. Furthermore, randomized trials show that while agents have improved over time on simple tasks with a clearly identified best choice, they exhibit strong position biases -- varying across providers and model versions, and persisting even in text-only "headless" interfaces -- undermining any universal notion of a ``top'' rank. Agents also consistently penalize sponsored tags while rewarding platform endorsements, and sensitivities to price, ratings, and reviews vary sharply across models. Finally, we demonstrate that sellers can respond: a seller-side agent making simple, query-conditional description tweaks can drive significant gains in market share. These findings reveal that agentic markets are volatile and fundamentally different from human-centric commerce, highlighting the need for continuous auditing and raising questions for platform design, seller strategy and regulation.

cs.AI↗

arXiv2025

Deep Reinforcement Learning for Inventory Networks: Toward Reliable Policy Optimization

We argue that inventory management presents unique opportunities for the reliable application of deep reinforcement learning (DRL). To enable this, we emphasize and test two complementary techniques. The first is Hindsight Differentiable Policy Optimization (HDPO), which uses pathwise gradients from offline counterfactual simulations to directly and efficiently optimize policy performance. Unlike standard policy gradient methods that rely on high-variance score-function estimators, HDPO computes gradients by differentiating through the known system dynamics. Via extensive benchmarking, we show that HDPO recovers near-optimal policies in settings with known or bounded optima, is more robust than variants of the REINFORCE algorithm, and significantly outperforms generalized newsvendor heuristics on problems using real time series data. Our second technique aligns neural policy architectures with the topology of the inventory network. We exploit Graph Neural Networks (GNNs) as a natural inductive bias for encoding supply chain structure, demonstrate that they can represent optimal and near-optimal policies in two theoretical settings, and empirically show that they reduce data requirements across six diverse inventory problems. A key obstacle to progress in this area is the lack of standardized benchmark problems. To address this gap, we open-source a suite of benchmark environments, along with our full codebase, to promote transparency and reproducibility. All resources are available at github.com/MatiasAlvo/Neural_inventory_control.

cs.LG↗

arXiv2024

The Fault in Our Recommendations: On the Perils of Optimizing the Measurable

Recommendation systems are widespread, and through customized recommendations, promise to match users with options they will like. To that end, data on engagement is collected and used. Most recommendation systems are ranking-based, where they rank and recommend items based on their predicted engagement. However, the engagement signals are often only a crude proxy for utility, as data on the latter is rarely collected or available. This paper explores the following question: By optimizing for measurable proxies, are recommendation systems at risk of significantly under-delivering on utility? If so, how can one improve utility which is seldom measured? To study these questions, we introduce a model of repeated user consumption in which, at each interaction, users select between an outside option and the best option from a recommendation set. Our model accounts for user heterogeneity, with the majority preferring ``popular'' content, and a minority favoring ``niche'' content. The system initially lacks knowledge of individual user preferences but can learn them through observations of users' choices over time. Our theoretical and numerical analysis demonstrate that optimizing for engagement can lead to significant utility losses. Instead, we propose a utility-aware policy that initially recommends a mix of popular and niche content. As the platform becomes more forward-looking, our utility-aware policy achieves the best of both worlds: near-optimal utility and near-optimal engagement simultaneously. Our study elucidates an important feature of recommendation systems; given the ability to suggest multiple items, one can perform significant exploration without incurring significant reductions in engagement. By recommending high-risk, high-reward items alongside popular items, systems can enhance discovery of high utility items without significantly affecting engagement.

cs.IR↗

arXiv2023

Dynamic Spatial Matching

Motivated by a variety of online matching platforms, we consider demand and supply units which are located i.i.d. in [0,1]^d, and each demand unit needs to be matched with a supply unit. The goal is to minimize the expected average distance between matched pairs (the "cost"). We model dynamic arrivals of one or both of demand and supply with uncertain locations of future arrivals, and characterize the scaling behavior of the achievable cost in terms of system size (number of supply units), as a function of the dimension d. Our achievability results are backed by concrete matching algorithms. Across cases, we find that the platform can achieve cost (nearly) as low as that achievable if the locations of future arrivals had been known beforehand. Furthermore, in all cases except one, cost nearly as low in terms of scaling as the expected distance to the nearest neighboring supply unit is achievable, i.e., the matching constraint does not cause an increase in cost either. The aberrant case is where only demand arrivals are dynamic, and d=1; excess supply significantly reduces cost in this case.

math.PR↗

arXiv2023

Dynamic Resource Allocation: Algorithmic Design Principles and Spectrum of Achievable Performances

Dynamic resource allocation problems are ubiquitous, arising in inventory management, order fulfillment, online advertising, and other applications. We initially focus on one of the simplest models of online resource allocation: the multisecretary problem. In the multisecretary problem, a decision maker sequentially hires up to $B$ out of $T$ candidates, and candidate ability values are drawn i.i.d. from a distribution $F$ on $[0,1]$. First, we investigate fundamental limits on performance as a function of the value distribution under consideration. We quantify performance in terms of regret, defined as the additive loss relative to the best performance achievable in hindsight. We present a novel fundamental regret lower bound scaling of $Ω(T^{1/2 - 1/2(1 + β)})$ for distributions with gaps in their support, with $β$ quantifying the mass accumulation of types (values) around these gaps. This lower bound contrasts with the constant and logarithmic regret guarantees shown to be achievable in prior work, under specific assumptions on the value distribution. Second, we introduce a novel algorithmic principle, Conservativeness with respect to Gaps (CwG), which yields near-optimal performance with regret scaling of $\tilde{O}(T^{1/2 - 1/2(1 + β)})$ for any distribution in a class parameterized by the mass accumulation parameter $β$. We then turn to operationalizing the CwG principle across dynamic resource allocation problems. We study a general and practical algorithm, Repeatedly Act using Multiple Simulations (RAMS), which simulates possible futures to estimate a hindsight-based approximation of the value-to-go function. We establish that this algorithm inherits theoretical performance guarantees of algorithms tailored to the distribution of resource requests, including our CwG-based algorithm, and find that it outperforms them in numerical experiments.

math.PR↗

arXiv2023

The Competition for Partners in Matching Markets

We study the competition for partners in two-sided matching markets with heterogeneous agent preferences, with a focus on how the equilibrium outcomes depend on the connectivity in the market. We model random partially connected markets, with each agent having an average degree $d$ in a random (undirected) graph, and a uniformly random preference ranking over their neighbors in the graph. We formally characterize stable matchings in large markets random with small imbalance and find a threshold in the connectivity $d$ at $\log^2 n$ (where $n$ is the number of agents on one side of the market) which separates a ``weak competition'' regime, where agents on both sides of the market do equally well, from a ``strong competition'' regime, where agents on the short (long) side of the market enjoy a significant advantage (disadvantage). Numerical simulations confirm and sharpen our theoretical predictions, and demonstrate robustness to our assumptions. We leverage our characterizations in two ways: First, we derive prescriptive insights into how to design the connectivity of the market to trade off optimally between the average agent welfare achieved and the number of agents who remain unmatched in the market. For most market primitives, we find that the optimal connectivity should lie in the weak competition regime or at the threshold between the regimes. Second, our analysis uncovers a new conceptual principle governing whether the short side enjoys a significant advantage in a given matching market, which can moreover be applied as a diagnostic tool given only basic summary statistics for the market. Counterfactual analyses using data on centralized high school admissions in a major USA city show the practical value of both our design insights and our diagnostic principle.

cs.GT↗

arXiv2022

Large Deviations Optimal Scheduling of Closed Queueing Networks

We study the design of dynamic scheduling controls in closed queueing networks with a fixed number of jobs. Each time a server becomes available, the controller has (limited) flexibility in choosing the buffer from which to serve a job. If no jobs are available at any compatible buffer, the server idles. If the job is served, it relocates to a ``destination'' buffer. We study how to maximize throughput in steady state via a large deviations analysis. We propose a family of simple state-dependent policies called Scaled MaxWeight (SMW) policies that dynamically manage the distribution of jobs in the network. We prove that under a complete resource pooling condition (analogous to the condition in Hall's marriage theorem), any SMW policy leads to exponential decay of throughput-loss probability as the number of jobs scales to infinity. Further, there is an SMW policy that achieves the optimal loss exponent among all scheduling policies, and we analytically specify this policy in terms of the service rates and routing probabilities. The optimal SMW policy maintains high job levels adjacent to structurally under-supplied servers. Our methodology also applies to the open network setting and leads to exponent-optimal policies.

math.PR↗

arXiv2022

Decentralized Online Convex Optimization in Networked Systems

We study the problem of networked online convex optimization, where each agent individually decides on an action at every time step and agents cooperatively seek to minimize the total global cost over a finite horizon. The global cost is made up of three types of local costs: convex node costs, temporal interaction costs, and spatial interaction costs. In deciding their individual action at each time, an agent has access to predictions of local cost functions for the next $k$ time steps in an $r$-hop neighborhood. Our work proposes a novel online algorithm, Localized Predictive Control (LPC), which generalizes predictive control to multi-agent systems. We show that LPC achieves a competitive ratio of $1 + \tilde{O}(ρ_T^k) + \tilde{O}(ρ_S^r)$ in an adversarial setting, where $ρ_T$ and $ρ_S$ are constants in $(0, 1)$ that increase with the relative strength of temporal and spatial interaction costs, respectively. This is the first competitive ratio bound on decentralized predictive control for networked online convex optimization. Further, we show that the dependence on $k$ and $r$ in our results is near optimal by lower bounding the competitive ratio of any decentralized online algorithm.

math.OC↗

arXiv2022

Blind Dynamic Resource Allocation in Closed Networks via Mirror Backpressure

We study the problem of maximizing payoff generated over a period of time in a general class of closed queueing networks with a finite, fixed number of supply units which circulate in the system. Demand arrives stochastically, and serving a demand unit (customer) causes a supply unit to relocate from the ``origin'' to the ``destination'' of the customer. The key challenge is to manage the distribution of supply in the network. We consider general controls including customer entry control, pricing, and assignment. Motivating applications include shared transportation platforms and scrip systems. Inspired by the mirror descent algorithm for optimization and the backpressure policy for network control, we introduce a rich family of \emph{Mirror Backpressure} (MBP) control policies. The MBP policies are simple and practical, and crucially do not need any statistical knowledge of the demand (customer) arrival rates (these rates are permitted to vary in time). Under mild conditions, we propose MBP policies that are provably near optimal. Specifically, our policies lose at most $O(\frac{K}{T}+\frac{1}{K} + \sqrt{ηK})$ payoff per customer relative to the optimal policy that knows the demand arrival rates, where $K$ is the number of supply units, $T$ is the total number of customers over the time horizon, and $η$ is the demand process' average rate of change per customer arrival. An adaptation of MBP is found to perform well in numerical experiments based on data from ride-hailing.

math.OC↗

arXiv2020

Matching while Learning

We consider the problem faced by a service platform that needs to match limited supply with demand but also to learn the attributes of new users in order to match them better in the future. We introduce a benchmark model with heterogeneous "workers" (demand) and a limited supply of "jobs" that arrive over time. Job types are known to the platform, but worker types are unknown and must be learned by observing match outcomes. Workers depart after performing a certain number of jobs. The expected payoff from a match depends on the pair of types and the goal is to maximize the steady-state rate of accumulation of payoff. Though we use terminology inspired by labor markets, our framework applies more broadly to platforms where a limited supply of heterogeneous products is matched to users over time. Our main contribution is a complete characterization of the structure of the optimal policy in the limit that each worker performs many jobs. The platform faces a trade-off for each worker between myopically maximizing payoffs (exploitation) and learning the type of the worker (exploration). This creates a multitude of multi-armed bandit problems, one for each worker, coupled together by the constraint on availability of jobs of different types (capacity constraints). We find that the platform should estimate a shadow price for each job type, and use the payoffs adjusted by these prices, first, to determine its learning goals and then, for each worker, (i) to balance learning with payoffs during the "exploration phase," and (ii) to myopically match after it has achieved its learning goals during the "exploitation phase."

cs.LG↗

arXiv2020

Dynamic Reserve Prices for Repeated Auctions: Learning from Bids

A large fraction of online advertisement is sold via repeated second price auctions. In these auctions, the reserve price is the main tool for the auctioneer to boost revenues. In this work, we investigate the following question: Can changing the reserve prices based on the previous bids improve the revenue of the auction, taking into account the long-term incentives and strategic behavior of the bidders? We show that if the distribution of the valuations is known and satisfies the standard regularity assumptions, then the optimal mechanism has a constant reserve. However, when there is uncertainty in the distribution of the valuations, previous bids can be used to learn the distribution of the valuations and to update the reserve price. We present a simple, approximately incentive-compatible, and asymptotically optimal dynamic reserve mechanism that can significantly improve the revenue over the best static reserve. The paper is from July 2014 (our submission to WINE 2014), posted later here on the arxiv to complement the 1-page abstract in the WINE 2014 proceedings.

cs.GT↗

arXiv2015

The set of solutions of random XORSAT formulae

The XOR-satisfiability (XORSAT) problem requires finding an assignment of $n$ Boolean variables that satisfy $m$ exclusive OR (XOR) clauses, whereby each clause constrains a subset of the variables. We consider random XORSAT instances, drawn uniformly at random from the ensemble of formulae containing $n$ variables and $m$ clauses of size $k$. This model presents several structural similarities to other ensembles of constraint satisfaction problems, such as $k$-satisfiability ($k$-SAT), hypergraph bicoloring and graph coloring. For many of these ensembles, as the number of constraints per variable grows, the set of solutions shatters into an exponential number of well-separated components. This phenomenon appears to be related to the difficulty of solving random instances of such problems. We prove a complete characterization of this clustering phase transition for random $k$-XORSAT. In particular, we prove that the clustering threshold is sharp and determine its exact location. We prove that the set of solutions has large conductance below this threshold and that each of the clusters has large conductance above the same threshold. Our proof constructs a very sparse basis for the set of solutions (or the subset within a cluster). This construction is intimately tied to the construction of specific subgraphs of the hypergraph associated with an instance of $k$-XORSAT. In order to study such subgraphs, we establish novel local weak convergence results for them.

cs.DM↗

arXiv2014

The size of the core in assignment markets

Assignment markets involve matching with transfers, as in labor markets and housing markets. We consider a two-sided assignment market with agent types and stochastic structure similar to models used in empirical studies, and characterize the size of the core in such markets. Each agent has a randomly drawn productivity with respect to each type of agent on the other side. The value generated from a match between a pair of agents is the sum of the two productivity terms, each of which depends only on the type but not the identity of one of the agents, and a third deterministic term driven by the pair of types. We allow the number of agents to grow, keeping the number of agent types fixed. Let $n$ be the number of agents and $K$ be the number of types on the side of the market with more types. We find, under reasonable assumptions, that the relative variation in utility per agent over core outcomes is bounded as $O^*(1/n^{1/K})$, where polylogarithmic factors have been suppressed. Further, we show that this bound is tight in worst case. We also provide a tighter bound under more restrictive assumptions. Our results provide partial justification for the typical assumption of a unique core outcome in empirical studies.

cs.GT↗