Searcharxiv⌕ Search

arXiv subjects

Boxiao Chen

Publications and source records attributed to Boxiao Chen.

5 recordsLinked to original sources

Reducing Batch Makespan for LLM Agent Workloads through Inference-Environment Coordination

Efficient batch execution of agent tasks is critical to LLM training and development workflows, including benchmark evaluation, synthetic-data generation, and RL rollouts. Since these tasks alternate between LLM inference and environment interaction, we present coordinated optimizations to reduce batch makespan, the time required to complete all tasks in a batch. For hybrid LLMs, a workload-based resource model guides GPU-memory allocation between key-value (KV) cache and recurrent state and provides a reference sample concurrency to exploit parallelism while limiting recomputation. Trajectory-aware routing and scheduling promote cache reuse while balancing worker load. Under constrained environment capacity, we decouple sample admission from environment execution and introduce Chain-Advancing Action Scheduling (CAAS), which prioritizes long chains using observed progress to shorten the batch tail. Controlled replay combines actual inference with recorded environment responses and durations for ablations of individual optimizations. Live evaluations with Ling-3.0-Flash and Ling-3.1-Flash on the BFCL, SWE-bench Multilingual, WideSearch, and MCP-Atlas benchmarks demonstrate makespan reductions of 10.4% to 65.0%, while evaluation scores remain aligned. Furthermore, these methods have been successfully applied to training and evaluation workflows for the Ling-3.x model series, accelerating the model-development cycle.

cs.DC↗

Decentralized Multi-product Pricing: Diagonal Dominance, Nash Equilibrium, and Price of Anarchy

Decentralized pricing in multi-product firms and marketplaces can lose revenue because each decision maker ignores how its price affects the demand for other products. We quantify this loss through the price of anarchy of a game in which the price of each product is set to maximize that product's own revenue. Demand is linear, and cross-price effects may be substitutes or complements and need not be symmetric. Under a two-sided diagonal dominance condition with parameter $μ\in[0,1)$, which bounds the aggregate absolute cross-price effects in each row and column relative to the own-price effect, we show that the game has a unique Nash equilibrium and that equilibrium revenue is at least a fraction $4(1-μ)/(2-μ)^2$ of the centralized optimum. This bound was previously known for symmetric gross substitutes; we show that it remains valid for signed and asymmetric interactions, and that it is tight. Because asymmetry breaks the simultaneous diagonalization underlying existing analyses, we instead evaluate the revenue gap at the equilibrium itself. This yields an exact formula for the worst-case price of anarchy of each demand matrix in terms of a normalized interaction matrix, which can be much sharper than the scalar bound in heterogeneous interaction networks. The same argument extends the guarantee to jointly concave nonlinear demand under pointwise dominance of the demand Jacobian, for every pure-strategy equilibrium.

cs.GT↗

Adaptive Risk Mitigation in Demand Learning

We study dynamic pricing of a product with an unknown demand distribution over a finite horizon. Departing from the standard no-regret learning environment in which prices can be adjusted at any time, we restrict price changes to predetermined points in time to reflect common retail practice. This constraint, coupled with demand model ambiguity and an unknown customer arrival pattern, imposes a high risk of revenue loss, as a price based on a misestimated demand model may be applied to many customers before it can be revised. We develop an adaptive risk learning (ARL) framework that embeds a data-driven ambiguity set (DAS) to quantify demand model ambiguity by adapting to the unknown arrival pattern. Initially, when arrivals are few, the DAS includes a broad set of plausible demand models, reflecting high ambiguity and revenue risk. As new data is collected through pricing, the DAS progressively shrinks, capturing the reduction in model ambiguity and associated risk. We establish the probabilistic convergence of the DAS to the true demand model and derive a regret bound for the ARL policy that explicitly links revenue loss to the data required for the DAS to identify the true model with high probability. The dependence of our regret bound on the arrival pattern is unique to our constrained dynamic pricing problem and contrasts with no-regret learning environments, where regret is constant and arrival-pattern independent. Relaxing the constraint on infrequent price changes, we show that ARL attains the known constant regret bound. Numerical experiments further demonstrate that ARL outperforms benchmarks that prioritize either regret or risk alone by adaptively balancing both without knowledge of the arrival pattern. This adaptive risk adjustment is crucial for achieving high revenues and low downside risk when prices are sticky and both demand and arrival patterns are unknown.

cs.LG↗

Learning to Order for Inventory Systems with Lost Sales and Uncertain Supplies

We consider a stochastic lost-sales inventory control system with a lead time $L$ over a planning horizon $T$. Supply is uncertain, and is a function of the order quantity (due to random yield/capacity, etc). We aim to minimize the $T$-period cost, a problem that is known to be computationally intractable even under known distributions of demand and supply. In this paper, we assume that both the demand and supply distributions are unknown and develop a computationally efficient online learning algorithm. We show that our algorithm achieves a regret (i.e. the performance gap between the cost of our algorithm and that of an optimal policy over $T$ periods) of $O(L+\sqrt{T})$ when $L\geq\log(T)$. We do so by 1) showing our algorithm cost is higher by at most $O(L+\sqrt{T})$ for any $L\geq 0$ compared to an optimal constant-order policy under complete information (a well-known and widely-used algorithm) and 2) leveraging its known performance guarantee from the existing literature. To the best of our knowledge, a finite-sample $O(\sqrt{T})$ (and polynomial in $L$) regret bound when benchmarked against an optimal policy is not known before in the online inventory control literature. A key challenge in this learning problem is that both demand and supply data can be censored; hence only truncated values are observable. We circumvent this challenge by showing that the data generated under an order quantity $q^2$ allows us to simulate the performance of not only $q^2$ but also $q^1$ for all $q^1<q^2$, a key observation to obtain sufficient information even under data censoring. By establishing a high probability coupling argument, we are able to evaluate and compare the performance of different order policies at their steady state within a finite time horizon. Since the problem lacks convexity, we develop an active elimination method that adaptively rules out suboptimal solutions.

math.OC↗

Joint Collaboration and Compression Design for Distributed Sequential Estimation in a Wireless Sensor Network

In this work, we propose a joint collaboration-compression framework for sequential estimation of a random vector parameter in a resource constrained wireless sensor network (WSN). Specifically, we propose a framework where the local sensors first collaborate (via a collaboration matrix) with each other. Then a subset of sensors selected to communicate with the FC linearly compress their observations before transmission. We design near-optimal collaboration and linear compression strategies under power constraints via alternating minimization of the sequential minimum mean square error. We show that the objective function for collaboration design can be non-convex depending on the network topology. We reformulate and solve the collaboration design problem using quadratically constrained quadratic program (QCQP). Moreover, the compression design problem is also formulated as a QCQP. We propose two versions of compression design, one centralized where the compression strategies are derived at the FC and the other decentralized, where the local sensors compute their individual compression matrices independently. It is noted that the design of decentralized compression strategy is a non-convex problem. We obtain a near-optimal solution by using the bisection method. In contrast to the one-shot estimator, our proposed algorithm is capable of handling dynamic system parameters such as channel gains and energy constraints. Importantly, we show that the proposed methods can also be used for estimating time-varying random vector parameters. Finally, numerical results are provided to demonstrate the effectiveness of the proposed framework.

eess.SP↗