In briefArtificial Currency and Separate Selling: A 3-Approximation for Consumer Surplus
SummaryThis paper studies how to allocate indivisible items to buyers whose values are drawn independently, aiming to maximize consumer surplus after subtracting burnt payments. The authors propose artificial-currency mechanisms, giving each buyer a personalized budget of currency that has no value outside the mechanism. The better of separate selling and these mechanisms achieves a 3-approximation to optimal consumer surplus, holding at every Bayes–Nash equilibrium without equilibrium selection. Both branches are shown to be necessary, since either one alone may capture only a vanishing share of the optimum.
In briefSecond-best trade achieves half of first-best efficiency in multidimensional markets
SummaryThe paper proves that at least half of the first-best gains from trade can always be achieved in multidimensional two-sided markets with additive values and costs and arbitrary downward-closed feasibility constraints. The guarantee is delivered by a mechanism that is Bayesian incentive compatible, interim individually rational, and strongly budget balanced. The result extends the known tight one-half bound for single-item bilateral trade, and the proof applies the Cai–Devanur–Weinberg duality framework.
WhyThe authors extend the tight one-half gains-from-trade bound to multidimensional markets with additive values and downward-closed constraints, via a new dual-based proof.
In briefExplicit restricted isometries with sparsity k = m^{2/3−o(1)}
SummaryThe paper constructs an explicit family of real m×n restricted isometry matrices with m→∞, m/n→0, δ_k→0, and k = m^{2/3−o(1)}. It combines Gershgorin's circle theorem with a flat RIP estimate to reduce the problem to selecting a large vertex set in a strongly regular graph whose small subsets span few edges. This vertex set is chosen using an incidence geometry result due to Zahl. The authors show that proof techniques using only entrywise absolute values of the Gram matrix face a barrier of m^{2/3}, not the conventionally assumed m^{1/2}, and that their construction meets this barrier.
WhyThe authors break the conventional m^{1/2} barrier of Gram-entrywise-absolute-value methods, reaching m^{2/3}, and give an explicit restricted isometry construction.
In briefNew linear codes over GF(7), GF(8), GF(9) via quasi-cyclic codes and Construction X
SummaryThe paper presents new linear codes over the finite fields GF(7), GF(8) and GF(9), including [88,8,64] and [90,8,65] over GF(7), [98,7,75] and [97,8,72] over GF(8), and [44,7,31] over GF(9). Each code's minimum distance exceeds the lower bound in Grassl's tables of best known linear codes by one. Four of the five are built from a quasi-cyclic code and a block-constant vector using Construction X with a repetition code. Parameters and weight counts were verified by exhaustive enumeration.
In briefInformation-theoretic study of how clustered regulatory T cells filter IL-2 interference
SummaryThis paper treats two neighboring activation sites in lymph nodes as a molecular communication system, studying how diffusible IL-2 couples their responses. Using a three-dimensional IL-2 field coupled to T cell population dynamics, the authors find that clustered regulatory T cells reduce neighboring-site interference more than dispersed ones while largely preserving strong local responses. Information-theoretic analysis shows clustering improves discrimination between weak and strong local sources at short separation.
Alexander Wietfeld, Kristina A. Ganzinger, Wolfgang Kellerer et al.
In briefAdversarial risk analysis with fully probabilistic designs
SummaryThis paper examines adversarial risk analysis (ARA), which supports decision-making against intelligent opponents and uncertain outcomes, and combines it with fully probabilistic design (FPD). FPD is a control-theory method that selects the decision distribution whose outcomes are closest to an ideal distribution. The authors use FPD to model and forecast opponents' decisions within ARA and show how to handle single- and multi-stage security games in this framework.
Mario Chac\'on-Falc\'on, Tatiana Valentine Guy, Miroslav K\'arn\'y et al.
In briefPerformance Portability Explained: Lessons from Speeding Up ICON Microphysics
SummaryThis paper examines how scientific computing codes can achieve reasonable performance on HPC hardware shaped by AI workloads without major engineering effort. The authors evaluate seven portability frameworks, spanning pragma-based, DSL, and data-centric approaches, on a microphysics kernel of the ICON climate model. Their benchmarks show that different HPC systems benefit from different combinations of optimizations. They therefore introduce a measure of framework expressivity to assess how easily each framework can customize portable implementations.
Faveo H\"orold, Georgiana Mania, Panagiotis Adamidis et al.
