-
Smoothness Meets Autobidding: Tight Price of Anarchy Bounds for Simultaneous First-Price Auctions
Authors:
Riccardo Colini-Baldeschi,
Sophie Klumper,
Twan Kroll,
Stefano Leonardi,
Guido Schäfer,
Artem Tsikiridis
Abstract:
Online advertising systems have recently transitioned to autobidding, enabling advertisers to delegate bidding decisions to automated agents. Each advertiser directs their agent to optimize a valuation-dependent objective subject to return-on-investment (ROI) or budget constraints. Given their relevance, there has been a surge in literature studying the liquid welfare price of anarchy (POA) of cor…
▽ More
Online advertising systems have recently transitioned to autobidding, enabling advertisers to delegate bidding decisions to automated agents. Each advertiser directs their agent to optimize a valuation-dependent objective subject to return-on-investment (ROI) or budget constraints. Given their relevance, there has been a surge in literature studying the liquid welfare price of anarchy (POA) of core auction formats in autobidding, among which simultaneous first-price auctions (FPA). These models capture a large range of heterogeneous agent behaviors, requiring advanced proofs to derive tight POA bounds. Recently, Deng et al. (NeurIPS 2024) showed that the POA of FPA for mixed autobidders (i.e., value and utility maximizers) under ROI is 2.18 for additive valuations.
We extend the smoothness framework of Syrgkanis and Tardos (STOC 2013) to autobidding. A key contribution is a technique to balance smoothness parameters across heterogeneous agent types. Finding the best POA bound reduces to solving a POA-revealing mathematical program. Our approach has three strengths: (1) Simplicity: We prove smoothness for single-item FPA. Results for simultaneous FPA follow via our theorem. For example, by showing smoothness for value and utility maximizers, we obtain the tight POA of 2.18 for mixed autobidding. (2) Extendibility: Our Extension Theorem adapts to simultaneous FPA with reserve prices and agents with fractionally subadditive valuations and heterogeneous payment sensitivities and target ROI parameters. We establish the first (mostly) tight POA bounds for several models beyond the autobidding state of the art. (3) Generality: Our framework bounds the POA of coarse correlated equilibria (CCE), which arise when hybrid agents employ regret-minimizing algorithms. Building on Kolumbus and Nisan (WWW 2022), we show that CCE from such agents have properties that keep their POA low.
△ Less
Submitted 25 June, 2025;
originally announced June 2025.
-
To Trust or Not to Trust: Assignment Mechanisms with Predictions in the Private Graph Model
Authors:
Riccardo Colini-Baldeschi,
Sophie Klumper,
Guido Schäfer,
Artem Tsikiridis
Abstract:
The realm of algorithms with predictions has led to the development of several new algorithms that leverage (potentially erroneous) predictions to enhance their performance guarantees. The challenge is to devise algorithms that achieve optimal approximation guarantees as the prediction quality varies from perfect (consistency) to imperfect (robustness). This framework is particularly appealing in…
▽ More
The realm of algorithms with predictions has led to the development of several new algorithms that leverage (potentially erroneous) predictions to enhance their performance guarantees. The challenge is to devise algorithms that achieve optimal approximation guarantees as the prediction quality varies from perfect (consistency) to imperfect (robustness). This framework is particularly appealing in mechanism design contexts, where predictions might convey private information about the agents. In this paper, we design strategyproof mechanisms that leverage predictions to achieve improved approximation guarantees for several variants of the Generalized Assignment Problem (GAP) in the private graph model. In this model, first introduced by Dughmi & Ghosh (2010), the set of resources that an agent is compatible with is private information. For the Bipartite Matching Problem (BMP), we give a deterministic group-strategyproof (GSP) mechanism that is $(1 +1/γ)$-consistent and $(1 + γ)$-robust, where $γ\ge 1$ is some confidence parameter. We also prove that this is best possible. Remarkably, our mechanism draws inspiration from the renowned Gale-Shapley algorithm, incorporating predictions as a crucial element. Additionally, we give a randomized mechanism that is universally GSP and improves on the guarantees in expectation. The other GAP variants that we consider all make use of a unified greedy mechanism that adds edges to the assignment according to a specific order. Our universally GSP mechanism randomizes over the greedy mechanism, our mechanism for BMP and the predicted assignment, leading to $(1+3/γ)$-consistency and $(3+γ)$-robustness in expectation. All our mechanisms also provide more fine-grained approximation guarantees that interpolate between the consistency and the robustness, depending on some natural error measure of the prediction.
△ Less
Submitted 6 March, 2024;
originally announced March 2024.
-
Partial Allocations in Budget-Feasible Mechanism Design: Bridging Multiple Levels of Service and Divisible Agents
Authors:
Georgios Amanatidis,
Sophie Klumper,
Evangelos Markakis,
Guido Schäfer,
Artem Tsikiridis
Abstract:
Budget-feasible procurement has been a major paradigm in mechanism design since its introduction by Singer (2010). An auctioneer (buyer) with a strict budget constraint is interested in buying goods or services from a group of strategic agents (sellers). In many scenarios it makes sense to allow the auctioneer to only partially buy what an agent offers, e.g., an agent might have multiple copies of…
▽ More
Budget-feasible procurement has been a major paradigm in mechanism design since its introduction by Singer (2010). An auctioneer (buyer) with a strict budget constraint is interested in buying goods or services from a group of strategic agents (sellers). In many scenarios it makes sense to allow the auctioneer to only partially buy what an agent offers, e.g., an agent might have multiple copies of an item to sell, they might offer multiple levels of a service, or they may be available to perform a task for any fraction of a specified time interval. Nevertheless, the focus of the related literature has been on settings where each agent's services are either fully acquired or not at all. The main reason for this, is that in settings with partial allocations like the ones mentioned, there are strong inapproximability results. Under the mild assumption of being able to afford each agent entirely, we are able to circumvent such results in this work. We design a polynomial-time, deterministic, truthful, budget-feasible $(2+\sqrt{3})$-approximation mechanism for the setting where each agent offers multiple levels of service and the auctioneer has a discrete separable concave valuation function. We then use this result to design a deterministic, truthful and budget-feasible $O(1)$-approximation mechanism for the setting where any fraction of a service can be acquired and the auctioneer's valuation function is separable concave (i.e., the sum of concave functions). For the special case of a linear valuation function, we improve the best known approximation ratio for the problem from $1+φ$ (by Klumper & Schäfer (2022)) to $2$. This establishes a separation between this setting and its indivisible counterpart.
△ Less
Submitted 28 May, 2025; v1 submitted 14 July, 2023;
originally announced July 2023.
-
Budget Feasible Mechanisms for Procurement Auctions with Divisible Agents
Authors:
Sophie Klumper,
Guido Schäfer
Abstract:
We consider budget feasible mechanisms for procurement auctions with additive valuation functions. For the divisible case, where agents can be allocated fractionally, there exists an optimal mechanism with approximation guarantee $e/(e-1)$ under the small bidder assumption. We study the divisible case without the small bidder assumption, but assume that the true costs of the agents are bounded by…
▽ More
We consider budget feasible mechanisms for procurement auctions with additive valuation functions. For the divisible case, where agents can be allocated fractionally, there exists an optimal mechanism with approximation guarantee $e/(e-1)$ under the small bidder assumption. We study the divisible case without the small bidder assumption, but assume that the true costs of the agents are bounded by the budget. This setting lends itself to modeling economic situations in which the goods represent time and the agents' true costs are not necessarily small compared to the budget. Non-trivially, we give a mechanism with an approximation guarantee of 2.62, improving the result of 3 for the indivisible case. Additionally, we give a lower bound on the approximation guarantee of 1.25. We then study the problem in more competitive markets and assume that the agents' value over cost efficiencies are bounded by some $θ\ge 1$. For $θ\le 2$, we give a mechanism with an approximation guarantee of 2 and a lower bound of 1.18. Both results can be extended to settings with different agent types with a linear capped valuation function for each type. Finally, if each agent type has a concave valuation, we give a mechanism for which the approximation guarantee grows linearly with the number of agent types.
△ Less
Submitted 1 September, 2022; v1 submitted 27 June, 2022;
originally announced June 2022.
