About

Itai Gurvich is a Professor at the Kellogg School of Management, Northwestern University. He earned his Ph.D. from Columbia University’s Graduate School of Business in 2008 and joined Kellogg the same year. From 2016 to 2020, he was on the faculty of Cornell University’s campus in New York City (Cornell Tech) before returning to Kellogg in 2021. His research focuses on the performance analysis and optimization of processing networks, as well as the theory of stochastic-process approximations. His work has been recognized with the INFORMS Applied Probability Society’s Best Publication Award. Gurvich has served as the Stochastic Models Area Editor for Operations Research and as Chair of the INFORMS Applied Probability Society. His academic positions include roles at Northwestern University and Cornell University, with a background in Decisions, Risk and Operations, and Operations Research. His teaching interests encompass Operations Management, Service Systems, Queueing Systems, and Applied Probability.

Research topics

• Computer Science
• Mathematics
• Mathematical optimization
• Statistics
• Mathematical economics
• Economics
• Operations research
• Microeconomics
• Applied mathematics
• Geometry
• Mathematical analysis

Selected publications

• Erratum to Online Allocation and Pricing: Constant Regret via Bellman Inequalities

  Operations Research · 2026-03-30

  articleSenior author

  Theorem 3 of the paper Online Allocation and Pricing: Constant Regret via Bellman Inequalities (Operations Research 69(3):821–840) states a constant regret result for a menu-pricing problem. In this erratum, the authors correct the theorem’s proof.

  PublisherDOI

• The Production of Service: A Workload View of Complementarity and Substitution

  SSRN Electronic Journal · 2025-01-01 · 1 citations

  preprintOpen accessSenior author
  PublisherDOI

• Technical Note—What’s in a Constraint? On the Ambiguity of Standard Delay Targets

  Operations Research · 2025-05-15

  articleSenior author

  Ambiguities in Average Speed of Answer Targets Staffing problems are often formulated as satisfization problems, in which the cost of servers is minimized subject to quality of service constraints. These constraints are intended to capture customers’ disutility from waiting or, at least, its structure. In “What’s in a constraint? On the ambiguity of standard delay targets,” Soh and Gurvich show that such targets—especially the popular average speed of answer—are ambiguous: they give rise to multiple optimal solutions (prioritization policies), each consistent with different assumptions about how customers value their time. By choosing, among all optimal solutions, the one that minimizes a weighted index of diversion (a generalization of variance for the multiclass queue), a service provider can ensure that its ASA-based staffing and prioritization decisions align with a convex model of customer delay disutility. Nonambiguity can also be enforced by restricting attention to fixed queue ratio priority policies.

  PublisherDOI

• Goggin's corrected Kalman Filter: Guarantees and Filtering Regimes

  ArXiv.org · 2025-02-19

  preprintOpen accessSenior author

  In this paper we revisit a non-linear filter for {\em non-Gaussian} noises that was introduced in [1]. Goggin proved that transforming the observations by the score function and then applying the Kalman Filter (KF) to the transformed observations results in an asymptotically optimal filter. In the current paper, we study the convergence rate of Goggin's filter in a pre-limit setting that allows us to study a range of signal-to-noise regimes which includes, as a special case, Goggin's setting. Our guarantees are explicit in the level of observation noise, and unlike most other works in filtering, we do not assume Gaussianity of the noises. Our proofs build on combining simple tools from two separate literature streams. One is a general posterior Cramér-Rao lower bound for filtering. The other is convergence-rate bounds in the Fisher information central limit theorem. Along the way, we also study filtering regimes for linear state-space models, characterizing clearly degenerate regimes -- where trivial filters are nearly optimal -- and a {\em balanced} regime, which is where Goggin's filter has the most value. \footnote{This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible.

  PublisherOA PDFDOI

• The Cost of Impatience in Dynamic Matching: Scaling Laws and Operating Regimes

  Management Science · 2024-07-19 · 1 citations

  articleSenior author

  We study matching queues with abandonment. The simplest of these is the two-sided queue with servers on one side and customers on the other, both arriving dynamically over time and abandoning if not matched by the time their patience elapses. We identify nonasymptotic and universal scaling laws for the matching loss due to abandonment, which we refer to as the “cost of impatience.” The scaling laws characterize the way in which this cost depends on the arrival rates and the (possibly different) mean patience of servers and customers. Our characterization reveals four operating regimes identified by an operational measure of patience that brings together mean patience and utilization. The four regimes subsume the regimes that arise in asymptotic (heavy-traffic) approximations. The scaling laws, specialized to each regime, reveal the fundamental structure of the cost of impatience and show that its order of magnitude is fully determined by (i) a “winner-take-all” competition between customer impatience and utilization, and (ii) the ability to accumulate inventory on the server side. Practically important is that when servers are impatient, the cost of impatience is, up to an order of magnitude, given by an insightful expression where only the minimum of the two patience rates appears. Considering the trade-off between abandonment and capacity costs, we characterize the scaling of the optimal safety capacity as a function of costs, arrival rates, and patience parameters. We prove that the ability to hold inventory of servers means that the optimal safety capacity grows logarithmically in abandonment cost and, in turn, slower than the square-root growth in the single-sided queue. This paper was accepted by Baris Ata, stochastic models and simulation. Funding: This work was supported by the National Science Foundation [Grant CMM-2137286]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.01513 .

  PublisherDOI

• Dynamic Allocation of Reusable Resources: Logarithmic Regret in Overloaded Networks

  Operations Research · 2024-07-05 · 5 citations

  article

  How to dynamically allocate limited capacity to service requests? The problem studied in this paper is common in service applications, such as hotels, car rentals, and consulting services. These applications have limited capacity that must be allocated among incoming service requests. Different requests may require resources for varying durations, and some requests might yield higher rewards than others when fulfilled. The decision maker, who controls this capacity, must decide upon each request’s arrival whether to accept it (thus committing resources for a certain period) or reject it in order to reserve capacity for potentially more valuable future requests. This paper expands our understanding of such resource allocation problems by developing an appealingly simple dynamic allocation algorithm with proven effectiveness. In an appropriate operating regime, the suboptimal gap of this algorithm is at most logarithmic in the maximum achievable reward. In other words, for large-scale systems, the suboptimality translates to a negligible percentage deviation from optimal performance. Paper’s Title: Dynamic Allocation of Reusable Resources: Logarithmic Regret in Overloaded Networks

  PublisherDOI

• Feature-Driven Priority Queuing

  SSRN Electronic Journal · 2024-01-01 · 2 citations

  preprintOpen accessSenior author
  PublisherDOI

• A Hierarchical Approach to Robust Stability of Multiclass Queueing Networks

  Operations Research · 2024-11-01

  article

  Robust Stability in Multiclass Queueing Networks: A New Approach In “A Hierarchical Approach to Robust Stability of Multiclass Queueing Networks,” F. Zhao, I. Gurvich, and J. Hasenbein introduce a framework to identify sufficient conditions under which a network’s stability is robust to the distributed choices of resources about their (local) prioritization of jobs. The framework produces sufficient conditions for such stability by relating it to robust optimization problems where the collection of priority policies plays the role of the uncertainty set. Interestingly, within the studied family of policies, robust stability for any policy is inherited from the stability of the special “corner” policies, which are none other than simple static-priority policies.

  PublisherDOI

• Dynamic Resource Allocation: The Geometry and Robustness of Constant Regret

  Mathematics of Operations Research · 2024 · 2 citations

  • Computer Science
  • Mathematics
  • Mathematical optimization

  We study a family of dynamic resource allocation problems, wherein requests of different types arrive over time and are accepted or rejected. Each request type is characterized by its reward, arrival probability, and resource consumption. An upper bound for the collected reward is given by a linear optimization problem with a random right-hand side. This type of problem, known as packing linear program (LP), is ubiquitous in resource allocation. We provide a detailed characterization of the parametric structure of this packing LP. Relying on this geometric understanding, we revisit and expand on BudgetRatio algorithms that achieve constant regret by resolving this same packing LP in each period and accepting requests scored as sufficiently valuable. We illustrate the benefits of the geometric view in proving that (i) BudgetRatio achieves constant regret relative to the offline (full information) upper bound in the presence of inventory that is (slowly) restocked, and (ii) within explicitly identifiable bounds, the algorithm’s regret is robust to misspecification of the model parameters. This gives bounds for the bandits version of the problem in which the parameters have to be learned. (iii) The algorithm has an equivalent formulation as a generalized bid-price algorithm in which the bid prices can be adaptively and efficiently computed. Our analysis focuses on the evolution of the remaining inventory—in turn of the LP that drives BudgetRatio—as a stochastic process. We prove that it is attracted to sticky regions of the state space in which the online algorithm takes actions consistent with the optimal basis of the offline upper bound, a basis that is revealed only in hindsight at the horizon’s end. Funding: This work was supported by the U.S. Department of Defense [Grant W911NF-20-C-0008].

  PublisherDOI

• Dynamic Matching: Characterizing and Achieving Constant Regret

  Management Science · 2023 · 22 citations

  Senior authorCorresponding
  • Computer Science
  • Computer Science
  • Mathematical optimization

  We study how to optimally match agents in a dynamic matching market with heterogeneous match cardinalities and values. A network topology determines the feasible matches in the market. In general, a fundamental tradeoff exists between short-term value—which calls for performing matches frequently—and long-term value—which calls, sometimes, for delaying match decisions in order to perform better matches. We find that in networks that satisfy a general position condition, the tension between short- and long-term value is limited, and a simple periodic clearing policy (nearly) maximizes the total match value simultaneously at all times. Central to our results is the general position gap ϵ; a proxy for capacity slack in the market. With the exception of trivial cases, no policy can achieve an all-time regret that is smaller, in terms of order, than [Formula: see text]. We achieve this lower bound with a policy, which periodically resolves a natural matching integer linear program, provided that the delay between resolving periods is of the order of [Formula: see text]. Examples illustrate the necessity of some delay to alleviate the tension between short- and long-term value. This paper was accepted by David Simchi-Levi, revenue management and market analytics. Funding: This work was supported by the National Science Foundation [Grant CMM-2010940] and the U.S. Department of Defense [Grant STTR A18B-T007].

  PublisherDOI

Recent grants

• Dynamic Matching Problems with Application to Kidney Allocation

  NSF · $517k · 2020–2021

• Taylor Expansion Approximations for Dynamic Programming Problems

  NSF · $350k · 2017–2020

Frequent coauthors

• Jan A. Van Mieghem

  30 shared

• R. Kannan Mutharasan

  Northwestern Medicine

  11 shared

• Clyde W. Yancy

  Northwestern University

  6 shared

• Nicholas D. Soulakis

  Northwestern University

  6 shared

• Amy R. Ward

  University of Chicago

  6 shared

• Eric Park

  Wake Forest University

  5 shared

• Junfei Huang

  5 shared

• Allen S. Anderson

  The University of Texas Health Science Center at San Antonio

  5 shared

Labs

• Itai GurvichPI

Awards & honors

• INFORMS Applied Probability Society’s Best Publication Award…
• INFORMS The Operations Research Society of Israel Prize for…
• POMS College of Healthcare Operations Management Best Paper…
• NU Excellence in Research, Northwestern University