We are interested in developing mathematical models for understanding and improving complex systems. Some broad areas of interest are detailed below.

Learning, attribution and incentives in Operations

Here are some questions that are currently occupying us: How should information be collected and exploited in an online decision making? How does one use curated choices to both learn consumer preferences and optimally satisfy consumer demand? How does one handle very high dimensional product and consumer features? What if consumers are impatient and leave if they are offered satisfiable set of products while the algorithm learns? What if the “utility” from a product is multidimensional, e.g. a drug has both efficacy and toxicity? What happens when the customer arrival is adversarial? How does one attribute the overall value generated from a collection of actions to each individual action? What are the incentives that ensure that participants in an industry adopt a blockchain?

  • V. Goyal, G. Iyengar, and R. Udwani. Online Allocation of Reusable Resources via Algorithms Guided by Fluid Approximations. 2020. arXiv.

  • M-H. Oh, G. Iyengar, and A. Zeevi. Sparsity-Agnostic Lasso Bandit. 2020. arXiv.

  • G. Iyengar, F. Saleh, J. Sethuraman, and W. Wang. Economics of permissioned blockchain adoption. 2020. SSRN.

  • R. Singal, O. Besbes, A. Desir, V. Goyal and G. Iyengar. Shapley Meets Uniform: An axiomatic framework for attribution in online advertising. Under revision in Management Science. SSRN. Preliminary version appeared in WWW 2019.

  • C. Kocyigit, G. Iyengar, D. Kuhn, and W. Wiesemann. Distributionally robust mechanism design. Management Science, 66(1):159–189, 2020. SSRN.

  • O. Toubia, G. Iyengar, R. Bunnell, and A. Lemaire. Extracting features of entertainment products: A guided Latent Dirichlet Allocation approach informed by the psychology of media consumption. Journal of Marketing Research, 56(1):18–36, 2019. PDF.

  • M-H. Oh and G. Iyengar. Thompson sampling for multinomial logit contextual bandits. NeurIPS 2019.

  • M-H. Oh and G. Iyengar. Sequential anomaly detection using inverse reinforcement learning. KDD 2019.

  • H. Alsabah, B. Bernard, A. Capponi, G. Iyengar, and J. Sethuraman. Multiregional oligopoly with capacity constraints. Under review in Management Science. SSRN.

Large Scale Optimization

There is a revolution underway in optimization methods for a variety of reasons: faster CPUs, a large amount of very granular data about applications, advent of GPUs with their ability to do simple operations very fast, new methodologies, deep learning and related machine learning tasks. We are currently working on developing scalable optimization algorithms for inventory management with capacity constraints, and multiperiod multi-asset portfolio optimization with taxes.

  • R. Mazumder, A. Choudhury, G. Iyengar, and B. Sen. A computational framework for multivariate convex regression and its variants. Journal of the American Statistical Association, 114(525):318–331, 2019. arXiv.

  • R. A. Carrasco, G. Iyengar, and C. Stein. Resource cost aware scheduling. EJOR, 269(2):621–632, 2018. IDEAS.

  • A. Federgruen, C. D. Guetta, and G. Iyengar. Two-echelon distribution systems with random demands and storage constraints. Naval Research Logistics (NRL), 65(8):594–618, 2018. PDF.

  • Y. Chen, R. Iyengar, and G. Iyengar. Modeling multimodal continuous heterogeneity in conjoint analysis: a sparse learning approach. Marketing Science, 36(1):140–156, 2017. Wharton Mkting Papers.

  • M. Haugh, G. Iyengar, and I. Song. A generalized risk budgeting approach to portfolio construction. Computational Finance, 21(2):29–60, 01 2017. SSRN.

  • M. Haugh, G. Iyengar, and C. Wang. Tax-aware dynamic asset allocation. Operations Research, 64(4):849–866, 2016. PDF.

Information in Biological Systems

Information is key for survival both at the level of a single cell as well as that of an organism. We are pursuing several research projects where the goal is to understand information acquisition and representation. With collaborators at the Simons Center for Living Machines at the National Center for Biological Sciences, Bangalore, we are investigating the role of glycans (sugar coated proteins) in cell identity. We are attempting to understand mechanisms, in particular the role of time, that allow the immune system to distinguish self vs non-self antigens, and the role of the gut microbiota.

  • A. Yadav, Q. Vagne, P. Sens, G. Iyengar, and M. Rao. Glycan processing in the Golgi – optimal information coding and constraints on cisternal number and enzyme specificity. 2020. bioRxiv.

  • S. G. Das, M. Rao, and G. Iyengar. Cascade of transitions in molecular information theory. Journal of Statistical Mechanics: Theory and Experiment, 2018(9):093402, 2018. arXiv.

  • S. G. Das, M. Rao, and G. Iyengar. Universal lower bound on the free-energy cost of molecular measurements. Physical Review E, 95(6):062410, 2017. arXiv.

  • G. Iyengar and M. Rao. A cellular solution to an information-processing problem. PNAS, 111(34):12402–12407, 2014. PNAS.