Research
Dr. Spall’s research is focused on stochastic systems, statistical analysis, and computational algorithms in optimization and related areas. He has given many invited presentations at conferences, research labs, and universities and he has over 190 refereed publications, with the majority of the publications having Dr. Spall as the sole or principal author. Dr. Spall’s publications include multiple books and several highly cited papers in the IEEE or other journal literature, including the all-time most-cited paper in the Johns Hopkins APL Technical Digest (from over 1300 papers). His book, Introduction to Stochastic Search and Optimization (Wiley), is the most cited book in the general area of stochastic optimization (per Google Scholar). His work has been extensively cited, with over 90 percent of the citations coming from documents for which he is the sole or principal author. Methods and algorithms that Dr. Spall and colleagues have developed are in extensive use worldwide, including in applications areas such as resource management, quantum computing, defense and transportation systems, artificial intelligence, engineering reliability analysis, data science, and finance.
Research areas: Stochastic systems, parameter estimation, stochastic optimization, Monte Carlo methods and simulation, neural networks, dynamical systems and control, system identification and Kalman filtering, mathematical statistics, optimization theory, and uncertainty calculation
Web site for simultaneous perturbation stochastic approximation (SPSA) algorithm: www.jhuapl.edu/SPSA. While this SPSA site has not been updated since 2012, the material that is at the website remains fully valid and potentially useful. The references listed give a selection of early papers (2012 or earlier) devoted to methods and applications.
Books
Spall, J. C. (2003), Introduction to Stochastic Search and Optimization: Estimation, Simulation, and Control, Wiley, Hoboken, NJ (618 pages). (Book Web site: www.jhuapl.edu/ISSO)
Spall, J. C. (2024), “Solutions Manual for Introduction to Stochastic Search and Optimization: Estimation, Simulation, and Control ” (332 pages).
Spall, J. C. (editor and coauthor) (1988), Bayesian Analysis of Time Series and Dynamic Models, Marcel Dekker, New York (576 pages).
Partial list of current and recent research projects
Uncertainty calculation for state estimates in the extended Kalman filter for nonlinear/non-Gaussian processes (with Shihong Wei, JHU doctoral student)
High-dimensional stochastic optimization for neural networks and other areas (with Shiqing Sun, recent JHU doctoral student) .
Monte Carlo-based random sampling methods beyond Markov chain Monte Carlo (with Shiqing Sun, recent JHU doctoral student).
Evaluation of distributed and cyclic optimization in multi-agent control and tracking with incomplete information about the environment (with Jiahao Shi, recent JHU M.S. student and current doctoral student at the University of Michigan, and Karla Hernández, recent JHU doctoral student).
Reliability estimation from subsystem and full system tests (with Coire Maranzano, JHU/APL staff member).
Analysis of different forms of empirical Fisher information matrices from the point of view of validity in statistical inference and accuracy of confidence regions (with Geng Zhang, recent JHU M.S. student).
Model-free controller for supply chain management (with Chenghan Ying, recent JHU M.S. student).
Simulation-based optimization with constrained SPSA for water distribution networks (with Aimee Dalsimer, recent JHU M.S. student).
Application of discrete and mixed-variable simultaneous perturbation stochastic approximation towards developing optimal public health strategies for containing the spread of influenza and Covid-19 given limited societal resources (with Zewei Li, recent JHU M.S. student and current doctoral student at Northwestern University).
Proper scaling methods for SPSA when parameters/variables being optimized have large differences in magnitude (with Zewei Li, recent JHU M.S. student and current doctoral student at Northwestern University).
Optimal estimation of unknown parameters in model of interest based on integrating data from the subsystems and the full system (with Long Wang, recent JHU doctoral student).
Stochastic optimization for problems when the parameters being optimized are a mixture of continuous and discrete variables (with Long Wang, recent JHU doctoral student).
Computable accuracy assessment for stochastic approximation with constant step sizes (with Jingyi Zhu, recent JHU doctoral student).
Isolating faults in mathematical models with applications in financial systems and decision tree-based classifiers (with Yin Bao, recent JHU M.S. student, and Zihao Zhao, current JHU undergraduate student).
Development and evaluation of method for general constrained stochastic optimization (with Zhichao Jia, recent JHU M.S. student and current doctoral student at Georgia Tech, and Ziyi Wei, recent JHU M.S. student and current doctoral student at Virginia Tech).
Study of convergence and stationarity of Metropolis-Hastings algorithm for Monte Carlo random sampling (with Stacy Hill, JHU/APL staff member).
Formal analysis of PID controller relative to model-free controller (with James Peng, recent JHU M.S. student and current doctoral student at Virginia Tech).