About

My research interests revolve around nonsmooth and nonconvex optimization, game theory, and bilevel optimization, with applications in machine learning, systems control, and economics. I received my PhD in 2020 jointly from KU Leuven (Belgium) and IMT Lucca (Italy). Prior to joining IMT Lucca, I held a Postdoc position with Research Foundation Flanders (FWO) at KU Leuven. You can read more about my research in my publications.

Publications

Preprints

• [P1]Adeoye, A. D., Latafat, P., Bemporad, A. (2025). “A proximal augmented Lagrangian method for nonconvex optimization with equality and inequality constraints,” arXiv:2509.02894
• [P2]Pethick, T., Latafat, P., Lewandowski, B., Xu, Z., Kairouz, P., Patrinos, P., Cevher, V. (2025). “iFedDR: Auto-tuning local computation with inexact Douglas–Rachford splitting in federated learning”

Journal Papers

• [J1]Upadhyaya, M., Latafat, P., Giselsson, P. “A Lyapunov analysis of Korpelevich's extragradient method with fast and flexible extensions,” Mathematical Programming (published online), 2026
• [J2]Evens, B., Latafat, P., Patrinos, P. “Spingarn's method and progressive decoupling beyond elicitable monotonicity,” Computational Optimization and Applications (to appear), 2026
• [J3]Ou, H., Latafat, P., Themelis, A. “Linesearch-free adaptive Bregman proximal gradient for convex minimization without relative smoothness,” Journal of Optimization Theory and Applications (to appear), 2026
• [J4]Latafat, P., Themelis, A., Stella, L., Patrinos, P. “Adaptive proximal algorithms for convex optimization under local Lipschitz continuity of the gradient,” Mathematical Programming 213(1):433–471, 2025
• [J5]Evens, B., Latafat, P., Patrinos, P. “Convergence of the Chambolle–Pock algorithm in the absence of monotonicity,” Journal of Optimization Theory and Applications 206(1):7, 2025
• [J6]Latafat, P., Themelis, A., Villa, S., Patrinos, P. “On the convergence of proximal gradient methods for convex simple bilevel optimization,” Journal of Optimization Theory and Applications 204(3):51, 2025
• [J7]Evens, B., Pas, P., Latafat, P., Patrinos, P. “Convergence of the preconditioned proximal point method and Douglas–Rachford splitting in the absence of monotonicity,” Mathematical Programming 214(1–2):247–301, 2025
• [J8]Behmandpoor, P., Latafat, P., Themelis, A., Moonen, M., Patrinos, P. “SPIRAL: A superlinearly convergent incremental proximal algorithm for nonconvex finite sum minimization,” Computational Optimization and Applications 88(1):71–106, 2024
• [J9]Latafat, P., Themelis, A., Ahookhosh, M., Patrinos, P. “Bregman Finito/MISO for nonconvex regularized finite sum minimization without Lipschitz gradient continuity,” SIAM Journal on Optimization 32(3):2230–2262, 2022
• [J10]Latafat, P., Themelis, A., Patrinos, P. “Block-coordinate and incremental aggregated proximal gradient methods for nonsmooth nonconvex problems,” Mathematical Programming 193(1):195–224, 2022
• [J11]Latafat, P., Patrinos, P. “Primal-dual algorithms for multi-agent structured optimization over message-passing architectures with bounded communication delays,” Optimization Methods and Software 37(6):2052–2079, 2022
• [J12]Latafat, P., Freris, N. M., Patrinos, P. “A new randomized block-coordinate primal-dual proximal algorithm for distributed optimization,” IEEE Transactions on Automatic Control 64(10):4050–4065, 2019
• [J13]Latafat, P., Patrinos, P. “Asymmetric forward–backward–adjoint splitting for solving monotone inclusions involving three operators,” Computational Optimization and Applications 68(1):57–93, 2017

Conference Papers

• [C1]Oikonomidis, K. A., Laude, E., Latafat, P., Themelis, A., Patrinos, P. “Adaptive proximal gradient methods are universal without approximation,” ICML 2024 (spotlight award)
• [C2]Latafat, P., Themelis, A., Patrinos, P. “On the convergence of adaptive first order methods: Proximal gradient and alternating minimization algorithms,” L4DC 2024, pp. 197–208
• [C3]Pethick, T., Fercoq, O., Latafat, P., Patrinos, P., Cevher, V. “Solving stochastic weak Minty variational inequalities without increasing batch size,” ICLR 2023
• [C4]Pethick, T., Latafat, P., Patrinos, P., Fercoq, O., Cevher, V. “Escaping limit cycles: Global convergence for constrained nonconvex-nonconcave minimax problems,” ICLR 2022 (spotlight award)
• [C5]Evens, B., Latafat, P., Themelis, A., Suykens, J., Patrinos, P. “Neural network training as an optimal control problem: An augmented Lagrangian approach,” CDC 2021, pp. 5136–5143
• [C6]Latafat, P., Patrinos, P. “Multi-agent structured optimization over message-passing architectures with bounded communication delays,” CDC 2018, pp. 1688–1693
• [C7]Latafat, P., Bemporad, A., Patrinos, P. “Plug and play distributed model predictive control with dynamic coupling: A randomized primal-dual proximal algorithm,” ECC 2018, pp. 1160–1165
• [C8]Latafat, P., Stella, L., Patrinos, P. “New primal-dual proximal algorithm for distributed optimization,” CDC 2016, pp. 1959–1964
• [C9]Latafat, P., Palumbo, P., Pepe, P., Kovács, L., Panunzi, S., De Gaetano, A. “An LMI-based controller for the glucose-insulin system,” ECC 2015, pp. 7–12

Contribution to Books

• [B1]Latafat, P., Patrinos, P. “Primal-dual proximal algorithms for structured convex optimization: A unifying framework,” in Large-Scale and Distributed Optimization (Giselsson, P., Rantzer, A., eds.), vol. 2227, Lecture Notes in Mathematics, Springer, pp. 97–120, 2018

PhD Thesis

• [PT]Latafat, P. (2020) “Distributed proximal algorithms for large-scale structured optimization,” Joint PhD, KU Leuven & IMT Lucca (“Excellent” evaluation with Doctor Europaeus certificate)

Teaching

Software

• AdaProx: Julia implementation of the adaptive proximal gradient (adaPGM) and adaptive primal-dual (adaPDM) methods, based on the results in [J4].
• CIAOAlgorithms: Julia implementation of several stochastic and incremental methods for finite-sum minimization, based on the results in [J10].
• AdaBiM: Julia solver for simple bilevel optimization problems, based on the results in [J6].
• ProximalAlgorithms: generic Julia solver for nonsmooth optimization, implementing, among other methods, algorithms from [J13], [J12], and [B1].
• fs-palm: Python implementation of the feasible-start proximal augmented Lagrangian method, based on [P1].

See also JuliaFirstOrder and KUL-OPTEC on GitHub for related packages.

Contact

• Email
• Address IMT School for Advanced Studies Lucca, Piazza S. Francesco 19, 55100 Lucca, Italy
• GitHub github.com/pylat
• Google Scholar scholar.google.com
• ORCID 0000-0002-7969-8565
• ResearchGate Profile