DEPARTMENT OF APPLIED MATHEMATICS 109 Email zaikun.zhang@polyu.edu.hk Qualification BSc (Jilin University) PhD (Chinese Academy of Sciences) ORCID ID 0000-0001-8934-8190 Dr ZHANG Zaikun Assistant Professor Research Areas Theory and algorithms of mathematical optimisation, especially randomised algorithms, methods that do not need derivatives (derivative-free methods) and noisy problems Representative Publications • Comput. Optim. Appl., 72: 525–559, 2019 • SIAM J. Sci. Comput., 40(4): A2174–A2199, 2018 • IMA J. Numer. Anal., 38(3): 1579–1597, 2018 • SIAM J. Optim., 25(3): 1515–1541, 2015 • Math. Program., 146(1–2): 77–96, 2014 Research Outputs • PDFO, a cross-platform package providing Python and MATLAB interfaces for a class of derivative-free optimization methods. It has been downloaded more than 46,000 times and included in industrial software such as GEMSEO (https://gemseo.readthedocs.io/en/stable/). URL: http://www.pdfo.net. • COBYQA, a derivative-free solver for general nonlinear optimization. It uses only function values of the objective function and nonlinear constraint functions. No derivative information is needed. URL: http://www.cobyqa.com • PRIMA, a package of state-of-the-art derivative-free optimization solvers for unconstrained, bound-constrained, linearly constrained, and nonlinearly constrained problems. URL: http://www.libprima.net Awards and Achievements • Member of the 2023 SIAG/OPT Early Career Prize Selection Committee, 2022–2023 • Best Paper Award, Department of Applied Mathematics, 2021 • Comput. Optim. Appl. 2019 Best Paper Prize, 2020 • Member of the Local Organizing Committee of SIAM Conference on Optimization 2020, 2019–2020 • Junior Council Member of the Mathematical Programming Branch of OR Society of China, 2018–present • Member of the Organizing Committee of The Fourth International Conference on Continuous Optimization, 2013, Lisbon, Portugal Effectiveness of trust-region methods with respect to eta Effectiveness of trust-region methods with respect to error Effectiveness of the GNU Octave fminunc versus the stablised fminunc
RkJQdWJsaXNoZXIy Mjc5OTU=