Optimization, Learning and Control Group (OLC)

Research



3 PhD positions funded by EU-H2020-MSCA-ITN

I am recruiting 3 PhD candidates supported by two EU-H2020-MSCA-ITN projects:
These are full positions and the salary will be according to European Commission funding rate. You can get more information about salary and ITNs in Information Note for ITN Fellows.

I am looking for candidates with strong mathematical and computations skills (optimization, linear algebra, control) to work on theoretical and algorithmic projects at the interface between optimization, machine learning and control. The positions are for 3 years and they are available immediately (before July 2020). Interested candidates are invited to send a CV and the name of two references by email to me at: ion.necoara@acse.pub.ro.

Team

Members of the OLC group are affilieted with University Politehnica Bucharest and also with "Gheorghe Mihoc-Caius Iacob" Institute of Mathematical Statistics and Applied Mathematics of the Romanian Academy
Prof. dr. Ion Necoara
ion.necoara@upb.ro
Download CV
Bio: Ion Necoara received the B.Sc. degree in mathematics and the M.Sc. degree in optimization and statistics from the University of Bucharest, in 2000 and 2002. After graduating he worked as a Ph.D. student at the Delft Center for Systems and Control, Delft University of Technology, The Netherlands. For the period 2006-2009, he completed a Postdoctoral Fellowship at the Katholieke Universiteit Leuven, Belgium. Since 2009 he is a staff member of the Faculty of Automatic Control and Computers, University Politehnica Bucharest, where he is now a Professor of Numerical Methods and Optimization. He is also a senior researcher with "Gheorghe Mihoc-Caius Iacob" Institute of Mathematical Statistics and Applied Mathematics of the Romanian Academy. His main research interests cover various topics in developing new optimization algorithms with a focus on structure exploiting and applying optimization techniques for developing new advanced controller design algorithms for complex systems.
Senior Researcher Dr. Gabriela Marinoschi
gabriela.marinoschi@acad.ro
Bio: Gabriela Marinoschi received the B.Sc. degree in mathematics from the University of Bucharest in 1979 and a Ph.D. in mathematics from the same university in 1989. Since 1998 she is with "Gheorghe Mihoc-Caius Iacob" Institute of Mathematical Statistics and Applied Mathematics of the Romanian Academy, where she is now a senior researcher and from 2020 director of the institute. Her research interests cover various topics such as optimal control, inverse problems, dynamical systems modeling and partial differential equations.
Assoc. prof. dr. Tudor Ionescu
tudor.ionescu@upb.ro
Bio: Tudor Ionescu received an M.Sc. from the Politehnica University of Bucharest (2004) and Ph.D. from University of Groningen, The Netherlands (2009). He was a Research Associate at Imperial College London, UK (2009-2013) and at the University of Sheffield, UK (2013-2015). Currently, he is an Associate Professor at Politehnica University of Bucharest and a senior researcher with "Gheorghe Mihoc-Caius Iacob" Institute of Mathematical Statistics and Applied Mathematics of the Romanian Academy. His research interests include modelling and control of nonlinear systems with focus on modelling and model order reduction with preservation of physical structure.
Assoc. prof. dr. Lucian Toma
lucian_toma_ro@yahoo.com
Bio: Lucian Toma received an M.Sc. from the Politehnica University of Bucharest (2003) and Ph.D. from Politehnica University of Bucharest on Power Systems Control (2010). Currently, he is an Associate Professor at Politehnica University of Bucharest. His research interests include modelling and control of power systems, smart grids and smart cities.
Phd. eng. Lupu Daniela
daniela.lupu@upb.ro
Bio: Lupu Daniela received her B.Sc. and M.Sc. degree in automatic control from the University Politehnica of Bucharest in 2017 and 2019, and she is currently a Ph.D. student at the same university. Her main research interests include stochastic optimization, machine learning, modeling complex systems and predictive control.
Phd. Liliana Ghinea
Liliana.Ghinea@ugal.ro
Bio: Liliana Ghinea received her B.Sc and M.Sc. degree in mathematics and computer science from the "Dunarea de Jos" University of Galati in 2017 and 2019, and she is currently a Ph.D. student at the "Dunarea de Jos" University of Galati (Professor Marian Barbu) in co-supervision with Professor Ion Necoara from University Politehnica of Bucharest. Her main research interests include optimization, control, modeling complex systems and big data optimization.
Phd. Nitesh Kumar Singh
ns103213@gmail.com
Bio: Nitesh Kumar Singh received his B.Sc. degree in Physics, Chemistry, Mathematics from CSJM University and M.Sc. degree in Applied Mathematics from South Asian University, New Delhi, India in 2017 and 2020, respectively. He is currently a Ph.D. Student at University Politehnica of Bucharest, Romania. His main research interests include Stochastic Optimization, Machine Learning and Deep Neural Networks.
Eng. Mirel Puchianu
puchianu.m@gmail.com
Bio: Mirel Puchianu received his B.Sc in automatic control from the University Politehnica of Bucharest in 2019. His bachelor thesis was on efficient optimization methods for AC-OPF problem arising in power systems. He is currently pursuing a master at the same university. His main research interests include optimization, control, machine learning and applications in power systems.


Former members

Dr. eng. Andrei Patrascu
andrei.patrascu@acse.pub.ro
Bio: Andrei Patrascu received the M.Sc. and Ph.D. degrees in automatic control from the University Politehnica of Bucharest, Romania, in 2012 and 2015, respectively. His research interests include numerical algorithms for large scale sparse optimization problems and their application in systems and control theory.
Dr. eng. Valentin Nedelcu
valentin.nedelcu@acse.pub.ro
Bio: Valentin Nedelcu received the M.Sc. and Ph.D. degrees in automatic control from the University Politehnica of Bucharest, Romania, in 2011 and 2013, respectively. His main research interests include distributed optimization and networked control systems.
Drd. eng. Dragos Clipici
dragos.clipici@acse.pub.ro
Bio: Dragos Clipici received his M.Sc. degree in automatic control from the "Politehnica" University of Bucharest in 2010, and is currently a Ph.D. student at the "Politehnica" University of Bucharest. His main research interests include distributed optimization, robust and stochastic optimization and predictive control.




Project - ScaleFreeNet



Unitatea Executiva pentru Finantarea Invatamantului Superior, a Cercetarii, Dezvoltarii si Inovarii (UEFISCDI, Exploratory Research Project, PN-III-P4-ID-PCE-2016-0731, 2017-2019)

Scale-free modeling and optimization techniques for control of complex networks

(ScaleFreeNet)

CONTRACT NR. 39/2017

Abstract: Today's society has reached a high level of interconnection and dynamics due to the developement of technology. It is proper, in this context, to characterize the systems as large and complex, thus our interest in the subject. The first step of the project was a survey of the main existing modelling, control, and optimization techniques for complex network systems. The following period was dedicated to developing new scalable algorithms for modeling and optimization of complex network systems alongside with MPC. Expected results (6 ISI journal papers), achieved results (13 ISI journal papers accepted or provissionally accepted).

Project's research team: Prof. I. Necoara, Assoc. Prof. L. Toma,Assoc. Prof. T. Ionesc, Dr. A. Patrascu, Phd. D. Lupu



Publications


Papers in ISI Journals

  1. A. Patrascu, I. Necoara, Nonasymptotic convergence of stochastic proximal point methods for constrained convex optimization, Journal of Machine Learning Research, 18(198): 1–42, 2018

  2. A. Patrascu, I. Necoara, On the convergence of inexact projection first order methods for convex minimization, IEEE Transactions on Automatic Control,63(10): 3317–3329, 2018

  3. I. Necoara, Yu. Nesterov, F. Glineur, Linear convergence of first order methods for non-strongly convex optimization, Mathematical Programming, 175(1): 69–107, 2019

  4. A. Nedich, I. Necoara, Random minibatch subgradient algorithms for convex problems with functional constraints, Applied Mathematics and Optimization, 80(3): 801-–833, 2019

  5. I. Necoara, P. Richtarik, A. Patrascu, Randomized projection methods for convex fea sibility problems: conditioning and convergence rates, Siam Journal on Optimization, 29(4): 2814–2852, 2019

  6. I. Necoara, Faster randomized block Kaczmarz algorithms, Siam Journal on Matrix Analysis and Applications, 40(4), 1425--1452, 2019

  7. T. Sun, I. Necoara, Q. Tran-Dinh, Composite Convex Optimization with Global and Local Inexact Oracles, to appear in Computational Optimization and Applications, 2020

  8. I. Necoara, M. Takac, Randomized sketch descent methods for non-separable linearly constrained optimization, to appear in IMA Journal of Numerical Analysis, 2020

  9. I. Necoara, T. Ionescu, H2 model reduction of linear network systems by moment matching and optimization, IEEE Transactions on Automatic Control, 65(12), 1--8, 2020

  10. I. Necoara, T. Ionescu, Optimal H2 moment matching-based model reduction for linear systems by (non)convex optimization, partially accepted in Siam Journal of Control and Optimization, 2018

  11. I. Necoara, Random block projection algorithms with extrapolation for convex feasibility problems, submitted to Applied Mathematics and Optimization, 2019

  12. I. Necoara, A. Nedich, Minibatch stochastic subgradient-based projection algorithms for solving convex inequalities, submitted to Computational Optimization and Applications, 2019

  13. I. Mezghani, Q. Tran-Dinh, I. Necoara, A. Papavasiliou, A globally convergent Gauss-Newton algorithm for AC optimal power flow, submitted to European Journal of Operational Research, 2019.



Papers in conferences


  1. D. Lupu, I. Necoara, Primal and dual first order methods for SVM: applications to driver fatigue monitoring, International Conference on System Theory, Control and Computing, Sinaia, 2018.

  2. I. Necoara, M. Takac, Random coordinate descent methods for linearly constrained convex optimization, International Symposium on Mathematical Programming, Bordeaux, 2018.

  3. I. Necoara, A. Patrascu, OR-SAGA: Over-relaxed stochastic average gradient mapping algorithms for finite sum minimization, European Control Conference, Limassol, 2018

  4. I. Necoara, Random gradient algorithms for convex minimization over intersection of simple sets, European Control Conference, Napoli, 2019

  5. I. Necoara, T. Ionescu, Parameter selection for best H2 moment matching-based model approximation through gradient optimization, European Control Conference, Napoli, 2019

  6. T. Ionescu, O. Iftime, Q. Zhong, Model reduction by moment matching: case study of a FIR system, European Control Conference, Napoli, 2019.

  7. O. Fercoq, A. Alacaoglu, I. Necoara, V. Cevher, Almost surely constrained convex optimization, International Conference on Machine Learning (A* conference), Long Beach, 2019.

  8. A. Nedich, I. Necoara, Random minibatch projection algorithms for convex feasibility problems, Conference on Decision and Control, Nice, 2019

  9. A. Radu, M. Eremia, L. Toma, Optimal charging coordination of electric vehicles considering distributed energy resources, IEEE PES PowerTech Conference, Milano, 2019.

  10. D. Sidea, L. Toma, M. Sanduleac, I. Picioroaga, V. Boicea, Optimal BESS Scheduling Strategy in Microgrids Based on Genetic Algorithms, IEEE PES PowerTech Conference, Milano,2019

  11. X. Cheng, I. Necoara, A suboptimal H2 clustering-based model reduction approach for linear network systems, European Control Conference, 2020

Project - MoCOBiDS



Unitatea Executiva pentru Finantarea Invatamantului Superior, a Cercetarii, Dezvoltarii si Inovarii (UEFISCDI, Human Resources, 2015-2017)

Modelling, Control and Optimization for Big Data Systems

(MoCOBiDS)

CONTRACT NR. 176/2015

Abstract: Experiments, observations and numerical simulations in many areas of science and business are currently generating terabytes of data. Analyses of the information contained in these data sets have already led to major breakthroughs in fields ranging from genomics to power grids and process industry. The availability of these massive data sets is transforming society and the way we think about information storage, retrieval and data processing. Not only because our team has already acquired expertise on Big Data Systems, but also because of the potential for future applications, we have identified modeling, control and optimization for big data systems as the common theme for this research proposal. The central objective of this proposal is the analysis, design and implementation of data-driven mathematical methods and numerical algorithms for the analysis and optimization of Big Data Systems, as well as modeling and control challenges. While inspired by concrete cases from application ranging from data access networks, power grids to process industry, the real focus in this project will be on tackling generic problems starting from quantitative measured data collected from Big Data Systems and developing efficient numerical algorithms for solving them. We will develop novel algorithms for modeling, control and optimization of Big Data Systems, implement the new algorithms in a programming language, test them in a wide variety of applications and include them in a toolbox.

Project's research team: Prof. Ion Necoara, Dr. Andrei Patrascu, Dr. Valentin Nedelcu, PhD Dragos Clipici.

Expected results:




Publications


Papers in ISI Journals

  1. I. Necoara, Yu. Nesterov, F. Glineur, Random block coordinate descent for linearly-constrained optimization over networks, Journal of Optimization Theory and Applications, to appear, pp. 1-26, 2016

  2. I. Necoara, A. Patrascu, Iteration complexity analysis of dual first order methods for conic convex programming, Optimization Methods & Software, 31(3):645--678, 2016.

  3. N.A. Nguyen, S. Olaru, P. Rodriguez-Ayerbe, M. Hovd, I. Necoara, Constructive solution of inverse parametric linear/quadratic programming problems, Journal of Optimization Theory & Applications, DOI 10.1007/s10957-016-0968-0, 2016.

  4. A. Patrascu, I. Necoara, Q. Tran-Dinh, Adaptive inexact fast augmented Lagrangian methods for constrained convex optimization, Optimization Letters, DOI:10.1007/s11590- 016-1024-6: 1-18, 2016.

  5. I. Necoara, D. Clipici, Parallel random coordinate descent methods for composite minimization: convergence analysis and error bounds, SIAM Journal on Optimization, vol. 26, no. 1, pp. 197-226, 2016.

  6. A. Patrascu, I. Necoara, On the convergence of inexact projection first order methods for convex minimization, IEEE Transactions on Automatic Control, 2017.

  7. I. Necoara, A. Patrascu and F. Glineur, Complexity of first order Lagrangian and penalty methods for conic convex programming, Optimization Methods and Software, 2017.


Book chapters

  1. I. Necoara, Coordinate gradient descent methods, chapter in Taylor & Francis LLC - CRC Press, pp. 1-30, 2016.

  2. I. Necoara, A. Patrascu, A. Nedich, Complexity certifications of first order inexact Lagrangian methods for general convex programming, chapter in Springer, pp. 1–22, 2015.


Papers in conferences


  1. I. Necoara, A. Patrascu, P. Richtarik, Randomized projection methods for convex feasibility problems, submitted to SIAM Conference on Optimization 2017.

  2. I. Necoara, V. Nedelcu, D. Clipici, L. Toma, On fully distributed dual first order methods for convex network optimization, submitted to IFAC World Congress, 2017.

  3. T. Ionescu, I. Necoara, A scale-free moment matching-based model reduction technique of linear networks, submitted to IFAC World Congress, 2017.

  4. A. Patrascu, I. Necoara, Inexact projection primal first order methods for strongly convex minimization, submitted to IFAC World Congress, 2017.

  5. A. Patrascu, I. Necoara, Complexity certifications of inexact projection primal gradient method for convex problems: application to embedded MPC, Mediterranean Conference on Control and Automation, 2016.

  6. I. Necoara, L. Toma, V. Nedelcu, Optimal voltage control for loss minimization based on sequential convex programming, IEEE Conference Innovative Smart Grid Technologies, 2016.

  7. I. Necoara, Yu. Nesterov, F. Glineur, Linear convergence of first order methods for nonstrongly convex optimization, invited paper in session: Recent advances on convergence rates of first-order methods, International Conference Continuous Optimization, 2016.

  8. I. Necoara, Linear convergence of gradient type methods for non-strongly convex optimization, invited paper in session: Analyse non-lisse et optimisation, Colloque Franco- Roumain de Mathematiques Appliquees, 2016.

  9. I. Necoara, A. Patrascu, F. Glineur, Complexity of first order inexact Lagrangian and penalty methods for conic convex programming, invited paper in session: First order methods for convex optimization problems, European Conference on Operational Research, 2016.

Applications

In our Optimization, Learning and Control Lab, we have developed a number of applications, from which we mention the following:



Software

  • The DuQuad optimization toolbox solves convex quadratic programs using Dual First Order Optimization Algorithms. The algorithms has predictable and fast convergence, low memory footprint, and uses only basic arithmetic and logical operations. DuQuad is therefore suited to be utilized by real-time applications running on low-cost hardware such as simple microcontrollers.DuQuad has an user friendly Matlab interface for maximum productivity, and the algorithms are implemented in efficient C-code. DuQuad is open source and can be downloaded from the git repository (for downloading DuQuad, click here).


  • Primal-Dual Toolbox for SVM (PD-SVM): Matlab code toolbox for solving large-scale SVM problems - download


  • Parallel Optimization Toolbox (POPT): C code toolbox for solving large-scale structured QPs - download

  • Pagina creata la data de 15/11/2015. Ultima modificare:4/6/2020.