Kiran Sagar Kollepara

Personal details: ----------------- Mail: <sagar1990@gmail.com>, <kiran-sagar.kollepara@ec-nantes.fr> Nationality: Indian Education: ---------- Erasmus Master of Science in Computational Mechanics (2015-2017) [UPC, Barcelona, Spain](http://www.upc.edu "Universitat Politècnica de Catalunya"), [ECN, Nantes, France](http://www.ec-nantes.fr "Ecole Centrale de Nantes") Bachelor of Technology in Aerospace Engineering (2008-2012) [IIST, Thiruvananthapuram, India](http://www.iist.ac.in "Indian Institute of Space Science and Technology") Research Experience: -------------------- Postdoctoral Research : Non-intrusive reduced order modelling for non-cartesian parametric domains using sparse subspace learning approach, 2023-current [Research Institute in Civil and Mechanical Engineering](https://gem.ec-nantes.fr "GeM"), [ECN, Nantes, France](http://www.ec-nantes.fr "Ecole Centrale de Nantes") PhD thesis titled "Low-rank and sparse approximations for contact mechanics" , defended on 5th July 2022 [Research Institute in Civil and Mechanical Engineering](https://gem.ec-nantes.fr "GeM"), [ECN, Nantes, France](http://www.ec-nantes.fr "Ecole Centrale de Nantes") > Limitations of low-rank models to solve contact mechanics problems with regards to the local nature of contact position is investigated. To mitigate the errors due to these local effects, overcomplete dictionary-based models explored. Two different strategies are proposed to resolve the dictionary-based approximation problem. Finally, DTW-based non-linear interpolations are utilized to reduce the offline cost of creating overcomplete dictionaries. Master Thesis: Enforcing the tensor structure of non-separable and non-linear problems [Institut de Calcul Intensif](https://ici.ec-nantes.fr/ "ICI"), [ECN, Nantes, France](http://www.ec-nantes.fr "Ecole Centrale de Nantes") > Proper Generalized Decomposition (PGD) is a well known tensorization based technique to efficiently solve high-dimensional PDEs. Solving non-linear PDEs in the framework of PGD is non-trivial, as a tensorized form of the non-linear term is not readily available. To this end, interpolative techniques were explored to perform tensorization of the nonlinear term. Numerical analysis of retrograde flow in the lamellipodium of cell [Laboratori de Càlcul Numèric](https://www.lacan.upc.edu/ "LaCàN"), [UPC, Barcelona, Spain](http://www.upc.edu "Universitat Politècnica de Catalunya") > Keratocytic cells use a polymerization-depolymerization cycle known as actin treadmilling as a mechanism of cell transport. A mathematical model of tread-milling process is solved numerically. A coupled system of transport equations that model the treadmiling process, are modeled using Finite Element method aided by SUPG stabilization. Internship: Real-time computation of the Incident Wave fields on the boundary of a harbour [Laboratori de Càlcul Numèric](https://www.lacan.upc.edu/ "LaCàN"), [UPC, Barcelona, Spain](http://www.upc.edu "Universitat Politècnica de Catalunya") > Reduced models to compute incident wave fields were developed for prediction of wave behaviour at harbours Teaching Experience ------------------- Tutorials: Linear Algebra, Ordinary differential equations Practicals: Programming (C++), Data structures (Python), Numerical methods and optimization (MATLAB) Skills ------ Language: French (intermediate), English (fluent) Programming: Python, MATLAB (advanced) and C++ (intermediate), OpenMP (beginner)