CV

                                                         Tahamina Akter

PhD Candidate in Numerical Mathematics | Numerical Linear Algebra & Scientific Computing

Berlin, Germany • +49 176 47679506 • taha.akter@yahoo.com

linkedin.com/in/tahaminaakter • arxiv.org/abs/2609.11363

RESEARCH PROFILE

DFG-funded doctoral researcher at Technische Universität Braunschweig, supervised by Prof. Dr. Matthias Bollhöfer. Develops scalable approximate selected-inversion methods for large sparse matrices using incomplete LDLᵀ and multilevel factorizations, Schur complements, sparse inverse approximations, and spectral / low-rank corrections. Experienced in algorithm design, numerical error analysis, reproducible MATLAB experiments, and high-performance scientific computing; currently developing Julia implementations.

RESEARCH EXPERIENCE

Doctoral Researcher / Research Assistant Feb 2023 – Present

Institute for Numerical Mathematics, Technische Universität Braunschweig, Germany

Research within the DFG project “Numerical Algorithms, Frameworks, and Scalable Technologies for Extreme-Scale Computing” (GEPRIS 470857344).

Develop single-level approximate selected-inversion approaches using incomplete LDLᵀ factorizations; compare SelInv, Neumann-based inversion, hybrid Mix, and Mix-SPAI refinements.

Investigate multilevel ILDL, coarse Schur-complement operators, inverse-based preconditioning, and adaptive spectral corrections for accuracy and robustness.

Design numerical benchmarks on sparse test matrices, studying diagonal/selected-entry errors, fill, computational cost, and PCG convergence; use MATLAB, Julia (in development), JANUS, ILUPACK, Linux/HPC.

Student Assistant 2016 – 2018

Julius-Maximilians-Universität Würzburg, Germany

Supported mathematics research and teaching activities.

EDUCATION

PhD Candidate in Mathematics Feb 2023 – Present

Technische Universität Braunschweig, Germany

Supervisor: Prof. Dr. Matthias Bollhöfer | Expected PhD completion: March 2027

M.Sc. Mathematics International 2016 – 2020

Julius-Maximilians-Universität Würzburg, Germany

Grade: 2.1 (German grading scale), 120 ECTS | Thesis: “Feature Reconstruction in Imaging”

M.Sc. Mathematics 2014

National University, Bangladesh

First Class (74.5%).

B.Sc. Mathematics 2013

National University, Bangladesh

First Class (70.42%).

PUBLICATIONS & MANUSCRIPTS

Research preprint (2026). Akter, T., Bollhöfer, M., and Schenk, O. “Scalable Approximate Selected Inversion Based on Single-Level Incomplete LDLᵀ Factorization and Spectral Corrections for Large Sparse Systems.” arXiv:2609.11363 (posted September 2026).

Manuscript in preparation. Akter, T., Bollhöfer, M., and Schenk, O. “Scalable Coarse-to-Fine Approximate Selected Inversion Based on Multilevel Incomplete LDLᵀ Factorization and Spectral Corrections for Large Sparse Systems” (working title).

SELECTED CONFERENCE PRESENTATIONS

SIAM Conference on Parallel Processing for Scientific Computing (PP26), Berlin, Germany, 3–6 Mar 2026 — minisymposium talk.

27th International Linear Algebra Society Conference (ILAS 2026), Virginia Tech, USA, 18–22 May 2026 — minisymposium talk.

Preconditioning Techniques for Scientific and Industrial Applications, ICMS, Edinburgh, UK, 27–29 May 2026 — contributed talk.

Workshop on Approximate Computing in Numerical (Multi-)Linear Algebra, Magdeburg, Germany, 31 Aug–1 Sep 2026 — contributed talk.

TEACHING AND ACADEMIC SERVICE

Lecturer in Mathematics 2015 – 2016

Dhaka City College, Dhaka, Bangladesh

Taught Mathematics for Economics, Business Mathematics, Differential Calculus, and Integral Calculus.

Conference participation: EMOSC 2025 — Energy-Based Modeling, Simulation and Control of Dynamical Systems.

TECHNICAL EXPERTISE

Numerical methods: Approximate selected inversion; sparse direct and iterative methods; ILU / ILDLᵀ; multilevel preconditioning; Schur complements; sparse approximate inverses; spectral / low-rank corrections; PCG.

Programming and software: MATLAB; Julia (ongoing research implementation); SQL; LaTeX; Linux; JANUS; ILUPACK; HPC computing environments.

AWARDS, LANGUAGES & MEMBERSHIP

Scholarships: Bavarian State Scholarship (2018); DAAD STIBET Scholarship (2017).

Languages: Bengali (native); English (professional); German (B2).

Membership: Society for Industrial and Applied Mathematics (SIAM).

ACADEMIC REFERENCE

Prof. Dr. Matthias Bollhöfer | Institute for Numerical Mathematics, Technische Universität Braunschweig | m.bollhoefer@tu-bs.de