Vassilis Kalantzis - Senior Research Scientist, IBM Research 
 



Personal Homepage of Vassilis Kalantzis



Contact


IBM Research
Thomas J. Watson Research Center
1101 Kitchawan Rd, Yorktown Heights, NY 10598, USA
vkal (at) ibm (dot) com


About


I am a Senior Research Scientist at IBM Research. Prior to my current role I was a Herman H. Goldstine Memorial Postdoctoral Fellow. I received my PhD (2018) and M.Sc (2016) in Computer Science and Engineering from the University of Minnesota, Twin Cities, U.S., and M.Sc (2014) and M.Eng (2011) in Computer Engineering and Informatics from the University of Patras, Greece.


Journal Publications

[24.]   Regenerative Ulam-von Neumann Algorithm: An Innovative Markov chain Monte Carlo Method for Matrix Inversion
S. Ghosh, L. Horesh, V. Kalantzis, Y. Lu, and T. Nowicki
SIAM Journal on Matrix Analysis and Applications (2025), To Appear
[23.]   Stable Iterative Refinement Algorithms for Solving Linear Systems
C. W. Wu, M. S. Squillante, V. Kalantzis, and L. Horesh
Journal of Computational and Applied Mathematics (2025), To Appear
[22.]   On the Variance of Schatten p-Norm Estimation with Gaussian Sketching Matrices
S. Ghosh, L. Horesh, V. Kalantzis, Y. Lu, and T. Nowicki
Monte Carlo Methods and Applications (2025), To Appear
DOI: 10.1515/mcma-2025-2006
[21.]   Single-Pass Top-N Subgraph Centrality of Graphs via Subspace Projections
V. Kalantzis, G. Kollias, S. Ubaru, N. Abe, and L. Horesh
Journal of Complex Networks (2025), Vol. 13, No. 1
DOI: 10.1093/comnet/cnae049
[20.]   Straggler-tolerant Stationary Methods for Linear Systems
V. Kalantzis, Y. Xi, L. Horesh, and Y. Saad
SIAM Journal on Scientific Computing (2025), To Appear
DOI: 10.1137/24M1673346
[19.]   A Rational Filtering Algorithm for Sequences of Shifted Symmetric Linear Systems with Applications to Frequency Response Analysis
A. P Austin, L. Horesh, and V. Kalantzis
SIAM Journal on Scientific Computing (2024), Vol. 46, No. 6, pp. A3552-A3573
DOI: 10.1137/23M1578474
Software: Matlab
[18.]   A Parallel Algorithm for Computing Partial Spectral Factorizations of Matrix Pencils via Chebyshev Approximation
T. Xu, A. P. Austin, V. Kalantzis, and Y. Saad
SIAM Journal on Scientific Computing (2024), Vol. 46, No. 2, pp. S324-S351
DOI: 10.1137/22M1501155
Software: Github
[17.]   Directed Graph Transformers
Q. Wang, G. Kollias, V. Kalantzis, N. Abe, and M. J. Zaki
Transactions on Machine Learning Research (2024)
OpenReview Archive
Software: Github
[16.]   parGeMSLR: A Parallel Multilevel Schur Complement Low-Rank Preconditioning and Solution Package for General Sparse Matrices
T. Xu, V. Kalantzis, R. Li, Y. Xi, G. Dillon, and Y. Saad
Parallel Computing (2022), Vol. 113
DOI: 10.1016/j.parco.2022.102956
Software: Github
[15.]   Enhanced Algebraic Substructuring for Symmetric Generalized Eigenvalue Problems
V. Kalantzis and L. Horesh
Numerical Linear Algebra with Applications (2022), Vol. 30, No. 2
DOI: 10.1002/nla.2473
[14.]   Fast Randomized non-Hermitian Eigensolvers Based on Rational Filtering and Matrix Partitioning
V. Kalantzis, Y. Xi, and L. Horesh
SIAM Journal on Scientific Computing (2021), Vol. 43, No. 5, pp. S791-S815
DOI: 10.1137/19M1280004
[13.]   A Domain Decomposition Rayleigh-Ritz Algorithm for Symmetric Generalized Eigenvalue Problems
V. Kalantzis
SIAM Journal on Scientific Computing (2020), Vol. 42, No. 6, pp. C410-C435
DOI: 10.1137/19M1280004
[12.]   A Spectral Newton-Schur Algorithm for the Solution of Symmetric Generalized Eigenvalue Problems
V. Kalantzis
Electronic Transactions on Numerical Analysis (2020), Vol. 52, pp. 132-153
DOI: 10.1553/etna_vol52s132
[11.]   TeraPCA: a Fast and Scalable Method to Study Genetic Variation in Tera-scale Genotypes
A. Bose, V. Kalantzis, E. M. Kontopoulou, M. Elkadi, P. Paschou, and P. Drineas
Bioinformatics (2019), Volume 35, Issue 19, pp. 3679-3683
DOI: 10.1093/bioinformatics/btz157
Software: C++
[10.]   Stable Computation of Generalized Matrix Functions via Polynomial Interpolation
J. L. Aurentz, A. P. Austin, M. Benzi, and V. Kalantzis
SIAM Journal on Matrix Analysis and Applications (2019), Vol. 40, No. 1, pp. 210-234
DOI: 10.1137/18M1191786
[9.]   EIGENREC: Generalizing PureSVD for Effective and Efficient Top-N Recommendations
A. N. Nikolakopoulos, V. Kalantzis, E. Gallopoulos, and J. D. Garofalakis
Knowledge and Information Systems (2019), Vol. 58, No. 1, pp. 59-81
DOI: 10.1007/s10115-018-1197-7
[8.]   Beyond Automated Multilevel Substructuring: Domain Decomposition with Rational Filtering
V. Kalantzis, Y. Xi, and Y. Saad
SIAM Journal on Scientific Computing (2018), Vol. 40, No. 4, pp. C477-C502
DOI: 10.1137/17M1154527
SIAM Student Paper Prize 2020
[7.]   A Hierarchical Low-Rank Schur Complement Preconditioner for Indefinite Linear Systems
G. Dillon, V. Kalantzis, Y. Xi, and Y. Saad
SIAM Journal on Scientific Computing (2018), Vol. 40, No. 4, pp. A2234-A2252
DOI: 10.1137/17M1143320
[6.]   Domain Decomposition Approaches for Accelerating Contour Integration Eigenvalue Solvers for Symmetric Eigenvalue Problems
V. Kalantzis, J. Kestyn, E. Polizzi, and Y. Saad
Numerical Linear Algebra with Applications (2018), Vol. 25, No. 5
DOI: 10.1002/nla.2154
[5.]   A Scalable Iterative Dense Linear System Solver for Multiple Right-Hand Sides in Data Analytics
V. Kalantzis, C. Malossi, C. Bekas, A. Curioni, E. Gallopoulos, and Y. Saad
Parallel Computing (2018), Vol. 74, pp. 136-153
DOI: 10.1016/j.parco.2017.12.005
Software: Fortran
[4.]   Cucheb: A GPU Implementation of the Filtered Lanczos Procedure
J. L. Aurentz, V. Kalantzis, and Y. Saad
Computer Physics Communications (2017), Vol. 220, pp. 332-340
DOI: 10.1016/j.cpc.2017.06.016
Software: C++/CUDA
[3.]   Spectral Schur Complement Techniques for Symmetric Eigenvalue Problems
V. Kalantzis, R. Li, and Y. Saad
Electronic Transactions on Numerical Analysis (2016), Vol. 45, pp. 305-329
[2.]   Estimating the Trace of the Matrix Inverse by Interpolating from the Diagonal of an Approximate Inverse
L. Wu, J. Laeuchli, V. Kalantzis, A. Stathopoulos, and E. Gallopoulos
Journal of Computational Physics (2016), Vol. 326, pp. 828-844
DOI: 10.1016/j.jcp.2016.09.001
[1.]   Accelerating Data Uncertainty Quantification by Solving Linear Systems with Multiple Right-Hand Sides
V. Kalantzis, C. Bekas, A. Curioni, and E. Gallopoulos
Numerical Algorithms (2013), Vol. 62, No. 4, pp. 637-653
DOI: 10.1007/s11075-012-9687-2


Conference Publications

[13.]   Multi-Sense Embeddings for Language Models and Knowledge Distillation
Q. Wang, M. Zaki, G. Kollias, and V. Kalantzis
In Proceedings of Findings of the 63rd Annual Meeting of the Association for Computational Linguistics (Findings of ACL 2025)
[12.]   Asynchronous Randomized Trace Estimation
V. Kalantzis, S. Ubaru, C. W. Wu, G. Kollias, and L. Horesh
In Proceedings of the 27th International Conference on Artificial Intelligence and Statistics (AISTATS 2024)
[Acceptance rate: 28%]
Proceedings of Machine Learning Research
Software: Github
[11.]   Topological Data Analysis on Noisy Quantum Computers
I. Y. Akhalwaya, S. Ubaru, K. L. Clarkson, M. S. Squillante, V. Jejjala, Y. He, K. Naidoo, V. Kalantzis, and L. Horesh
In Proceedings of the 12th International Conference on Learning Representations (ICLR 2024)
Selected for oral presentation (1.2% of total submissions)
[Acceptance rate: 31%]
OpenReview Archive
[10.]   Solving Sparse Linear Systems via Flexible GMRES with In-Memory Analog Preconditioning                                         
V. Kalantzis et al.
In Proceedings of the 27th IEEE Conference on High Performance Extreme Computing (HPEC 2023)
DOI: 10.1109/HPEC58863.2023.10363502
[9.]   Matrix Resolvent Eigenembeddings for Dynamic Graphs
V. Kalantzis and P. Traganitis
In Proceedings of the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2023)
DOI: 10.1109/ICASSP49357.2023.10096476
[8.]   Quantum Graph Transformers
G. Kollias, V. Kalantzis, T. Salonidis, and S. Ubaru
In Proceedings of the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2023)
DOI: 10.1109/ICASSP49357.2023.10096345
[7.]   Accelerating Matrix Trace Estimators by Aitken’s Δ-squared Process
V. Kalantzis, G. Kollias, S. Ubaru, and T. Salonidis
In Proceedings of the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2023)
DOI: 10.1109/ICASSP49357.2023.10095081
[6.]   Directed Graph Auto-Encoders
G. Kollias, V. Kalantzis, T. Ide, A. Lozano, and N. Abe
In Proceedings of the 36th AAAI Conference on Artificial Intelligence (AAAI 2022)
[Acceptance rate: 15%]
DOI: 10.1609/aaai.v36i7.20682
Software: Github
[5.]   Solving Sparse Linear Systems with Approximate Inverse Preconditioners on Analog Devices
V. Kalantzis et al.
In Proceedings of the 25th IEEE Conference on High Performance Extreme Computing (HPEC 2021)
DOI: 10.1109/HPEC49654.2021.9622816
[4.]   Projection Techniques to Update the Truncated SVD of Evolving Matrices with Applications
V. Kalantzis, G. Kollias, S. Ubaru, A. N. Nikolakopoulos, L. Horesh, and K. L. Clarkson
In Proceedings of the 38th International Conference on Machine Learning (ICML 2021)
[Acceptance rate: 21%]
Proceedings of Machine Learning Research
[3.]   Sparse Graph-Based Sketching for Numerical Linear Algebra
D. Hu, S. Ubaru, A. Gittens, L. Horesh, K. L. Clarkson, and V. Kalantzis
In Proceedings of the 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2021)
DOI: 10.1109/ICASSP39728.2021.9414030
[2.]   Factored Proximity Models for Top-N Recommendations
A. N. Nikolakopoulos, V. Kalantzis, E. Gallopoulos, and J. D. Garofalakis
In Proceedings of the 8th IEEE International Conference on Big Knowledge (ICBK 2017)
[Acceptance rate: 27%]
DOI: 10.1109/ICBK.2017.14
Software: Fortran
Best Paper Award
[1.]   PFEAST: A High Performance Sparse Eigenvalue Solver Using Distributed-Memory Linear Solvers
J. Kestyn, V. Kalantzis, E. Polizzi, and Y. Saad
In Proceedings of the 2016 ACM/IEEE Supercomputing Conference (SC 2016)
[Acceptance rate: 18%]
DOI: 10.1109/SC.2016.15




PhD Thesis


I obtained my PhD (Advisor: Yousef Saad) in 09/2018 from the University of Minnesota, Twin Cities. My dissertation is titled "Domain decomposition algorithms for the solution of sparse symmetric generalized eigenvalue problems". You can view/download a copy here.





Patents


A list of filed patents can be found here.




Disclaimer


Any opinions, statements, or conclusions expressed in the above reports do not necessarily reflect the views of the acknowledged funding sponsors or International Business Machines Corporation, Inc.