1979-1984, Carnegie-Mellon University, Ph.D. (Applied Mathematics)
1973-1977, University of Buenos Aires, M.S. (Mathematics)


Email: easocolovsky@nsu.edu
Phone: 757-823-2327

Address:

Center for Biotechnology and Biomedical Sciences
Norfolk State University
700 Park Avenue
Norfolk, VA 23504

I am responsible for Bioinformatics R&D and Computational Resources at the CBBS, including providing guidance and support for all aspects of computation and the use of software and hardware. The core computational facility consists of four 64 bit Xeon Linux workstations and a server; its main tool is the SYBIL Software suite for Protein Modeling and Computational Drug Design.

I have been involved in Applied Math and Scientific Computation since my early years in college, when I had to travel to the University of La Plata, to use their computers and be mentored by Prof. P. E. Zadunaisky, a well known astronomer and numerical analyst, on the computation and application of Least Squares approximate solutions. After graduation I worked for Pfizer, and then took positions at the School of Engineering of the University of Buenos Aires and the National Institute of Hydric Science and Technology, under the direction of Dr. G. Marshall a computational fluid dynamics expert. Later I joined Computation Facility of the National Atomic Energy Commission in Buenos Aires.

At these jobs I worked on applied problems like, the simulation of fluoride pollution from an aluminum smelter in Puerto Madryn (Patagonia), the selection of representative subsets of meteorological observation stations, and the calibration of neutron sensors in a nuclear reactor, among other projects. After finishing an MS in Math, I enrolled in a PhD program at Carnegie-Mellon University in Pittsburgh, to be able to tackle more complex problems, in particular, physical phenomena modeled by partial differential equations. There under the direction of Prof George Fix, I worked on finite element and finite differences modeling of electric arcs, and under the supervision of my advisor Prof Richard C MacCamy in collaboration with Prof Morton E. Gurtin, I worked on finite speed diffusion problems that appear, for example, in porous media flow and population dynamics.

Modeling solidification and crystal growth using Phase Field type equations, in collaborative work with Prof G. Caginalp at the University of Pittsburgh, it became necessary to use Cray supercomputers and produce movies depicting the simulations. Since then I became a practitioner and advocate of high performance computation and graphics, including cluster and distributed computing. During a sabbatical at NASA Langley Research Center (LaRC), development needs at LaRC directed me into data mining research, aimed at providing software tools for knowledge discovery from massive distributed data sets to be used in an immersive environment for planning planetary missions.

Currently, I am also co-PI of SPHERE (http://sphere.pcs.cnu.edu/ ), a NASA funded program that involves students in NASA Research, where I supervise students in data mining research and software development and testing. A sample of my current and previous research is listed below.

Current Research

·         Computational Drug Design and Protein Functional Modeling]

Model proteins using the Sybil software suite, starting from their amino-acid sequence to their 3-dimensional structure by sequence and structure homology, gap filling, side-chain addition and energy minimization, virtual screening of ligands, determination of conserved domains and active sites, and scored docking of ligands with simultaneous visualization of molecules and binding sites, their VDW and Connolly surfaces, colored by lipophylic and electrostatic potentials. Multiple sequence alignments and conserved/functional domain identification with NCBS software (CDTree, Cn3D, BLAST) and EBI software (Kalign, Clustal, MAFFT, etc) and corresponding databases.

·         Algorithms and Software for Virtual Screening of Compounds and Genetic Classification of Species from Primer Data

Fingerprints reflecting the properties and characteristics of the data are clustered with highly effective algorithms. One algorithm can potentially be orders of magnitude faster, since its computational complexity has a factor log(N) instead of a factor N in traditional methods. Additionally, it eliminates the memory limitations for large scale data sets that traditional methods have, since it does not require all the compounds loaded in main memory.

·         Algorithms For Clustering Heterogeneously and Homogeneously Distributed Data Sets

        Distributed data sets are clustered without sending all data to a central location, which is often unfeasible due to the storage and network requirements it would create. Further, they are clustered considering the reliability and accuracy of the distributed data sources (instruments), using new dissimilarity measures  obtained as weighted combinations of local dissimilarity measures. Each one of the local measures can be tailored to a specific data source, as long as it satisfies the triangle inequality. Examples of such local measures are any distance, and the “sine” dissimilarity measure designed and analyzed for high- and infinite dimensional data in inner product spaces.

Other Research

·         Computational phase transitions and interface evolution using a phase field approach.

Unified approximation of different multi-scale physical phenomena like: faceted and dendritic crystal growth, coarsening, instabilities, motion by mean curvature, traveling waves and approximation of sharp interface problems. Interface determined as a level set, eliminating the need to track or impose interface conditions requiring the interface normal or curvature. Identification of physical parameters in equations. Study of different phase function double well forms in the free energy functional, in particular to let interface width be a free parameter without capillary length constraints.

·         Finite speed diffusion and reaction-diffusion problems.

Efficient methods to compute the solution and track the interfaces, even in the case of stationary interfaces and front formation which required developing schemes free of the Gibbs oscillatory effect. Both, nonlinear semi-group theory and reformulation in Lagrangean coordinates approaches were applied. Convergence, stability and existence results were obtained, including in (the non-reflexive) L¹ space. Application to the dynamics of  populations that  avoid crowding, and to porous media flow

·         Selection of optimal subsets of observation variables.

Developed numerically stable parallelizable algorithms for the stepwise forward selection and backward discarding of actual observation variables (instead of linear combinations as obtained using Principal Components).

Publications 

1.       D. Smith, P.N. Tosso, J.C. Hall and E.A. Socolovsky, ”Use of Computer-Assisted Drug Design Tools to Identify Binding Sites of Protein D/E and Structural Regions Involved in Sperm/Egg Fusion”, Poster 904, 47^th Annual Meeting of the American Society for Cell Biology, Dec. 2007

2.       E. A. Socolovsky, “Clustering Heterogeneously Distributed Data”, Proceedings of the Clustering and Classification Conference, Joint Annual Meeting of the Interface and the Classification Society of North America, Washington University School of Medicine, St. Louis, MO, June 8-12, 2005

3.       E. A. Socolovsky, "A Dissimilarity Measure for Clustering High- and Infinite Dimensional Data that Satisfies The Triangle Inequality”, ICASE I Rep # 43, NASA/CR-2002-212136, Dec 2002.

4.       G. Caginalp and E. A. Socolovsky, "Phase Field Computations of Single Needle Crystals, Crystal Growth and Motion by Mean Curvature", SIAM J. Sci. Comp. 15(1994)106-126

5.       G. Caginalp and E. A. Socolovsky, "A Unified Computational Approach to Phase Boundaries by Spreading: Single Needle, Crystal Growth and Motion by Mean Curvature", in "Free Boundary Problems Involving Solids", J. Chaddam and H. Rasmussen (eds.), Pitman Res. Notes in Math. 281, Longman Sci., 1993

6.       G. Caginalp and E. A. Socolovsky, "Computation of Sharp Phase Boundaries by Spreading: The Planar and Spherically Symmetric Cases", J. Comp. Physics 95,85-100, 1991      

7.       G. Caginalp and E. A. Socolovsky, "Efficient Computation of a Sharp Interfaces by Spreading Via Phase Field Methods", Appl. Math. Lett. 2(1989),117-120      

8.       E. A. Socolovsky, "Lagrangian Non-oscillatory and F.E.M. Schemes for the porous Media Equation", Computers and Mathematics with Applications 15(1988),611-617      

9.       E. A. Socolovsky, "On the Numerical Approximation of Finite Speed Diffusion Problems", Numerische Mathematik 53(1988),97-105

10.    E. A. Socolovsky, "Difference Schemes for Degenerate Parabolic Equations", Mathematics of Computation, 47(1986), 411-420.

11.    R. C. MacCamy and E. A. Socolovsky, "Numerical Procedures for the Porous Media Equation", Computers and Mathematics with Applications, 11(1985), 315-319.

1.       M. E. Gurtin, R. C. MacCamy and E. A. Socolovsky, "A Coordinate Transformation for the Porous Media Equation that Renders the Free Boundary Stationary", Quarterly of Applied Mathematics, 42(1984), 345-357.

2.       G. K. Leaf, E. A. Socolovsky, "Analysis of the Asymptotic Behavior of the Linearized Stagnation Flow Equations of Kuramoto-Sivashinsky Type", in Proc. of the Focused Research Program on Spectral Theory and Boundary Value Problems, ANL-87-26, vol 2, pp 167-192, H. G. Kaper, M. K. Kwong and A. Zettl (eds.), Argonne National Laboratory, Illinois (1988).

3.       E. A. Socolovsky, "A Finite Element Procedure for the Porous Media Problem", in Advances in Computer Methods for PDE’s - V, R. Vichnevetsky and R. S. Stepleman Eds, (1984), 130-133

4.       E. A. Socolovsky, "Computation of the Linearity of an Amplifier", Computing Center, National Atomic Energy Commission, Buenos Aires, Argentina, 1979.

5.       E. A. Socolovsky, "Fast Poisson Solvers in the Numerical Solution of Elliptic Problems", in Numerical Methods in Continuum Mechanics, G. Marshall (editor), EUDEBA (1978).

6.       E. A. Socolovsky et al., "Selection of Optimal Subsets of Observation Variables", VII Latin American Congress of Meteorology, Santiago, Chile (1976).