Holography · String theory · Numerical methods

Veselin Filev

  • Associate Professor, Department of Algebra and Logic, Institute of Mathematics and Informatics, Bulgarian Academy of Sciences
  • Adjunct Associate Professor, American University in Bulgaria
  • Research Associate, School of Theoretical Physics, Dublin Institute for Advanced Studies

I study holography and string theory, and the numerical methods, from lattice Monte Carlo to neural networks, that put holography to the test.

Publications INSPIRE-HEP arXiv Contact
Anti-de Sitter space as a tiled Poincaré disk. Move the pointer: the arc is the minimal (Ryu–Takayanagi) curve for the highlighted boundary interval, and its length gives the interval’s entanglement entropy.
About

Physics at strong coupling

Portrait of Veselin Filev

I am a theoretical physicist at the Institute of Mathematics and Informatics of the Bulgarian Academy of Sciences, where I have been Associate Professor since 2021. I also teach mathematics and physics at the American University in Bulgaria.

My research is in theoretical high-energy physics, in particular superstring theory and the holographic (AdS/CFT) correspondence. Much of it concerns flavour: adding quarks to holographic gauge theories through probe and backreacted D-branes, and following how external fields, temperature and finite size shape the resulting phases. I have used lattice Monte Carlo simulations of matrix models as non-perturbative tests of gauge/gravity duality, and most recently neural networks as variational solvers for holographic problems.

Before moving to Sofia in 2017 I was a postdoctoral fellow at the Dublin Institute for Advanced Studies and at the Max Planck Institute for Physics in Munich. I did my PhD at the University of Southern California with Clifford V. Johnson.

Career

  1. 2021 – nowAssociate ProfessorIMI-BAS, Department of Algebra and Logic
  2. 2021 – nowAdjunct Associate ProfessorAmerican University in Bulgaria
  3. 2016 – nowResearch AssociateSchool of Theoretical Physics, DIAS
  4. 2017 – 2021Assistant, then Chief Assistant ProfessorIMI-BAS
  5. 2012 – 2016Postdoctoral FellowSchool of Theoretical Physics, DIAS
  6. 2010 – 2012Marie Curie / IRCSET Postdoctoral FellowMax Planck Institute for Physics, Munich
  7. 2008 – 2010IRCSET Postdoctoral FellowSchool of Theoretical Physics, DIAS
  8. 2008PhD in PhysicsUniversity of Southern California
  9. 2003MSc in Theoretical and Mathematical PhysicsSofia University
  10. 2002BSc in PhysicsSofia University
Research

Research themes

Five lines of work, from semiclassical strings in 2003 to neural networks today. Each theme lists its most cited papers; “All papers” opens the full list filtered to that theme.

Current focus

Neural networks for holography

2025 –

Neural networks used as variational solvers. A network parametrises a probe-brane embedding or a Ryu–Takayanagi surface, and the action or the area is minimised directly, with no equations of motion to derive. The same idea runs backwards to reconstruct bulk geometry from boundary data: entanglement entropy together with Wilson loops recovers the metric.

    Try it: a neural network learns a minimal surface

    A small network (one hidden layer, 10 units) bends a curve between two boundary points of AdS₃. Its loss is the hyperbolic length, and gradient descent pulls it onto the exact geodesic, shown in grey. Drag the endpoints to set a new problem.

    Training step0
    Network length–
    Exact length–
    Relative error–

    Holographic flavour dynamics

    2006 – 2019

    Probe D-branes add fundamental matter (quarks) to holographic gauge theories. This work maps how external magnetic and electric fields, temperature, compact spaces and defects reshape the phase structure: magnetic catalysis of chiral symmetry breaking, meson melting, quantum critical points, and topology-changing transitions.

      Matrix models & numerical holography

      2012 – 2018

      Lattice Monte Carlo simulations of matrix quantum mechanics, namely the BFSS and BMN models and their flavoured versions, used as non-perturbative tests of gauge/gravity duality. Alongside them, analytic studies of commuting and multi-matrix models at general coupling.

        Backreacted flavour

        2011 – 2022

        Going beyond the probe limit: supergravity backgrounds that include the backreaction of many flavour branes. These give unquenched holographic models of magnetic catalysis, an anisotropic quark–gluon plasma, non-commutative theories and the D0/D4 system.

          Spinning strings

          2003 – 2005

          Semiclassical quantisation of rotating strings and the operators with large quantum numbers they describe, including giant gravitons. This is where my work on AdS/CFT began.

            Publications

            Papers and preprints

            Every paper links to arXiv, the journal and its INSPIRE-HEP record. Full profiles: INSPIRE-HEP · arXiv · Google Scholar.

            Papers–2003 – 2026
            Citations–INSPIRE-HEP
            h-index–INSPIRE-HEP

            Papers by year of first appearance

            Each dot is one paper, coloured by research theme. Hover or tap a dot for details, and click it to jump to the paper. Use the theme buttons to filter.

              Talks

              Talks and lectures

              Invited talks, lecture courses and seminars since 2019. Earlier talks, from 2005 to 2018, are listed at the end.

              Show earlier talks, 2005 – 2018 ()
              Teaching

              Courses at the American University in Bulgaria

              Courses I teach across physics, mathematics, statistics and quantitative finance.

              Physics

              PHY 110Physics I
              PHY 210Classical Mechanics
              PHY 310Thermodynamics and Statistical Physics

              Mathematics

              MAT 105Elementary Linear Algebra and Analytical Geometry
              MAT 1002Finite Mathematics
              MAT 1003Calculus I

              Statistics & quantitative finance

              STA 1005Statistics
              MAT 4051Mathematical Finance

              Lectures at schools and earlier teaching

              • 2026, PhD School “Advanced Topics in Contemporary Theoretical Physics” (SEENET-MTP), Sofia University: two lectures on matrix models and holography
              • 2017, CERN–SEENET-MTP PhD Training Program, Sofia: Introduction to AdS/CFT with flavours
              • 2015, CERN European School of High-Energy Physics: discussion leader
              • 2003 – 2008, University of Southern California: graduate assistant for Fundamentals of Physics I (Mechanics and Thermodynamics) and III (Optics and Modern Physics)
              Beyond fundamental physics

              Practical applications

              Beyond fundamental physics, I am interested in practical applications of mathematical and computational methods, including quantitative finance, machine learning for insurance, and agentic coding and AI.

              Quantitative finance

              Stochastic models, derivative pricing and risk, and the Monte Carlo techniques they share with lattice field theory. I also teach Mathematical Finance at AUBG.

              Machine learning for insurance

              Statistical learning for actuarial questions such as pricing, risk segmentation and claims modelling.

              Agentic coding and AI

              AI agents as collaborators in research and software engineering, and neural networks as numerical solvers in physics.

              Try it: Monte Carlo pricing of a European call option

              Simulated price paths under geometric Brownian motion (S₀ = 100, r = 3%). As paths accumulate, the Monte Carlo estimate and its 95% band close in on the Black–Scholes price.

              Monte Carlo–
              Black–Scholes–
              Paths0
              Contact

              Get in touch

              Institute of Mathematics and Informatics

              Bulgarian Academy of Sciences
              Acad. Georgi Bonchev Str., Block 8, Office 567
              1113 Sofia, Bulgaria
              Email: vfilev [at] math.bas.bg

              American University in Bulgaria

              Main campus
              1 Georgi Izmirliev Sq.
              2700 Blagoevgrad, Bulgaria
              AUBG faculty profile

              Profiles