Alexander Patapoff

Computational physicist with 3+ years experience in theoretical and applied molecular dynamics with a focus on in-silico early stage drug discovery. Designed and maintained several impactful APIs, scientific software programs, and CUDA accelerated engines from first principles across wide ranging applications: intermolecular calculations, mass spec data analysis, python API wrappers...etc. High value team player with experience coordinating in-silico needs of several independent project teams through research, design, and implementation of unique solutions in a scalable manner.

Non-Hermitian gap closure and delocalization in interacting directed polymers

Abhijeet Melkani, Alexander Patapoff, and Jayson Paulose

We study a classical model of thermally fluctuating polymers confined to two dimensions, experiencing a grooved periodic potential, and subject to pulling forces both along and transverse to the grooves. The equilibrium polymer conformations are described by a mapping to a quantum system with a non-Hermitian Hamiltonian and with fermionic statistics generated by noncrossing interactions among polymers. Using molecular dynamics simulations and analytical calculations, we identify a localized and a delocalized phase of the polymer conformations, separated by a delocalization transition which corresponds (in the quantum description) to the breakdown of a band insulator when driven by an imaginary vector potential. We calculate the average tilt of the many-body system, at arbitrary shear values and filling density of polymer chains, in terms of the complex-valued non-Hermitian band structure. We find the critical shear value, the localization length, and the critical exponent by which the shear modulus diverges in terms of the branch points (exceptional points) in the band structure at which the bandgap closes. We also investigate the combined effects of non-Hermitian delocalization and localization due to both periodicity and disorder, uncovering preliminary evidence that while disorder favors localization at high values, it encourages delocalization at lower values.