Holography, machine learning, and stochastic thermodynamics.

  • Inverse digital holography using conditional generative models

    While holograms constitute powerful optical instruments, it can be difficult to find a suitable hologram for a given task. For instance, in holographic optical tweezers, holograms are used to shape laser light that is then used to exert forces on microscopic particles. While there are numerous algorithms for finding holograms, few are geared towards producing small laser patterns with high accuracy. Here, I explore conditional generative models for phase-retrival in digital holography. Read the preprint.

    ...   ...
  • Faster-than-free-diffusion reaction rates - Optimising Kramers problem

    Kramers' theory provides a comprehensive framework for thermally-activated reactions. Crucially, it explains the Arrhenius-like scaling of reaction rates in the limit of high activation-energy barriers. Following theoretical predictions by my coauthors, I show experimentally that for certain barrier shapes, this scaling law is inverted. In other words, reaction rates may actually increase with increasing barrier height. In my experiments, I construct fine-tuned barriers using an automated holographic optical tweezers setup and measure the mean first-passage time of escape. Read the paper... or a commentary (open access).

  • Fick-Jacobs theory and entropic transport in microfluidic systems

    Microfluidic channels that vary in width provide a simple toy model for entropically-driven transport. Entropic forces cause Brownian transitions across constrictions to be shorter than across straight parts of a channel. Crucially, in microfluidic channels, the hydrodynamic friction a particle experiences depends on its proximity to obstructions. When this effect is accounted for, mean first-passage times adhere closely to Fick-Jacobs predictions, which are based on a free-energy that is marginalised along the lateral direction. More coming soon...

  • Experimental evidence of symmetry breaking of transition-path times

    Various first-passage times adhere to a fundamental inversion symmetry, which is related to microscopic reversibility. For instance, the mean time of escape from a 1D interval [a,b] when initialised in the centre, x0=(a+b)/2, is equivalent at both exits: a and b. Crucially, this remains so under the influence of a (constant) drift-force f(x)=f0. In our paper, we systematically explore this symmetry and demonstrate how it can break down on the meso- and molecular scale. We furthermore provide examples how this might be useful for analyzing molecular processes. Read the paper (open access).

  • Broken detailed balance and non-equilibrium dynamics in living systems

    Living systems operate far from thermodynamic equilibrium. Enzymatic activity can induce broken detailed balance at the molecular scale. This breakdown of detailed balance is crucial to achieve biological functions such as high-fidelity transcription and translation, sensing, adaptation, biochemical patterning, and force generation. While biological systems such as motor enzymes violate detailed balance at the molecular scale, it remains unclear how non-equilibrium dynamics [...] Read the review article (open access).

  • Broken detailed balance of filament dynamics in active networks

    Myosin motor proteins drive vigorous steady-state fluctuations in the actin cytoskeleton of cells. Endogenous embedded semiflexible filaments such as microtubules, or added filaments such as single-walled carbon nanotubes are used as novel tools to noninvasively track equilibrium and nonequilibrium fluctuations in such biopolymer networks. Here, we analytically calculate shape fluctuations of semiflexible probe filaments in a viscoelastic environment, driven out of equilibrium by motor activity. [...] Read the first ...
    ... or a more in-depth paper.

About Me

Jannes Gladrow

Research engineer at Microsoft Research Cambridge.

My Career

Microsoft Research Cambridge, UK

Research Engineer

Sep 2020
Software Development/Machine Learning

Microsoft Research Cambridge, UK

AI Resident

Sep 2019
Software Development/Machine Learning

University of Cambridge

PhD Studies (stochastic thermodynamics, optical tweezers, and machine learning)

Oct 2015 - Sep 2019
PhD Student

nanoTemper Technologies, Munich

R&D intership (rapid prototyping of measurement devices for protein characterisation)

Jun - Sep 2015
R&D Intern

Georg-August University Göttingen

M.Sc. in physics (minor in computer science). Thesis in non-equilibrium thermodynamics with Christoph Schmidt.

Sep 2013 - May 2015
Student and Tutor (quantum- and statistical mechanics)

École Normale Supérieure, Paris

Exchange year (M1, ICFP). Includes a 6-month lab programming internship (java, imageJ) at Institut Curie.

Sep 2012 - Jul 2013

Penn-State University, State College, PA, USA

Developed LEED correction algorithms in python.

Aug - Oct 2011
Research intern

Georg-August University Göttingen

B.Sc. in physics (minor in computer science). Thesis in theoretical neuroscience with Marc Timme.

Oct 2009 - Mar 2012

My Skills (self-assessment)