Source author record

James A. Glazier

James A. Glazier appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

cond-mat.soft Machine Learning Cell Behavior cond-mat.dis-nn Mathematical Software physics.comp-ph Subcellular Processes Tissues and Organs

Catalog footprint

What is connected

7works

8topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2021arXiv

Deep learning approaches to surrogates for solving the diffusion equation for mechanistic real-world simulations

In many mechanistic medical, biological, physical and engineered spatiotemporal dynamic models the numerical solution of partial differential equations (PDEs) can make simulations impractically slow. Biological models require the simultaneous calculation of the spatial variation of concentration of dozens of diffusing chemical species. Machine learning surrogates, neural networks trained to provide approximate solutions to such complicated numerical problems, can often provide speed-ups of several orders of magnitude compared to direct calculation. PDE surrogates enable use of larger models than are possible with direct calculation and can make including such simulations in real-time or near-real time workflows practical. Creating a surrogate requires running the direct calculation tens of thousands of times to generate training data and then training the neural network, both of which are computationally expensive. We use a Convolutional Neural Network to approximate the stationary solution to the diffusion equation in the case of two equal-diameter, circular, constant-value sources located at random positions in a two-dimensional square domain with absorbing boundary conditions. To improve convergence during training, we apply a training approach that uses roll-back to reject stochastic changes to the network that increase the loss function. The trained neural network approximation is about 1e3 times faster than the direct calculation for individual replicas. Because different applications will have different criteria for acceptable approximation accuracy, we discuss a variety of loss functions and accuracy estimators that can help select the best network for a particular application.

preprint2021arXiv

Using Deep LSD to build operators in GANs latent space with meaning in real space

Generative models rely on the key idea that data can be represented in terms of latent variables which are uncorrelated by definition. Lack of correlation is important because it suggests that the latent space manifold is simpler to understand and manipulate. Generative models are widely used in deep learning, e.g., variational autoencoders (VAEs) and generative adversarial networks (GANs). Here we propose a method to build a set of linearly independent vectors in the latent space of a GANs, which we call quasi-eigenvectors. These quasi-eigenvectors have two key properties: i) They span all the latent space, ii) A set of these quasi-eigenvectors map to each of the labeled features one-on-one. We show that in the case of the MNIST, while the number of dimensions in latent space is large by construction, 98% of the data in real space map to a sub-domain of latent space of dimensionality equal to the number of labels. We then show how the quasi-eigenvalues can be used for Latent Spectral Decomposition (LSD), which has applications in denoising images and for performing matrix operations in latent space that map to feature transformations in real space. We show how this method provides insight into the latent space topology. The key point is that the set of quasi-eigenvectors form a basis set in latent space and each direction corresponds to a feature in real space.

preprint2015arXiv

3D simulations of wet foam coarsening evidence a self similar growth regime

In wet liquid foams, slow diffusion of gas through bubble walls changes bubble pressure, volume and wall curvature. Large bubbles grow at the expenses of smaller ones. The smaller the bubble, the faster it shrinks. As the number of bubbles in a given volume decreases in time, the average bubble size increases: i.e. the foam coarsens. During coarsening, bubbles also move relative to each other, changing bubble topology and shape, while liquid moves within the regions separating the bubbles. Analyzing the combined effects of these mechanisms requires examining a volume with enough bubbles to provide appropriate statistics throughout coarsening. Using a Cellular Potts model, we simulate these mechanisms during the evolution of three-dimensional foams with wetnesses of $ϕ=0.00$, $0.05$ and $ 0.20$. We represent the liquid phase as an ensemble of many small fluid particles, which allows us to monitor liquid flow in the region between bubbles. The simulations begin with $2 \times 10^5$ bubbles for $ϕ= 0.00$ and $1.25 \times 10^5$ bubbles for $ϕ= 0.05$ and $0.20$, allowing us to track the distribution functions for bubble size, topology and growth rate over two and a half decades of volume change. All simulations eventually reach a self-similar growth regime, with the distribution functions time independent and the number of bubbles decreasing with time as a power law whose exponent depends on the wetness.

preprint2015arXiv

libRoadRunner: A High Performance SBML Simulation and Analysis Library

This paper presents libRoadRunner, an extensible, high-performance, cross-platform, open-source software library for the simulation and analysis of models \ expressed using Systems Biology Markup Language (SBML). SBML is the most widely used standard for representing dynamic networks, especially biochemical networks. libRoadRunner supports solution of both large models and multiple replicas of a single model on desktop, mobile and cluster computers. libRoadRunner is a self-contained library, able to run both as a component inside other tools via its C++ and C bindings andnteractively through its Python interface. The Python Application Programming Interface (API) is similar to the APIs of Matlab and SciPy, making it fast and easy to learn, even for new users. libRoadRunner uses a custom Just-In-Time (JIT) compiler built on the widely-used LLVM JIT compiler framework to compile SBML-specified models directly into very fast native machine code for a variety of processors, making it appropriate for solving very large models or multiple replicas of smaller models. libRoadRunner is flexible, supporting the bulk of the SBML specification (except for delay and nonlinear algebraic equations) and several of its extensions. It offers multiple deterministic and stochastic integrators, as well as tools for steady-state, stability analyses and flux balance analysis. We regularly update libRoadRunner binary distributions for Mac OS X, Linux and Windows and license them under Apache License Version 2.0. http://www.libroadrunner.org provides online documentation, full build instructions, binaries and a git source repository.

preprint2010arXiv

Adhesion between cells, diffusion of growth factors, and elasticity of the AER produce the paddle shape of the chick limb

This paper has been withdrawn

preprint2007arXiv

Experimental growth law for bubbles in a "wet" 3D liquid foam

We used X-ray tomography to characterize the geometry of all bubbles in a liquid foam of average liquid fraction $ϕ_l\approx 17 %$ and to follow their evolution, measuring the normalized growth rate $\mathcal{G}=V^{-{1/3}}\frac{dV} {dt}$ for 7000 bubbles. While $\mathcal{G}$ does not depend only on the number of faces of a bubble, its average over $f-$faced bubbles scales as $G_f\sim f-f_0$ for large $f$s at all times. We discuss the dispersion of $\mathcal{G}$ and the influence of $V$ on $\mathcal{G}$.

preprint2003arXiv

Dynamics and Topological Aspects of a Reconstructed Two-Dimensional Foam Time Series Using Potts Model on a Pinned Lattice

We discuss a method to reconstruct an approximate two-dimensional foam structure from an incomplete image using the extended Potts mode with a pinned lattice we introduced in a previous paper. The initial information consists of the positions of the vertices only. We locate the centers of the bubbles using the Euclidean distance-map construction and assign at each vertex position a continuous pinning field with a potential falling off as $1/r$. We nucleate a bubble at each center using the extended Potts model and let the structure evolve under the constraint of scaled target areas until the bubbles contact each other. The target area constraint and pinning centers prevent further coarsening. We then turn the area constraint off and let the edges relax to a minimum energy configuration. The result is a reconstructed structure very close to the simulation. We repeated this procedure for various stages of the coarsening of the same simulated foam and investigated the simulation and reconstruction dynamics, topology and area distribution, finding that they agree to good accuracy.