Syllabus

Workshops are Mondays at 2 p.m. in PSF 372 (unless otherwise noted). Laptops are strongly recommended for live coding in all sessions unless otherwise noted. Some talks will be recorded and the videos are posted in the Becksteinlab YouTube Channel .

The workshop series is organized by Kathleen Clark; contact her with questions or use the comment section below.

* indicates a guest speaker

** indicates a program change

Week 1

11 June | Ian Kenney | Using a Queuing System and Supercomputers

An interactive how-to session on using a queuing system to increase productivity including how to run and suspend jobs. Will also cover the basics of supercomputer usage. (video)

Week 2

18 June | Taylor Colburn | Bash: Scripting, the Terminal, and Text Editors

A brief history of the Bourne-again shell (Bash), followed by an overview of it’s strengths and weaknesses, and a survey of useful commands. Specific examples will be explored with some degree of depth, giving the audience exposure to piping, grep, sed, find, xargs, awk, cut, and more. (video)

Week 3

25 June | Sean Seyler* | Hybrid Micro/Mesoscale Simulations

Numerical simulation methods must carefully balance computational cost in the context of the spatiotemporal scales over which the phenomena of interest occur. However, so-called mesoscale phenomena—from cytoplasmic streaming in cells to multi-species particle transport in colloidal solutions—often exhibit nonequilibrium and noisy dynamics spanning a very broad spectrum of scales, which challenges both particle-based and continuum-based simulation approaches: generally speaking, whereas particle methods can achieve high spatiotemporal resolution, continuum methods can access much larger and longer scales much more cheaply (computationally) at the cost of detail. Hybrid simulation (e.g., the particle-continuum method) and other mesoscale approaches balance computational cost by using higher resolution only where it is needed. We will discuss the challenges involved in numerically modeling mesoscale systems, some of the strategies for designing effective mesoscale simulation, and, more generally, how such efforts may eventually reconcile apparent conceptual differences between particle and continuum perspectives. (video)

Week 4

2 July | David Dotson* | MDAnalysis and datreant: How to analyze large sets of simulations productively

How to organize heterogeneous collections of data such as MD trajectories and analyze them in a scalable (and possibly parallel) fashion – using open source software tools datreant and MDAnalysis. This hands-on workshop will introduce datreant and the MDAnalysis library; uses Python. (video) (additional video)

Week 5

9 July | Kathleen Clark | Parallelizing with dask

In this live coding session, I will show how to use dask to parallelize work to increase productivity. By using workers and schedulers, many jobs can be run in parallel from notebooks or a terminal for better performance. We will also discuss a few libraries (e.g. daskml, pmda) which utilize dask. (video)

Week 6

16 July | Abhishek Singharoy* | Theoretical Concerns with Molecular Dynamics (no laptop needed)

MD simulations, even with the largest supercomputers of the day, are seriously limited by inadequacies in the classical force fields, choice of simulation boundary conditions, and most importantly, ergodic sampling of slower collective modes. In this workshop some of the practical considerations to overcome these issues will be discussed. (video)

Week 7

23 July | Ian Welland* | Non-equilibrium statistical mechanics (no laptop needed)

Non-equilibrium systems pose a variety of theoretical modeling challenges, and constitute many of the most studied systems in science and engineering. Here we examine the basic and established methods of the subject, along with a survey of applications and more advanced methods and problems. Since the topic is vast, the approach will be more discussion oriented, with the intention of focusing on applications or methods of interest to the audience. (Ian did not wish to be recorded.)

Week 8

30 July (Special Time: 11 a.m.) ** | Julio Candanedo | Introduction to stochastic equations (no laptop needed)

A stochastic process is one that includes an element of randomness. We will discuss the equations that are used to describe these processes, as well as how to use them. (video)

Week 9

6 August | Oliver Beckstein | Good software engineering practices

I will introduce tools and ideas that are common in professional software development but not widely known in the scientific community. By using version control (`git`), tests (often automated with continuous integration services), and documentation generators, scientists can be more productive, improve reproducibility and replicability, and generally treat code with the same kind of professionalism that is applied to theory and experimentation. I will also talk about the importance of licenses and how to make open source work for you. (video)

slides

miniworkshop_Beckstein_2018.pdf (GitHub) or slides (PDF)

example files repository

github.com/Becksteinlab/workshop_testing

Week 10

13 August ** | Shujie Fan | Machine learning and Bayesian statistics

I will give an introduction to machine learning by discussing various algorithms that are commonly used. Additionally, we will discuss Bayesian statistics: what it is and how to use it. (video)

The physicists' cow. Image from: https://en.wikipedia.org/wiki/Spherical_cow