Workshop on Basic Computing Services in the Physics Department - subMIT
Kolker Room (26-414)
The subMIT computing facility is a login pool that is designed to provide access to the basic research computing resources of the physics department and beyond.
This one day workshop will provide an overview and updates on the status and plans for the system and project, as well as topical presentations on a range of use cases.
Additionally, there will be tutorial sessions. Bring your laptop to run the examples in real-time or just sit back and listen to the accompanying instructional talk.
More information on the subMIT project as well as user documentation is available at https://submit.mit.edu
The workshop will take place in the Kolker Room and a Zoom connection will be available at https://mit.zoom.us/j/96743699673?pwd=b3h2Q3c3cVQwYW12blhMUG5SWXZCZz09
Refreshments will be provided during coffee breaks!
Please note: the details of the schedule of individual talks may change as we work to accommodate new requests for contributions.
-
-
1
Introduction/subMIT Project OverviewSpeaker: Joshua Bendavid (Massachusetts Institute of Technology)
-
2
Getting started on subMIT: Available Resources
-
Computational Resources
-
Documentation: Users Guide, GitHub Examples
-
Support: Help Desk & Chat Bot
Speaker: Zhangqier Wang -
-
3
Getting started on subMIT: How to Interact with subMIT
- ssh / terminal
- JupyterHub
- X2Go
- Visual Studio Code (Remote Development)
- 1-minute summary of batch jobs (see also later tutorial)
Speaker: Matthew Heine (Massachusetts Institute of Technology) -
4
Getting started on subMIT: Installing / Managing Software
- native environment
- installing software with conda
- software environment management tools
- containers overview
Speaker: Chad Freer (Massachusetts Institute of Technology) -
5
Navigating the recent Linux Upgrade
In this session we will outline key changes in the migration to the new version of Linux as well as how to access the older version via containers. We are also setting aside a portion of time to answer user questions about the upgrade, so please bring any questions you have, general or specific.
Speaker: Marianne Moore (MIT) -
10:20
Coffee Break
-
6
Experience thus far running a multi-experiment OpenShift Cluster at FNALSpeaker: Lindsey Gray (Fermi National Accelerator Laboratory)
-
7
Tutorial: Batch Job / Workflow Management: SLURM & HTCondor
This session will show how to use the batch schedulers/resource-managers SLURM & HTCondor to manage your computational tasks, distributing them across the shared resources. Depending on your workflow, this may provide an easy way to run your calculations in parallel, shortening time-to-result and eliminating some manual tasks.
Speaker: Chad Freer (Massachusetts Institute of Technology) -
8
Tutorial: GPUs: Introduction to Likelihood-Free Inference with PyTorch Lightning
This tutorial will give an overview of using pytorch lightning for building and training neural networks. A simple problem in inference: measuring the parameters of a line will be presented. The tutorial will also introduce distributed training with pytorch lightning
Speaker: Deep Chatterjee (Massachusetts Institute of Technology) -
12:30
Lunch
-
9
Parameter Estimation of Un-modeled Gravitational-wave Signals using Likelihood-free InferenceSpeaker: Deep Chatterjee (Massachusetts Institute of Technology)
-
10
DM21cm: a GPU-accelerated simulation of dark matter energy injection in 21cm cosmologySpeaker: Yitian Sun (Massachusetts Institute of Technology)
-
11
Embarrassingly parallel ray-tracing for the Arcus X-ray spectrometerSpeaker: Hans Guenther (Massachusetts Institute of Technology)
-
12
Employing snakemake to perform measurements with the LHCb experiment at CERNSpeaker: Blaise Delaney (Massachusetts Institute of Technology)
-
13
Tutorial: Introduction to the Snakemake workflow managerSpeaker: Blaise Delaney (Massachusetts Institute of Technology)
-
15:05
Coffee Break
-
14
CMS Analysis on subMIT with crab and dask jobqueueSpeaker: Simon Rothman (Massachusetts Institute of Technology)
-
15
Adaptive Brillouin-zone integration for optical conductivitySpeaker: Lorenzo Van Munoz (Massachusetts Institute of Technology)
-
16
Axion and high-frequency gravitational wave searches with ABRACADABRA and DMRadioSpeaker: Kaliroe Pappas (MIT laboratory for nuclear science)
-
17
Training spiking neural networks via adjoint sensitivity analysisSpeaker: Enrique Toloza (Massachusetts Institute of Technology)
-
18
ML for Science: Best PracticesSpeaker: Denis Boyda (Massachusetts Institute of Technology)
-
19
Open Discussion & Closing Remarks
-
1