Statewide Users Group Conference Minutes - October 4, 2018

Wednesday, October 3rd

4:00 - 6:00 pm

Allocations Committee (members only)

6:00 - 7:30 pm

SUG Executive Meeting (members only)                           

Thursday, October 4th 

9:00 - 10:00 am

 

 



10:00 - 11:00 am


 




 

11:00 - 11:45 am


11:45 am - 12:00 pm


12:00 - 12:55 pm

 



12:55 - 1:00 pm


1:00 - 1:50 pm

 

2:00 - 2:50 pm

 

2:50 - 3:00 pm

 

3:00 - 4:45 pm

 

 

4:45 pm

Breakfast Assortment

Hardware and Operations Committee Meeting (non-members welcome)

Software and Activities Committee Meeting (non-members welcome)



Breakout Sessions (determined by which you signed up for)

1. Pitzer Information Session

2. OSC Campus Champions

3. Data Analytics

4. Client Portal


OSC: Welcome and Presentation


Lunch Pick-up


Keynote Address:
Russ Fromkin, Intel
Americas HPC and HPDA Sales Director

 

Break


Flash Talks: Session 1

 

Flash Talks: Session 2

 

Break

 

Poster Session

Networking and Hors D'oeuvres

 

Poster and Flash Talk Winner Announcement

 

Keynote Address: 
Intel, Ross Fromkin, Americas HPC and HPDA Sales Director

Russ Fromkin is the Americas HPC and HPDA Sales Director for the Intel Corporation. Russ and his team are responsible for helping technical computing and data analytics customers implement cutting edge technologies developed by Intel Corporation. Russ joined Intel Corporation in 1999 as part of an acquisition of Dialogic Corporation. Previous to his current role, Russ helped lead Information Assurance, HPC and Cloud Computing activities for the Intel Federal team. Prior to the acquisition, Russ developed and managed the ISV program for Dialogic. Russ holds a BA from The Johns Hopkins University.

Flash Talk Winner

Dylan Wood, Graduate Research Associate at The Ohio State University
"Modeling of Failure Mechanisms for Flood Control Systems in a Hydrodynamic Storm Surge Model"Risk-informed decision making plays a critical role in mitigation of loss of life and assets during a catastrophic event such as the land fall of a hurricane or tropical cyclone. In this case, many factors, such as proximity to coastline and resilience of structures, influence the probable degree of risk posed to coastal regions by storm surge, though perhaps the most relevant and poorly understood factor is the ability of flood control systems to withstand forces attributed to storm surge and to protect coastal regions from inundation. We present on the development and application of techniques for resolving and assessing the risk of failure for flood defense structures within a large scale hydrodynamic storm surge model for forecasting risks posed by tropical cyclone impacts.

Flash Talk Runner-Up

Megan Smith, Graduate Fellow at The Ohio State University
"Disentangling the Process of Speciation Using Machine Learning"
Historically, investigations into the processes driving speciation have largely been isolated from systematic investigations into species limits. Recent advances in sequencing technology have led to a rapid increase in the availability of genomic data, and this, in turn, has led to the introduction of many novel methods for species delimitation. However, these methods have been limited to divergence-only scenarios and have not attempted to evaluate complex modes of speciation, such as those that include gene flow during early stages of divergence (sympatric speciation) or population size changes (founder effect speciation). This shortcoming is primarily the result of the computational demands of evaluating such models. To address this, we introduce delimitR, an R-package that enables biologists to infer species boundaries and evaluate the demographic processes that may have led to speciation. delimitR uses the binned multidimensional Site Frequency Spectrum and a machine-learning algorithm (Random Forests) to compare speciation models while minimizing the computational resources required. We use simulations to evaluate the accuracy of delimitR. When comparing models that include lineage divergence and gene flow for three populations, error rates are near zero with recent divergence times (<100,000 generations) and a modest number of Single Nucleotide Polymorphisms (SNPs; 1,500). When applied to a more complex model set (including divergence, gene flow, and population size changes), error rates are moderate (~0.15 with 10,000 SNPs), and misclassifications are generally between highly similar models.

Poster Winner

Changlong Zou, Graduate Research Associate at The Ohio State University
"Computational Studies of Zeolite Nanosheets as Pervaporation Membranes for Ethanol Extraction"
Pervaporation is a promising approach for anhydrous ethanol extraction from dilute biomass fermentation broth.  Using siliceous zeolite membranes can offer outstanding separation factors but relatively low fluxes. Zeolite nanosheets, synthesized with a thickness of a few nanometers, may overcome this limitation because of their intrinsically short diffusion paths. However, the potential of zeolite nanosheets as pervaporation membranes for ethanol extraction remains unknown. In this study, by employing molecular dynamics techniques, we demonstrated the outstanding performance of zeolite nanosheets for ethanol extraction. The ethanol flux of zeolite nanosheets was found to be orders of magnitude higher than currently available membranes reported in the literature, while a high separation factor can be still achieved. We have also discovered the critical role of nanosheet surfaces in the separation performance. The outcomes of this work are anticipated to be instrumental to the future development of zeolite nanosheet membranes.

Poster Runner-Up

Heather Hansen, Graduate Teaching Associate at The Ohio State University
"Using the Ohio Supercomputer Cluster to Measure Developmental Changes in Connectivity between the Amygdala Subnuclei and Occipit"Previous work in macaques has shown that connectivity from the lateral and basal amygdalar subnuclei to the occipitotemporal cortex decreases on a gradient from anterior to posterior, and refines across development. Can we delineate this connectivity in humans, and determine what functional regions might be driving it? Using Diffusion Weighted Imaging data, we ran probabilistic tractography with FSL between the amygdala subnuclei as seeds and the occipitotemporal cortical parcellations as targets. Given that connectivity between 170 parcellations was computed for each of 45 subjects, the use of parallel computing through the Ohio Supercomputer Center was essential. Results showed that the development of human amygdalar connectivity to the occipitotemporal cortex resembles that of macaques, and is especially driven by functional parcels such as PPA, PFS, LO, and OFA.

SUG Press Release

Please see our press release.