Postdoc Position at the University of California, Merced

[Noemi Petra]

Hi everyone,

There is an immediate opening for a postdoc position in my

research group in the School of Natural Sciences at the

University of California, Merced. The postdoctoral researcher

will work under an NSF-funded Collaborative Research (with UC

Merced, UT Austin and MIT) entitled: Integrating Data with

Complex Predictive Models under Uncertainty: An Extensible

Software Framework for Large-Scale Bayesian Inversion (see

https://hippylib.github.io for a general overview of the

project).

This project involves research in the field of large-scale

Bayesian inverse problems, and in particular on implementation

work on adding new features in

hIPPYlib (https://hippylib.github.io). hIPPYlib contains

state-of-the-art scalable adjoint-based algorithms for PDE-based

deterministic and Bayesian inverse problems. It builds on FEniCS

for the discretization of the PDEs, hence all the code

development tasks are tight to being able to work with FEniCS and

on python coding experience. There will be occasional C++

implementation needs as well, hence experience with C++ is a

plus.

The postdoc will contribute to the research dissemination and

will also help build the user/developer community by attending

and speaking at conferences, workshops and summer schools at

local and international events.

Interested candidates should contact Noemi Petra at

npetra at ucmerced.edu and apply at:

https://aprecruit.ucmerced.edu/apply/JPF00505.

Please feel free to forward this announcement as appropriate.

Thanks very much,

Noemi

-- 
Noemi Petra, PhD

Assistant Professor of Applied Mathematics
Secretary of the SIAM UQ Activity Group
SIAM Student Chapter Faculty Advisor
University of California, Merced
http://faculty.ucmerced.edu/npetra/

[Johan Hoffman]

Hi Noemi, 

I saw your announcement and found it very interesting, since I also want to combine PDE modelling and optimisation. 

I come from the other direction, being a co-founder of FEniCS in 2003 and since then focused on the development of scalable implementations and adaptive algorithms with a particular focus on turbulent flow simulation. Recently I have taken an interest in PDE constrained optimisation and UQ. 

For the last couple of years our group has developed the HPC branch FEniCS-HPC for massively parallel systems, based on the Dolfin-HPC branch, but still based on the FEniCS form language UFL and PETSc. 

https://link.springer.com/chapter/10.1007/978-3-319-32149-3_34

https://bitbucket.org/fenics-hpc/dolfin-hpc

How is hIPPYlib coupled to FEniCS? Do you think it would be straight forward to extend the coupling to FEniCS-HPC? 

Our purpose with FEniCS-HPC is to target the most advanced supercomputers available, and make the technology available open source, and also through cloud services to non-experts, e.g. through the MSO4SC project: 

http://mso4sc.eu

It would be interesting to find some common interests in the area. With respect to applications of the methods I am involved in general CFD, biomechanics and geosciences. 

All the best,

Johan

----------------------------------------------
Johan Hoffman
Professor
KTH Royal Institute of Technology
Stockholm, Sweden
jhof...@kth.se

--
You received this message because you are subscribed to the Google Groups "fenics-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fenics-dev+...@googlegroups.com.
To post to this group, send email to fenic...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.