The Theory Group at the MIT Plasma Science and Fusion Center seeks applicants for a postdoctoral position in machine learning applied to the development of fast surrogate models of computationally intensive radio frequency (RF) heating and current drive simulations. These surrogate models will enable enhanced real-time plasma control, better interpretation of experimental data, and efficient time-dependent integrated modeling of advanced tokamak plasmas.
The position is funded by a DoE SciDAC grant for machine learning applied to radio frequency (RF) modeling tools in fusion. Predictive capability of RF wave heating and current drive is critical not only for present-day fusion experiments but also to design and construct a fusion power plant. The “advanced tokamak” (AT) concept is a leading candidate for a steady-state fusion power plant. The AT makes use of the pressure-gradient-driven bootstrap current to sustain a majority of the required plasma current, augmented by non-inductive current drive actuators. Control of the RF heating and current drive profiles, through varying the launched power and wavenumber of the system, is one of the few direct current profile control knobs available on a tokamak. Although RF simulation tools are sophisticated and accurate, the computational resources required are considerable and widespread use in parametric scenario scoping studies, time-dependent modeling, and real-time control of experiments will benefit from reducing the time required for simulation modeling, which this project aims to accomplish by leveraging machine learning methods to develop fast surrogate models.
The postdoc will be expected to work with machine learning specialists and RF physicists in developing solutions to forward and inverse problems arising in the development of fast surrogate models for complex computer codes. Other duties may include managing training data, developing supervised learning pipelines and developing classifiers.
Requirements: a Ph.D. in applied math, computer science, physics or related discipline. The successful applicant will also have expertise in machine learning methods such as various neural network techniques and experience applying these methods to modeling of physics systems.
This position is for a one-year term with the possibility of one or two one year reappointments assuming satisfactory performance and the availability of funds.
Interested candidates may apply online at https://hr.mit.edu/careers with a CV, publication list, and statement of research interests. Please reference job number 19005. Applicants should also arrange for three letters of reference to be sent to John Wright at firstname.lastname@example.org.
MIT is an equal employment opportunity employer. All qualified applicants will receive consideration for employment and will not be discriminated against on the basis of race, color, sex, sexual orientation, gender identity, religion, disability, age, genetic information, veteran status, ancestry, or national or ethnic origin.
The National Academies of Science, Engineering, and Medicine
6 Days Ago
BACK TO TOP
Physics Today Jobs has listings for the latest assistant, associate, and full professor roles, plus scientist jobs in specialized disciplines like theoretical physics, astronomy, condensed matter, materials, applied physics, astrophysics, optics and lasers, computational physics, plasma physics, and others! Find a job here as an engineer, experimental physicist, physics faculty, postdoctoral appointee, fellow, or researcher.