I am a meticulous detail-oriented Nuclear Engineer, with an extensive mechanical background, eager to implement the state-of-the-art pattern recognition and machine learning algorithms to render feasible the repeated executions of computationally taxing engineering analyses such as Uncertainty Quantification (UQ), Sensitivity Analysis (SA), Data assimilation (DA) and Design Optimization (DO).
Mohammad G. Abdo
ANS WINTER MEETING
November, 2018
Orlando, FL
PHYSOR 2018 MEETING
April, 2018
Cancun, Mexico
LATEST AND UPCOMING EVENTS
ANS ANNUAL 2018
June, 2018
Philadelphia, PA
MY LATEST RESEARCH
Multi-Level ROM methodology claims that one can identify the active parameter subspace needed for the dimensionality reduction and surrogate construction for the high-fidelity model (lattice assembly) using the low-fidelity model (one or few pin cells) then apply it without violating the constructed error bounds.
The error due to projecting the cross sections onto the low-fidelity proposed active subspace and running the high-fidelity model did not exceed the predicted error of 10% (in fact, did not exceed 3%) with a reduction of 2-3 orders of magnitude in running time and from 49784 nominal dimension to 1500 reduced dimension.