Who am I? (abridged version)

• Name is pronounced Yeh-or-yee-os, but I'll accept George(ous)
• Born and raised in Cyprus, before heading to Boston for (u)grad
• Into musical theatre, camping, and generative art

• 2nd year Miller fellow
• Co-hosted by Prof. Mary Scott (MSE) and Prof. Joel Moore (Physics)
• Interested in imaging fields (temperature, magnetization, current density) inside materials w/ high resolution
• This is a high-dimensional, non-convex, inverse problem

Non-convex inverse problems (aka Sudoku for dummies)

• Sudoku is a constraint satisfaction problem:
• Each row contains only one of each of the numbers 1-9
• Each column contains only one of each of the numbers 1-9
• Each 3x3 block contains only one of each of the numbers 1-9
• Solution must be consistent with a (minimum) set of pre-defined clues
• Each constraint can be interpreted as a projection to a non-convex set
• This can now be solved using iterative proximal gradient algorithms

iteration 01

iteration 02

iteration 03

iteration 04

iteration 05

animated solution