AI & softwareResearch findings
Bayesian Uncertainty Quantification for Doping Profile Reconstruction in Semiconductor Devices
Original published 2026.06.29 23:18 KSTAdded to Blackdeck 2026.11.17 15:03 KST
What has changed?
A pushforward prior using sigmoid transformation enables efficient representation of piecewise-constant doping profiles with sharp interfaces in nonlinear PDEs
Key points
- Propose a Bayesian framework to reconstruct doping profiles from boundary flux measurements
- Introduce a pushforward prior with sigmoid transformation to model sharp interfaces efficiently
- Employ adjoint-based gradients with NUTS sampler for reliable uncertainty quantification
Connect it to your work
Spatially resolved uncertainty quantification is achieved for both planar and curved interfaces
What remains unproven
This research note is based on an abstract; check the original for test conditions and application scope.
Originals used for this note
Links open the original publisher in a new tab.