Robust Derivative Estimation with
Walk on Stars
Yu, Sawhney, Miller, Wu, Zhao
SIGGRAPH Asia (2025)
About
I'm a Senior Research Scientist in the High Fidelity Physics group at NVIDIA. Before joining NVIDIA, I did my PhD in Computer Science at Carnegie Mellon University, advised by Keenan Crane. I'm broadly interested in developing new algorithms at the intersection of geometry processing, simulation and rendering. My current research explores how core problems in PDE-based geometric computing can be efficiently and reliably solved via grid-free Monte Carlo methods without any volumetric mesh generation, taking inspiration from algorithms for photorealistic rendering. Checkout my PhD thesis for further details.
I've been fortunate to receive Best Paper and Dissertation awards at SIGGRAPH and CMU (2022, 2024, 2025), as well as an Nvidia Graduate Fellowship for work on Monte Carlo PDE solvers. I've also received the Symposium on Geometry Processing best software award for the Boundary First Flattening application to quickly and robustly generate UV maps. Previously, I worked at IrisVR as a core graphics engineer, and received a Bachelor’s in Physics and Computer Science from Columbia University. Find my CV here.
Research
Robust Derivative Estimation with
Harmonic Caching for
Walk on Spheres
Zhao, d’Eon, Sawhney, Jarosz
SIGGRAPH Asia (2025)
State of the Art in Monte Carlo Methods For Partial Differential Equations
Sawhney, Miller, Crane, Gkioulekas
SIGGRAPH & SGP Courses (2025, 2024)
Unbiased Differential Visibility Using Fixed-Step Walk-on-Spherical-Caps and Closest Silhouettes
Wu, Morrical, Bangaru, Sawhney, Zhao, Wyman, Ramamoorthi, Lefhon
SIGGRAPH (2025)
Solving partial differential equations in participating media
Miller, Sawhney, Crane, Gkioulekas
SIGGRAPH (2025)
Differential
Walk on Spheres
Miller, Sawhney, Crane, Gkioulekas
SIGGRAPH Asia (2024)
Walkin’ Robin: Walk on Stars with Robin Boundary Conditions
Miller*, Sawhney*, Crane†, Gkioulekas†
SIGGRAPH (2024)
Paper | Project Page | Code
Decorrelating ReSTIR Samplers via MCMC Mutations
Sawhney, Lin, Kettunen, Bitterli, Ramamoorthi, Wyman, Pharr
ACM Transactions on Graphics (2023)
Paper | Supplemental | Video | Talk
Walk on Stars: Grid-Free Monte Carlo for PDEs with Neumann Conditions
Sawhney*, Miller*, Gkioulekas†, Crane†
SIGGRAPH (2023)
Boundary Value Caching for
Walk on Spheres
Miller*, Sawhney*, Crane†, Gkioulekas†
SIGGRAPH (2023)
Grid-Free Monte Carlo for PDEs with Spatially Varying Coefficients
Sawhney*, Seyb*, Jarosz†, Crane†
SIGGRAPH (2022)
Best Paper (Honorable Mention)
Paper | Supplemental | Code | Talk
Monte Carlo Geometry Processing: A Grid-Free Approach to PDE-Based Methods on Volumetric Domains
Sawhney, Crane
SIGGRAPH (2020)
Paper | Project Page | Code | Talk
Boundary First Flattening
Sawhney, Crane
ACM Transactions on Graphics (2018)
SGP Best Software Award (2019)
Paper | Project Page | Talk | Web Demo
Software
ZOMBIE
Monte Carlo PDE Solvers
FCPW
Fast closest point queries
Boundary First Flattening
Quick and robust UV mapping
geometry-processing-js
3D geometry processing on the web
linear-algebra-js
Sparse & dense matrix routines on the web
