BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:Asia/Kolkata
X-WR-TIMEZONE:Asia/Kolkata
BEGIN:VEVENT
UID:201@cds.iisc.ac.in
DTSTART;TZID=Asia/Kolkata:20260603T113000
DTEND;TZID=Asia/Kolkata:20260603T123000
DTSTAMP:20260526T161148Z
URL:https://cds.iisc.ac.in/events/m-tech-research-thesis-defense-cds-03-ju
 ne-2026-a-physics-informed-framework-for-super-resolution-of-fluid-flows/
SUMMARY:M.Tech Research: Thesis Defense: CDS: 03\, June 2026 "A physics-inf
 ormed framework for super-resolution of fluid flows"
DESCRIPTION:DEPARTMENT OF COMPUTATIONAL AND DATA SCIENCES\nM.Tech Research 
 Thesis Defense\n\n\n\nSpeaker : Ms. Diya Nag Chaudhury\nS.R. Number : 06-1
 8-01-10-22-23-1-23885\nTitle : "A physics-informed framework for super-res
 olution of fluid flows"\nThesis examiner : Dr. Ganesh Kiran Vaidya\, Dept.
  of Mathematics\, IISc.\nResearch Supervisor : Prof. Sashikumaar Ganesan\n
 Date &amp\; Time : June 03\, 2026 (Wednesday) at 11:30 AM\nVenue : # 102 C
 DS Seminar Hall\n\nABSTRACT\nThe reconstruction of high-resolution flow fi
 elds from low-resolution data remains a persistent challenge within fluid 
 dynamics. Super-resolution using deep learning is a valuable tool for enha
 ncing the visualization of flow fields\, allowing for the recovery of deta
 iled features from low-resolution data. This work introduces the Bicubic F
 ourier Neural Operator (Bicubic FNO)\, which combines the benefits of both
  bicubic interpolation and FNO. We compare our architecture against standa
 rd FNO and two convolutional neural network (CNN) models reported in the l
 iterature for super-resolution: Super-resolution Convolutional Neural Netw
 ork (SRCNN)\, widely used for natural images\, and the Downsampled Skip-Co
 nnection/Multi-Scale (DSC/MS) model\, designed for reconstructing high-res
 olution turbulent flow fields.\nWe tested our model on two-dimensional dec
 aying homogeneous isotropic turbulence (HIT)\, wall-bounded channel flow\,
  and the Stanford flame dataset. Our results demonstrate a 10.47% improvem
 ent over the standard FNO and a 19.59% improvement over the DSC/MS model i
 n terms of Peak Signal-to-Noise Ratio (PSNR) for 2D HIT. We achieved a 63%
  reduction in Mean Squared Error (MSE) compared to pure bicubic interpolat
 ion\, confirming the FNO's ability to recover fine-scale structures.\n\nWe
  extend our research to integrate physical principles into the framework\,
  ensuring that the model can reconstruct physically consistent flow fields
  even with sparse data. Incorporating this domain knowledge into the netwo
 rk provides an additional 14% improvement in MSE\, indicating that physica
 l knowledge effectively complements data-driven learning.\n\nWe find that 
 physics constraints\, when carefully formulated and weighted\, can signifi
 cantly enhance learned models. To ensure physical consistency\, we enforce
  a divergence-free constraint that guarantees local mass conservation\, a 
 principle fundamental to incompressible flow physics. We then transition t
 o a self-supervised approach\, which opens new possibilities for applicati
 ons where generating high-fidelity training data is computationally prohib
 itive\, such as direct numerical simulations (DNS) at high Reynolds number
 s. Finally\, we employ Explainable AI (XAI) to examine how the model learn
 s physical structures\, respects conservation laws\, and processes turbule
 nce phenomena.\nThis research highlights the efficacy of the Bicubic FNO a
 nd its extension with physical constraints\, demonstrating that we can ext
 ract meaningful turbulent features while significantly reducing the relian
 ce on massive datasets.\n\n\n\nALL ARE WELCOME
CATEGORIES:Events,Thesis Defense
END:VEVENT
BEGIN:VTIMEZONE
TZID:Asia/Kolkata
X-LIC-LOCATION:Asia/Kolkata
BEGIN:STANDARD
DTSTART:20250603T113000
TZOFFSETFROM:+0530
TZOFFSETTO:+0530
TZNAME:IST
END:STANDARD
END:VTIMEZONE
END:VCALENDAR