Embedded Files
Codes
Codes
For full list of codes, see my GitHub page. Below are some selected codes.
Quantum linear solver: Fluid dynamics
Machine learning for fluid dynamics
Graph theory applications to fluid dynamics
CFD codes
Datasets
Datasets
Turbulent flows
Stably stratified turbulence
Published:
More on STRATA Group website: https://stratified-turbulence.github.io/web/Datasets/
Colliding thermals: Subgrid-scale effects in cloud-like atmospheric flows
Talks
Talks
Application of Network science to Turbulence
Quantum linear solvers for Fluid dynamics - 2025, Winter Classic Invitational Student Cluster Competition: talk link
Gallery
Gallery
Deep time-series ML models are developed for closure modeling of URANS equations for stably stratified turbulence. Models perform accurately in a priori and a posterior tests. Furthermore, the data requirements of the time-series ML models are explored by extracting physically relevant timescales of the complex system.
Network theory is used to formulate and characterize the web of interactions among vortical elements in wake flows. The nodes of these networks correspond to vortical elements in the flow field and the connections among them are weighted by the induced velocity. In particular, community detection is used to distill the overall interaction-based physics of the high-dimensional flow field into the vortical community centroids. ROM are formulated to predict the lift and drag in wake flows.
Network community detection is used to extract connector (high inter-community) and peripheral (low inter-community) structures that resemble shear-layer and vortex-core type structures, respectively, in 2- and 3-D isotropic turbulence. The effect of perturbing such structures to enhance local turbulence mixing is also explored.
The vortex dynamics in the wake of NACA airfoils are analyzed to understand the unsteady aerodynamic characteristics at low Reynolds number flows. Various canonical wake regimes are revealed at higher angles of attack and with the addition of Gurney flap, a passive flow modification device, at the trailing edge.
We introduce a physics-guided data-driven method for image-based multi-material decomposition for dual-energy computed tomography (CT) scans. The method is demonstrated on simulated breast models with calcifications as the dense material placed amongst the muscle tissues.
Two data-driven approaches are introduced, employing network analysis to elucidate specialized metabolite production by exogenous treatments. The direct route reveals the relationship between treatments and the production of known/putative specialized metabolites, whereas the auxiliary route distinguishes unique unknown analytes from abundantly produced ones.
Some interesting links
Some interesting links
Johns Hopkins Turbulence Database : Open source interactive turbulence database.
Awesome Network Analysis: Resources for network (graph) analysis.
Machine Learning Mastery: Intro to machine learning with results-first, hands-on approach (advanced topics are covered as well).
Useful tools
Useful tools
These are suggested based on my experience with them in Mac.
Draw.io or Diagrams.net : Open-source diagramming software. Does the same job as OmniGraffle but has LaTeX embedding. Install the desktop version using the root github repo or their latest installers.
ImageJ: Open-source image processing software. Simple movie maker from an image sequence (alternative for Sequimago).
Page updated
Google Sites
Report abuse