Research
This section collects research notes, speculative concepts, and project sketches. Most pages are working documents rather than finished papers. The common thread is an interest in statistical, computational, and AI-assisted systems that remain inspectable under uncertainty.
Current Threads
Agricultural And Geospatial Modeling
Notes on crop-risk modeling, temporal field behavior, remote sensing, and geospatial infrastructure.
- Crop Lifecycle Temporal Modeling Notes: early notes on Sentinel-1, Sentinel-2, PRISM, and SSURGO as temporal signals for management and lifecycle embeddings.
- Wind Flow Simulation: early notes on atmospheric dynamics and simulation structure.
Representation Learning And Intrinsic Structure
Notes that connect geometry, uncertainty, embeddings, and high-dimensional learning.
- Intrinsic Geometry of Mathematics: a speculative note on mathematical corpora, intrinsic dimension, and latent geometry.
- Scale-Aware Uncertain PCA: a toy idea for treating numerical precision as uncertainty inside linear algebra.
- Gaussian Smoothing Neural Networks: a neural architecture concept based on Gaussian aggregation and smoothing.
- Hybrid Gaussian-ReLU Neural Networks: follow-up notes on combining smooth statistical structure with ReLU-style switching.
AI-Assisted Learning And Research Workflows
Notes on AI systems, teaching workflows, and research traceability.
- AI Homework: a concept document for an instructor-controlled AI tutoring system.
- AI Homework Intake: rough seed workflow for turning problems into guided tutoring objects.
Related raw source material is staged in ../intake/ai-assisted-research/ before it is adapted into public research notes.
Proof Tooling And Mathematical Automation
Notes on learned priors, tactic selection, and formal proof traces.
- Learned Prior for Tactic Selection: research sketch for smooth-manifold proof automation.
- Proposal Draft: more formal version of the same idea.
How To Read This Section
- Use
concepts/for early-stage ideas and speculative frameworks. - Use project subfolders, such as
ai-homework/andprob-proof/, for more focused threads. - Treat status labels and file context seriously: some notes are seeds, not claims.
- Expect ideas to move from rough notes to clearer public writeups over time.
Comments