Advanced -> Gradient Descent¶

Theory-breadth smooth optimization taught first through fixed-step batch gradient descent on one-feature linear regression with squared loss.

Microtopics¶

Source	Type
Stanford CS229 notes on linear regression	`Course`
MIT Open Learning Library gradient descent chapter	`Course`
MIT nonlinear optimization notes on gradient descent	`Course`

Material	Type
Gradient Descent hot sheet	`quick reference`
Linear Regression Gradient Descent Benchmark	`flagship note`
Machine Learning Algorithms tutorial	`compare point`
Template Library exact starter route	`starter route`

Problem	Source	Difficulty	Context	Style	Prerequisites	Tags	Why it fits
Linear Regression Gradient Descent Benchmark	`Stanford CS229`	`Theory`	Smooth Optimization	Theory Benchmark; Deterministic Update Rule	Affine Models; Derivatives; Basic Algebra	Linear Regression; Squared Loss; Learning Rate	The cleanest first gradient-descent benchmark for this repo because one affine model and one smooth loss make the update rule completely explicit without pretending to cover modern optimizer families.

Problem	Source	Difficulty	Context	Style	Prerequisites	Tags	Why it fits
Gradient descent	`MIT Open Learning Library`	`Theory`	Further Theory	Course Notes; Theory Breadth	Gradient; Learning Rate	Convexity; Step Size; Convergence; Optimization	A good follow-up once the repo benchmark is clear and you want broader smooth-optimization intuition without widening the lane into full numerical optimization coverage.

Code	Title	Fit	Difficulty	Pattern	Note	Solution
`LINEARREGRESSIONGD`	Linear Regression Gradient Descent Benchmark	`primary`	`medium`	-	Note	Code

python3 scripts/generate_problem_catalog.py