Short On-Ramp Paths
Short On-Ramp Paths
The full learning paths run 10–160+ hours. These on-ramps don't — pick your persona and spend 5 minutes, 30 minutes, or 3 hours before deciding whether to commit to the full path.
How to Use These On-Ramps
Each persona below has three on-ramps of increasing depth. Every step links to a page that already exists elsewhere on the site — nothing here is new content. Time estimates are study time (reading plus, where noted, running something), the same convention used by the full learning paths. Each on-ramp ends with one self-check question so you can tell whether it landed before you decide to go deeper.
Learner
New to control theory or biomechanics.
5 Minutes
Goal: See where the Foundations path fits and how the site grades its own claims.
- Overview — ~5 min
Self-check: In AffineDrift’s core equation \(\dot{x} = f(x) + G(x)u\), which term is called “drift” and which is “control”? Answer: \(f(x)\) is drift — the state-dependent dynamics that remain when the control input is zero. \(G(x)u\) is the control (input) contribution.
30 Minutes
Goal: Refresh the Lagrangian mechanics the Foundations path assumes.
- Lagrangian Reference — ~30 min
Self-check: Why does the Lagrangian Reference say energy-based mechanics is often easier than force-based (Newtonian) mechanics for a system like the golf swing? Answer: Energy is scalar, so you only add numbers; the Newtonian approach requires resolving vector forces in multiple directions for every coupled part — torso, arms, bending shaft, and clubhead.
3 Hours
Goal: Start the Foundations path for real and meet the drift/input framework twice.
- Learning Path: Foundations — ~30 min (review the module map and select Module 1)
- Theory Part 1: Control-Affine Derivation — ~30 min
- Theory Part 2: Counterfactual Diagnostics — ~120 min (work through the drift/input separation and both counterfactual diagnostics)
Self-check: Theory Part 2 says the forward Zero Torque Counterfactual (ZTCF) “can define a future trajectory” while the Zero Velocity Counterfactual (ZVCF) “is a state evaluation.” If you want to know how the swing would continue from right now, which one actually answers that, and why doesn’t the other? Answer: ZTCF does — it integrates the drift dynamics forward from the current state, producing an actual forward trajectory. ZVCF only re-evaluates the model once at a single velocity-reset state, so it describes that one instant, not a continuation.
Researcher
Academic or scientific focus.
5 Minutes
Goal: See how AffineDrift labels literature it hasn’t reviewed yet, before you rely on any of it.
- Research Reviews — ~5 min
Self-check: The Research Reviews hub is marked “Planned (Scaffolding Phase).” What do these pages actually contain right now, and what should you not treat them as? Answer: Source-collection stubs — review questions and links to search, not paper-by-paper summaries. They should not be treated as synthesized literature reviews yet.
30 Minutes
Goal: Read the derivation the rest of the Theory series builds on.
- Theory Part 1: Control-Affine Derivation — ~30 min
Self-check: Theory Part 1’s Scope & Exclusions callout lists several things the model explicitly leaves out — name two, and explain why that matters for how you read the drift term \(f(x)\). Answer: It excludes aerodynamic drag, compliant or changing foot contact, marker dynamics, muscle activation delays, and impacts (any two). It matters because \(f(x)\) only captures what’s inside the declared model — anything excluded doesn’t vanish from the real swing, it just isn’t attributed by the decomposition.
3 Hours
Goal: Finish the five-part Theory series and preview where the research continues.
- Theory Part 2: Counterfactual Diagnostics — ~40 min
- Theory Part 3: Drift Invariance and Force Taxonomy — ~45 min
- Theory Part 4: Worked Pendulum Dynamics — ~60 min (work through the pendulum derivations)
- Theory Part 5: Numerical Verification Protocol — ~15 min
- Tangent-Space Series — ~20 min (preview only)
Self-check: Theory Part 5 says the worked pendulum in Part 4 supplies “independently checkable mathematical examples.” What does “independently checkable” mean here, and why does it matter more than whether the example is specifically about golf? Answer: It means the beam/pendulum derivation can be verified on its own — the math is reproducible without trusting the golf application. That matters because it separates “is the calculation internally consistent” from “does the model predict a real golfer,” which Part 4 explicitly says are different questions.
Integrator
Software developer integrating tools.
5 Minutes
Goal: Know what the Programming Companion actually is before installing it.
- Programming Companion Catalog — ~5 min
Self-check: The catalog is pinned to one exact UpstreamDrift commit hash rather than tracking “latest.” What does pinning to an exact commit buy you that a moving-target catalog wouldn’t? Answer: Every fact on the page stays checkable against one precise, unchanging, hash-verified snapshot — a “latest” catalog could describe software that’s already changed underneath you.
30 Minutes
Goal: Survey the software landscape before picking a tool to integrate.
- Software — ~30 min
Self-check: The Software page lists Pinocchio, MuJoCo, and Drake alongside OpenSim/MyoSuite, which it marks experimental. If you needed the most mature, well-supported path for a new integration, which tier would you look for? Answer: The non-experimental engines — Pinocchio, MuJoCo, and Drake — rather than the OpenSim/MyoSuite integrations the page flags as experimental.
3 Hours
Goal: Move from browsing the catalog to running a verified workflow locally.
- Programs and Models Directory — ~20 min
- Governed Workflows and Verification Tasks — ~20 min
- Install locally and run the
installation-verificationworkflow documented on that page — ~140 min
Self-check: The Workflow Registry has two kinds of rows: workflow and failure-fixture. What does a failure-fixture row verify, and why would a catalog of reference workflows deliberately include tests that are supposed to fail? Answer: A failure-fixture (for example “Reject Invalid Pendulum Parameters” or “Reject a Stale Companion Compatibility Version”) verifies that the software correctly rejects bad input or an incompatible state. They’re included so error handling gets the same exact-exit-code, SHA-256-verified rigor as the successful workflows, instead of going unchecked.
Experimentalist
Hands-on simulation and modeling.
5 Minutes
Goal: See the simulation inventory you’ll be choosing from.
- Programs and Models Directory — ~5 min
Self-check: Before you run any program from the 70-record directory, which column tells you whether its output is more than exploratory, and what does the Authority Boundary note explicitly say the inventory does not establish? Answer: The Maturity column (stable / beta / experimental); the Authority Boundary note says the inventory does not establish experimental validation or clinical applicability.
30 Minutes
Goal: Learn the decision protocol for picking model complexity before running anything.
- Planar-to-Spatial Model Ladder — ~30 min
Self-check: The Model Ladder page says it asks one limited question when choosing a model level. What is that question, and what does it deliberately avoid doing? Answer: Which is the least complex declared level that satisfies a particular task metric; it deliberately avoids ranking whole models against each other with one overall fidelity score.
3 Hours
Goal: Pick a reference workflow, run it, and check the result against real data.
- Governed Workflows and Verification Tasks — ~20 min (pick
reference-double-pendulum-simulation) - Run the workflow and inspect its output — ~120 min
- Datasets — ~40 min (compare against a real recorded dataset)
Self-check: The Model Ladder’s fixtures use manufactured, synthetic parameters, not participant data. Once you’ve run reference-double-pendulum-simulation, why is comparing its output against a recorded dataset a meaningfully different check than comparing it against another synthetic fixture? Answer: A synthetic fixture only confirms the code reproduces its own manufactured numbers — internal consistency. Comparing against a real recorded dataset checks whether the simulated behavior resembles an actual measured swing, which is a genuinely different, empirical kind of check.
Reviewer
Critical evaluation of claims.
5 Minutes
Goal: See how AffineDrift labels literature it hasn’t reviewed yet, before you rely on any of it.
- Research Review: Baseball Pitching — ~2 min
- Research Review: Shaft Flexibility — ~3 min
Self-check: Both review pages you just read carry an “Evidence Status” warning. What, specifically, do they say you should not do with the material they link to yet? Answer: Not treat the cross-sport pitching analogy or the shaft-flexibility source links as established claims about golf — they’re source collections, not yet-reviewed conclusions, until the cited papers are engaged with directly.
30 Minutes
Goal: Understand what’s already been formally challenged, and how.
- Critiques & Responses — ~30 min
Self-check: The Critiques & Responses page says these are “working documents, not peer-reviewed assessments.” What does that distinction change about how you should weigh an unresolved critique there versus a peer-reviewed one? Answer: An unresolved critique here is a transparently published, flagged concern under active engagement — not a vetted, externally adjudicated finding — so it should be read as an open question the framework hasn’t necessarily answered yet, not as settled peer-reviewed criticism.
3 Hours
Goal: Understand the framework being critiqued, then read representative critiques across categories.
- Theory Part 1: Control-Affine Derivation — ~30 min
- The Fallacy of Passive Drift and the Skeletal Baseline — ~20 min
- ZTCF Identifiability — ~20 min
- Scientific Claim-Audit Inventory — ~90 min
- Research Review: Interaction Forces — ~20 min
Self-check: Theory Part 1 explicitly excludes aerodynamic drag from the model. The ZTCF Identifiability critique argues this specific exclusion causes a concrete misattribution during a real analysis — what gets misattributed, and to whom? Answer: Because drag isn’t in the drift model, its effect still shows up in the measured motion; inverse dynamics then computes a lower total torque than the model expects, and the residual (input = total − drift) reads as negative “braking” — so aerodynamic drag gets misattributed to the golfer as active braking effort.
Contributor
Contributing to the project.
5 Minutes
Goal: See how to get involved and what already exists to build on.
- Collaborate — ~3 min
- Resources Hub — ~2 min
Self-check: Collaborate says submitting an issue or pull request doesn’t guarantee acceptance, endorsement, or scientific validation. What does a proposed claim actually have to pass before it counts as accepted? Answer: The repository’s own evidence and review contracts — the same bar every other claim on the site has to clear, not just a maintainer’s approval of the pull request.
30 Minutes
Goal: Understand the project’s authority and how to reach the maintainer.
- About — ~30 min
Self-check: About says the author’s biography is “not independent evidence for the scientific claims published on the site.” Why does the page bother saying that on its own about page? Answer: Because credentials or personal interest don’t substitute for the site’s evidence standards — who wrote something doesn’t make a claim validated; only the cited sources, derivations, and review contracts do.
3 Hours
Goal: Read the roadmap, then try exporting the program catalog yourself.
- Development Roadmap — ~20 min
- Programs and Models Directory — ~20 min
- Governed Workflows and Verification Tasks — ~20 min (find
program-catalog-export) - Run the
program-catalog-exportworkflow and review the output — ~120 min
Self-check: The Development Roadmap explicitly omits dates and percentages. What does it name as the actual authority for what’s been delivered, instead of a hand-maintained forecast? Answer: Protected merge history on GitHub — the live issue tracker and pull requests, where “protected” means changes landed only through required review and checks.
Curious Golfer / Coach
Physics, equipment, and swing mechanics.
Notice: AffineDrift is a scientific and theoretical mechanics resource. It does not provide swing instruction, swing fixes, or personal coaching.
5 Minutes
Goal: See what’s actually measured before the physics gets technical.
- Technology — ~5 min
Self-check: The Technology page marks each linked article’s state as “Available.” What does “Available” actually promise, and what does it explicitly not promise? Answer: It promises the article can be read — nothing more. It does not promise every claim on that page is validated; each article still has to be checked for what’s measured, derived, estimated, or simulated before you trust its conclusions.
30 Minutes
Goal: Read the chapter that ties clubhead delivery to ball flight.
- Chapter 28: Impact and Launch — ~30 min
Self-check: Chapter 28 says static loft plus attack angle does not determine dynamic loft, and that three scalars omitting speed and spin axis cannot determine ball flight. What has to be measured directly at impact that those simpler numbers can’t substitute for? Answer: The actual face attitude and contact conditions at the moment of impact, plus the full spin vector and ball speed — dynamic loft and launch depend on the club’s real state during contact, not on setup angles measured beforehand.
3 Hours
Goal: Read the golf-science path map and start the textbook from Chapter 1.
- Learning Path: Golf Science — ~20 min (skim the module map)
- The Physics of Golf — ~10 min (browse the table of contents)
- Chapter 1: Why Physics — ~80 min
- Chapter 3: The Double Pendulum — ~70 min
Self-check: Chapter 1 distinguishes three kinds of statement: an identity derived within a model, a prediction from fitted parameters, and an empirical claim about golfers. Why can’t a correctly derived identity alone tell you something true about a real golfer’s swing? Answer: An identity is only checked mathematically, within the model’s own declared assumptions — it says nothing about actual golfers until it’s connected to relevant measurements, the way an empirical claim requires.
Student
Coursework, robotics, and biomechanics.
5 Minutes
Goal: Get the vocabulary straight before lectures and readings use it.
- Mathematical Notation and Terminology — ~2 min
- Glossary — ~3 min
Self-check: The Notation page fixes \(G(x)\) as uppercase for the control-affine input map. If you see the input map written with a lowercase \(g\) in a swing-dynamics equation on this site, what should you conclude? Answer: That it’s a notation error against the site’s own normative convention, not a different mathematical object — the input map is always \(G(x)\), uppercase, throughout the control-affine equations.
30 Minutes
Goal: Refresh the Lagrangian mechanics used across the theory and biomechanics material.
- Lagrangian Reference — ~30 min
Self-check: The Lagrangian Reference says the Newtonian approach requires resolving every force as a vector, while the Lagrangian approach works with scalar energy functions instead. Why does that make the Lagrangian approach easier for a system like a golf swing with many coupled moving parts? Answer: Because energy is a scalar, the parts combine by simple addition instead of having to resolve and sum force vectors in multiple directions for every rotating torso, arm, shaft, and clubhead segment — both approaches give the same equations of motion, but the bookkeeping is far simpler.
3 Hours
Goal: Connect the math to a reproducible, hands-on example.
- Learning Path: Biomechanics & Motor Control — ~20 min (skim the module map)
- Theory Part 1: Control-Affine Derivation — ~30 min
- Theory Part 4: Worked Pendulum Dynamics — ~70 min (work through the beam and pendulum derivations)
- Run the
reference-double-pendulum-simulationworkflow from Governed Workflows yourself — ~60 min
Self-check: Theory Part 4’s beam and pendulum derivations are checked by an executable test suite that independently integrates the kinetic energy, differentiates the potential, and verifies the power balance. Why does verifying a derivation this way matter more than just trusting the algebra on the page? Answer: Because an independent numerical check catches errors that a correct-looking derivation can hide — if the integrated energy balance and the written equations disagree, you know something is wrong even before you compare the model to any real golfer.
Citation
@online{untitled,
author = {},
title = {Short {On-Ramp} {Paths}},
url = {https://affinedrift.com/resources/on-ramp-paths.html},
langid = {en}
}