PAI-101 · Education

Exploring Intelligence: The Matching Principle

One idea, carried from a 1928 antenna to a theory of mind. A wave meets a boundary and whatever does not match comes back, and that same geometry reappears as lost channel capacity, as a bandwidth budget you cannot overspend, as the heat of erasing a bit, as a robot's contact chattering against a wall, and finally as surprise. You start with no prerequisites and finish by writing one function that reproduces every answer you earned along the way. The Smith Chart Navigator and the passivity disk are your instruments throughout.

Beginner → Skilled·7 modules·14 lessons
▶ Start the course ← All courses
THE HORIZON

Where this sits, and what moves it.

Binding constraint · The reflected wave. Every interface returns whatever does not match, and the return is measurable in identical form at an antenna, a gearbox, a noisy channel, and a robot's contact patch.

Was impossible

Matching was a 1928 antenna result. That the same geometry sizes a gearbox, caps a channel's capacity, and prices the heat of erasing a bit was not something anyone could put in front of a student in one sitting, because the four fields did not share a notation. Building that shared notation is what this course is.

Is probable

The unification holds where it has been checked: the maximum power transfer theorem, the water-filling solution, and Landauer's kT ln 2 are the same statement in three costumes. What is not settled is how far it reaches. Whether surprise is genuinely the last reflected wave, or a good analogy that stops paying, is the open question the last module hands you rather than answers.

Becomes possible

It becomes a design rule the day someone sizes a learning system's interface the way an engineer sizes a matching network -- computing the mismatch, predicting the reflection, and measuring it where predicted. The gap is not conceptual; it is that this review did not locate a written-down impedance of a policy in units you can measure on a bench.

Every hard thing was impossible until the constraint that made it impossible was named. How we read a frontier →

Module 1

Foundations: the wave, the bit, and the loop

Build the vocabulary from nothing: what reflects, what a bit is, and where an agent touches its world.

  1. L1Waves, reflection, and the scariest chart in engineeringYou send a signal down a cable into a mismatched load. Where does the power that isn't absorbed go?Meet the founding image of the course: a wave hits a boundary, and whatever does not match comes back.→
  2. L1What is information?You measure the entropy of English three ways: letter frequencies, then letter pairs, then longer context. What happens to the number?Put a number on surprise, so that later we can ask what a channel costs.→
  3. L1What is an agent?You sweep how tightly an agent couples to its world, from barely touching it to gripping it completely. Where does performance peak?Draw the loop that makes something an agent, and find the one place it can touch its world.→
Module 2

The universal matching principle

Show that one theorem sizes an antenna and a gearbox, and that impedance is a language, not a metaphor.

  1. L2Impedance everywhere: the universal matching principleOne theorem sizes an antenna's load and a robot's gear ratio. What does it say they have in common?Show that the same theorem sizes an antenna and a gearbox.→
Module 3

The information spine

Carry matching into information: the optimal receiver, and the conservation law that caps every match.

  1. L2Channels, capacity, and the matched filterYou build four receivers for the same signal in noise. Which one achieves the best signal-to-noise ratio?Move the matching idea out of impedance and into information.→
  2. L2Bandwidth is a budget: Bode–FanoYou try to design a matching network that is perfect across a whole frequency band. How close can you get?Meet the conservation law that says a perfect match over a band is impossible.→
Module 4

The physical price of a bit

Find where information finally touches thermodynamics, and what forgetting costs.

  1. L2The physical price of a bit: LandauerIs there a fundamental minimum energy to ERASE one bit of information?Find the exchange rate between information and heat.→
Module 5

Physical AI: learning the interface

Follow the chart onto the die and into the model, and design a matching network with a net.

  1. L3Learned matching: when AI designs the interfaceYou distil an exhaustive matching search into a small neural network. What has the network actually learned?Hand the matching problem to a network and watch it become a coordinate system a model works in.→
Module 6

Embodied AI: contact and the body

Meet the disk again in mechanical dress: the passivity condition, and what a compliant body computes.

  1. L3Impedance control and contact: staying inside the diskYou render a virtual wall in software and make it stiffer and stiffer. What happens?Discover the safety condition for touching an unknown world, and find that it is the Smith chart's unit disk.→
  2. L3Morphological computation: the body as computerA task is too hard for the robot’s CONTROLLER to solve. Can changing the BODY (its shape and material) solve it without a smarter controller?Watch a body solve a problem its controller could not.→
Module 7

The unifying frontier

Reach surprise as the last reflected wave, and collapse the course into one function.

  1. L3The free energy principle: matching as cognitionOne quantity, minimised two different ways, gives you two familiar things. What are they?Reach the top of the ladder: surprise as the last reflected wave.→
  2. L3Curiosity: seeking the reflected waveTwo learners share one data budget. One takes whatever arrives, the other hunts the experiences it predicts worst. When does curiosity win?Turn the free-energy agent active, make it seek the mismatch, and find exactly when curiosity buys speed and when it does not.→
  3. L3Development: the U-shaped curveA learner faces a mostly-regular mapping with a few exceptions. Track its accuracy on the exceptions across training. What shape appears?Watch a learner *develop* (memorize, then over-generalize, then reconcile) the fingerprint that tells an entity apart from a controller.→
  4. L3Capstone: a theory of intelligence as matchingYou write the reflection coefficient once and apply it to an antenna, a digital wall, a power supply and a gearbox. How many of the four does it answer?Collapse the whole course into a single function, and make it reproduce every answer you earned.→