Intuition The ONE core idea
A system that repeats its motion traces a closed loop in an abstract plane of states (which we will call phase space in section 2) — and that loop can be described by just two honest numbers: how big it is (the action) and how far around it we are (the angle). Everything on the parent page is machinery for turning a messy repeating motion into that clean loop.
Before you can read the parent note, you must own its vocabulary. Below, every symbol and idea is built from nothing, in an order where each rung of the ladder rests on the one below it.
q (position) and p (momentum)
== q == is where a thing is (an angle, a distance — one number per way it can move). == p == is how much motion it carries — roughly mass times velocity, p = m v , for simple cases.
Intuition Why TWO numbers, not one?
If I only tell you a pendulum is "at the bottom", you can't predict the next instant: it might be swinging fast or sitting still . You need both where it is (q ) and how it's moving (p ). Knowing ( q , p ) now fixes the entire future — that pair is the complete "state".
Look at figure 1: one dot on the plane whose horizontal address is q and vertical address is p . That single dot is the whole system at one instant.
The plane (or higher space) whose axes are q and p . One point = one complete state. As time runs, the point moves , drawing a curve.
Intuition Why draw it this way?
Because a repeating motion draws a closed loop here. Out-and-back-and-out-again returns you to the same ( q , p ) , so the pen returns to where it started. That closed loop is the hero of the whole topic.
In figure 2 the swinging pendulum's dot travels round an oval, over and over, never escaping it.
Definition Periodic (bounded) motion
Motion that stays in a limited region and repeats after a fixed time T (the period). A swing, a planet, a mass on a spring.
Intuition Two flavours of "repeat"
Libration — back-and-forth like a swing: q goes up to a max, back to a min, back up. In phase space: a closed oval .
Rotation — round-and-round like a spinning wheel: q keeps increasing (angle 0 → 2 π → 4 π … ) but the state repeats every turn.
Both give a closed loop in phase space; the parent note's "there-and-back" factor of 2 comes from libration, and the "q increases through 2 π " contour comes from rotation.
Intuition The wrap-around for rotation (
q ≡ q + 2 π )
For rotation-type motion, q is an angle , so q = 0 and q = 2 π are the same physical place — we identify them, written q ≡ q + 2 π . Picture cutting the number line every 2 π and gluing the cuts into a circle: the "branch cut" at q = 2 π is just the seam where the paper is glued. That is why the rotation contour runs once through 2 π (not there-and-back) and still closes: you come back to the same seam. Miss this identification and you would wrongly think the loop never closes.
T and angular frequency ω
T = seconds for one full loop . == ω == (Greek "omega") = how many radians of loop-progress per second :
ω = T 2 π .
2 π ?
We agree to say one full loop = 2 π of angular progress , exactly like one full turn of a clock hand. So ω answers "how fast does the clock hand of this motion sweep?" This is the single number the whole topic works to extract.
A way to measure angle by arc length on a unit circle . A full circle is 2 π ≈ 6.283 radians. Half a circle is π .
Intuition Why radians here?
The angle variable θ is designed to advance by exactly 2 π over one loop. Radians make "one loop = 2 π " true by definition, so no clumsy conversion factors ever appear.
∫ f d q
"Slice the range of q into tiny strips of width d q , multiply each by the height f there, add them all up." Geometrically: the area under the curve f versus q . Here f is some function of q ; in the very next section the height we will use is f = p , the momentum read off the loop.
Intuition Why an integral for action?
The action is an area in phase space . Area = sum of thin vertical strips of height p (the momentum at that q ) and width d q . That sum is ∫ p d q . No mystery — it's area-counting.
∮ p d q
The ∮ (circle on the integral) means: add up p d q all the way around the closed loop , returning to the start. The result is the area the loop encloses .
Intuition Why the little circle matters
If you only integrate from q m i n to q m a x you get half the loop (the top). The circle tells you to also come back along the bottom — so the two halves together fence off the full enclosed area. Figure 3 shades exactly this area.
Definition Orientation / sign convention
Direction matters. We traverse the loop counterclockwise in the ( q , p ) plane (the direction the physical motion actually takes: at the top p > 0 so q increases, at the bottom p < 0 so q decreases). Going counterclockwise makes the enclosed area — and hence J — positive . Traverse it clockwise and you'd get the same number with a minus sign; we always pick the sense that makes J > 0 .
Common mistake Dropping the return trip
Integrating one-way and forgetting the bottom of the loop halves your answer. ∮ = whole loop, always .
J
J ≡ 2 π 1 ∮ p d q .
In words: take the loop's area, divide by 2 π . It's a single number saying "how big is this orbit".
2 π ?
So that its partner angle winds by a tidy 2 π per loop (proved on the parent page). J stays constant as the system evolves — the loop doesn't change size while you march around it — which is exactly why it makes a good "coordinate that never moves".
Definition Angle variable
θ
The partner of J . It says how far around the loop you are, and it advances at a constant rate:
θ ( t ) = ω t + θ 0 .
Here == θ 0 == is the initial phase — where on the loop you happened to start at time t = 0 (the clock hand's reading when you first looked). It just shifts the whole schedule; it does not change the frequency. After one full loop θ has increased by exactly 2 π .
Think of a clock hand pinned at the loop's centre, ticking at steady speed ω . θ is the hand's angle; θ 0 is the reading at the moment you started your stopwatch; J is the loop's size. Together, ( θ , J ) replace the messy ( q , p ) with "clock hand + loop size".
H ( q , p )
A recipe that eats a state ( q , p ) and returns its total energy (kinetic + potential). For motion that doesn't change with time, H stays fixed at a value we call E (the energy).
Intuition Why it matters here
The equation H ( q , p ) = E is the loop! It's the exact curve of all states with the same energy — the oval in figure 2. Solving it for p gives p ( q , E ) , the height we integrate to get the area. See Hamiltonian mechanics for the full engine.
∂ J ∂ H and dJ d E
A derivative is a rate of change / slope : "if I nudge the bottom variable a tiny bit, how much does the top one move?" The curly ∂ (partial) means "change this one while holding the others fixed"; the straight d is used when only one variable is in play. (Both H and E here are the Hamiltonian/energy just defined in section 10.)
Intuition Why derivatives give the frequency
The parent's punchline is ω = dJ d E : the frequency is the slope of the energy-vs-loop-size graph . Steeper graph → faster oscillation. This is why you can get ω without solving the motion — just measure a slope of areas.
Definition Canonical transformation
A change of coordinates ( q , p ) → ( θ , J ) that keeps Hamilton's rules intact — the physics is unchanged, only the description gets cleaner. See Canonical transformations .
Definition Generating function
W ( q , J )
A single function of the old position q and the new momentum J that manufactures the transformation through two defining relations:
p = ∂ q ∂ W , θ = ∂ J ∂ W .
Differentiate W with respect to q and you recover the old momentum p ; differentiate it with respect to J and you get the new angle θ . See Generating functions and the Hamilton–Jacobi equation where W comes from.
Intuition Why we need this gadget
We can't just declare new coordinates; they must respect the mechanics. W is the guaranteed-legal machine that produces ( θ , J ) from ( q , p ) without breaking any laws.
Definition Poisson bracket
{ F , G } is a number-valued operation on two quantities that measures how one changes as you flow along the other . If { F , G } = 0 they are "in involution" — compatible, non-interfering. See Poisson brackets .
Intuition Why for multi-DOF
For n moving parts, integrability needs n conserved quantities that all commute ({ F i , F j } = 0 ). That compatibility is what lets the many loops assemble into a single doughnut (Liouville–Arnold theorem ).
A doughnut surface. With n degrees of freedom the loop generalises to an n -dimensional torus, and motion becomes straight-line winding on it. Figure 4 shows the 2-torus with its two angles.
Intuition Why a doughnut?
Each degree of freedom has its own loop (its own circle). Two circles combined = the surface of a doughnut; n circles = an n -torus. Position on it = one angle per circle, exactly the θ i .
position q and momentum p
periodic motion draws a closed loop
period T and frequency omega
closed loop integral of p dq
action J equals loop area over two pi
angle theta winds at rate omega
Hamiltonian H equals E is the loop
canonical transform via generating W
clean coordinates theta and J
Poisson brackets in involution
Action angle variables topic
Cover the right side; can you answer each before reading the parent note?
What does a single point in phase space represent? One complete state of the system — a position q and momentum p together.
Why do we need both q and p , not just q ? Position alone can't predict the next instant; you also need how fast/which way it's moving. The pair fixes the whole future.
What shape does periodic motion draw in phase space? A closed loop (an oval for libration, a full circle for rotation).
What is the difference between libration and rotation? Libration is back-and-forth (there-and-back); rotation is round-and-round with q increasing through 2 π .
For rotation, why does the loop close even though q keeps increasing? Because q is an angle with the identification q ≡ q + 2 π — q = 2 π is the same place as q = 0 , so one trip through 2 π returns to the start.
What does ∮ mean and how does it differ from ∫ ? ∮ sums all the way around the closed loop (whole area); ∫ sums over an open range (often just half the loop).
Which way do you traverse the loop, and why? Counterclockwise in ( q , p ) — the physical sense of motion — which makes the enclosed area and J positive.
Define the action J in one line. The loop's enclosed phase-space area divided by 2 π : J = 2 π 1 ∮ p d q .
Why divide the area by 2 π ? So its partner angle θ advances by exactly 2 π per loop.
What does the angle variable θ track, and what is θ 0 ? θ = how far around the loop you are, advancing at rate ω ; θ 0 = the initial phase, where you started at t = 0 .
What is a derivative, in plain words? A slope — how much the top quantity changes when you nudge the bottom one.
How do you get the frequency ω from energy and action? ω = d E / dJ , the slope of the energy-versus-action graph — no trajectory needed.
What equation is the loop? H ( q , p ) = E — the set of all states with the same energy.
What are the two defining relations of the generating function W ( q , J ) ? p = ∂ W / ∂ q and θ = ∂ W / ∂ J .
When is a multi-DOF system integrable, and what shape does its motion live on? When it has n conserved quantities in involution ({ F i , F j } = 0 ); motion then winds on an n -torus (doughnut).