Open Toda: a complete learning sequence

How can a nonlinear system have enough structure to be solved systematically? The finite open Toda chain provides a concrete answer. Its particles interact through exponential forces, yet their motion can be encoded in a matrix with constant eigenvalues. This sequence takes you from the equations to a proof of integrability and a reproducible three-particle computation.

You will distinguish three statements that often get confused: a quantity is conserved, several quantities Poisson-commute, and their gradients are independent. Together, with the correct count on the stated phase space, these establish Liouville integrability. A stable-looking numerical trajectory establishes something different.

The open Toda problem

Our model has real canonical coordinates $(q_i,p_i)$, dimensionless variables, and Hamiltonian

$$H=\frac12\sum_{i=1}^{N}p_i^2+ \sum_{i=1}^{N-1}e^{q_i-q_{i+1}}.$$

“Open” means there is no bond joining site $N$ to site $1$. It does not prescribe fixed endpoint positions. For the central example take $N=3$ and

$$q(0)=(0,0,0),\qquad p(0)=(1,0,-1).$$

The model record fixes the regime and compares it with periodic and infinite chains. The Library derivation provides the full argument for any finite $N\geq2$. You can consult either alongside the lessons; neither requires completion of an unwritten course.

Check your starting point

You need differentiation, partial derivatives, and multiplication of small matrices. The first lesson introduces Hamilton’s equations; later lessons develop the particular Poisson-bracket and matrix calculations they use. Python is needed only to execute or modify the final experiment.

For a more gradual introduction to Hamiltonian mechanics, use the bracket bridge and oscillator lesson, then return here. They explain canonical variables and separate conservation from mutual involution through calculations small enough to do by hand.

Try these three calculations before continuing.

1. For $V(q_1,q_2)=e^{q_1-q_2}$, find both partial derivatives.
2. If $A=\begin{pmatrix}0&1\\0&0\end{pmatrix}$ and $D=\begin{pmatrix}2&0\\0&3\end{pmatrix}$, calculate $AD-DA$.
3. If $f(x,y)=x+y$ and $g(x,y)=2x+2y$, are their differentials independent?

Solutions and preparation

The derivatives are $V_{q_1}=e^{q_1-q_2}$ and $V_{q_2}=-e^{q_1-q_2}$. The chain rule supplies the second sign. In the Toda Hamiltonian, the force is the negative position derivative.

The only nonzero entry of $AD-DA$ is its upper-right entry, $3-2=1$, so the commutator equals $A$. Multiplication order matters even for simple matrices. Use $(AB)_{ij}=\sum_kA_{ik}B_{kj}$.

The differentials are dependent: $dg=2\,df$. Two expressions are not automatically two independent pieces of information. The later Jacobian calculation asks this same question for conserved quantities.

If a calculation was unfamiliar, follow the entry check and repair at the start of the relevant lesson. They supply the specific preparation used there.

Follow the six-step sequence

Step	Work	What you should be able to do
1	This orientation	Identify the phase space, boundary regime and learning goal.
2	Derive the open Toda equations	Obtain every endpoint force and check total momentum.
3	Build the Toda Lax pair	Verify the commutator entry by entry, including its signs and factors.
4	Extract spectral invariants	Derive trace invariants and relate them to characteristic coefficients.
5	Test independence and Poisson commutativity	Complete the missing integrability checks and diagnose a misleading zero minor.
6	Three-site Toda: equations, invariants & evolution	Reproduce a trajectory, measure convergence and expose numerical blind spots.

Each of the four lessons contains a worked example, guided practice, an independent problem, and a problem that changes the setting. Attempt the task before opening its hint or solution. The project then combines the mathematical and computational skills.

Readers who already know Hamiltonian mechanics can use the Library proof directly, then complete the independent and transfer exercises. The Lax reference keeps finite matrices, field compatibility and boundary conditions distinct. The bibliography identifies the source editions and provides citation export.

Demonstrate what you have learned

Keep a short calculation sheet and your numerical output. A successful completion should answer all of these questions.

• Can you derive the equations without importing a periodic closing bond?
• Can you recover the equations from $\dot L=[B,L]$ with the same convention?
• Can you calculate $P$, $H$ and $I_3$, show their brackets vanish, and exhibit a nonzero three-by-three Jacobian minor?
• Can you explain why an isospectral matrix alone does not prove Liouville integrability?
• Can you reproduce second-order numerical convergence against an independent reference, including nonsymmetric initial data?
• Can you predict the effect of translating and boosting the chain, and identify what its spectral invariants fail to record?

Use the solutions to locate a missing step, then redo that step with the unequal initial data supplied in the lessons. Reading every page is useful preparation; a calculation you can explain and transfer to a changed problem is the completion evidence.

Where this sequence leads

The same questions recur in more advanced work: how to build useful spectral data, which Poisson structure makes them commute, and what information is needed to reconstruct the physical solution. This sequence establishes those questions in a finite open chain. Periodic Toda, continuum limits, inverse scattering on the line and quantum chains require additional structures and hypotheses; the present proof does not automatically cover them.

The wider Classical Integrability overview maps future topics. Continue now with the equations of motion, or inspect the three-particle experiment to see the destination.