Algebraic and Geometric Formulations

Algebraic and geometric quantum mechanics reorganize familiar Hilbert-space theory around different primary structures.

The algebraic formulation begins with an observable algebra $\mathcal A$. States are normalized positive functionals

$$\omega:\mathcal A\to\mathbb C,$$

and autonomous reversible dynamics is a group of *-automorphisms

$$\alpha_t:\mathcal A\to\mathcal A.$$

The geometric formulation begins with the projective space $\mathbb P(\mathcal H)$ of pure-state rays. Transition probabilities define the Fubini–Study metric, the imaginary part of the inner product defines a symplectic form, and Schrödinger evolution is Hamiltonian flow generated by the energy expectation

$$h([\psi]) = \frac{\langle\psi\vert H\vert\psi\rangle}{\langle\psi\vert\psi\rangle}.$$

Neither formulation changes the predictions of ordinary quantum mechanics. They reveal different structures that become useful in different regimes: algebraic language for mixed states, infinite systems, equilibrium, sectors, and QFT; geometric language for pure-state distinguishability, Hamiltonian flow, variational manifolds, and geometric phase.

Helpful background. Operators and density operators supply the algebraic examples; Heisenberg evolution shows how unitary dynamics acts on observables; rays and global phase explain the geometric starting point.

States as expectation rules

The algebraic formulation packages a system as three layers:

$$\begin{gathered} \mathcal A \quad \text{observable algebra}, \\ \omega\in S(\mathcal A) \quad \text{state}, \\ \alpha_t\in\operatorname{Aut}(\mathcal A) \quad \text{dynamics}. \end{gathered}$$

For a finite $n$-level system with the full matrix algebra,

$$\mathcal A=M_n(\mathbb C),$$

and every state is represented by a unique density matrix:

$$\omega_\rho(A) = \operatorname{Tr}(\rho A), \qquad \rho\geq0, \qquad \operatorname{Tr}\rho=1.$$

For a time-independent Hamiltonian, unitary Heisenberg evolution acts by

$$\alpha_t(A) = U_t^*AU_t.$$

Its generator is the derivation

$$\delta(A) = \frac{i}{\hbar}[H,A]$$

when a Hamiltonian implementation and its domains are under control.

The finite matrix formulas are familiar. The conceptual payoff appears when there is no preferred representation, when inequivalent representations describe different sectors, or when a global Hamiltonian is not an element of the bounded observable algebra.

Geometric Dictionary

For the full observable algebra on a finite-dimensional Hilbert space, the geometric formulation organizes pure-state theory as

$$\begin{gathered} \mathbb P(\mathcal H) \quad \text{pure-state manifold}, \\ g_{\rm FS} \quad \text{distinguishability metric}, \\ \omega_{\rm FS} \quad \text{symplectic form}, \\ f_A([\psi]) = \langle A\rangle_\psi \quad \text{observable function}. \end{gathered}$$

For normalized representatives, with the distance convention used in this chapter,

$$d_{\rm FS}([\phi],[\psi]) = \arccos \left| \langle\phi\vert\psi\rangle \right|.$$

At a normalized representative $\psi$, horizontal tangents obey $\langle\psi|u\rangle=\langle\psi|v\rangle=0$. We use

$$\omega_{\rm FS}(u,v) = 2\hbar\, \operatorname{Im} \langle u\vert v\rangle.$$

With the convention $\iota_{X_h}\omega_{\rm FS}=dh$, the expectation function of $H$ generates the horizontal Schrödinger velocity:

$$X_h([\psi]) \quad\text{is represented by}\quad -\frac{i}{\hbar} \left( H-\langle H\rangle_\psi \right) \lvert\psi\rangle.$$

The induced Poisson bracket is

$$\{f_A,f_B\}_{\rm FS} = \frac{1}{i\hbar} f_{[A,B]}.$$

This is an exact rewriting of pure-state quantum mechanics, not a classical-limit approximation.

Where the Formulations Meet

For a finite system, let

$$\Phi_t([\psi]) = [U_t\psi]$$

be the projective state flow and let

$$\alpha_t(A) = U_t^*AU_t$$

be the observable automorphism. Then

$$\begin{aligned} f_A(\Phi_t([\psi])) &= \langle U_t\psi\vert A \vert U_t\psi\rangle\\ &= \langle\psi\vert U_t^*AU_t \vert\psi\rangle\\ &= f_{\alpha_t(A)}([\psi]). \end{aligned}$$

The same expectation value can therefore be read in two ways:

• move the state point through projective Hilbert space and keep $A$ fixed;

• move $A$ through the observable algebra and keep the state fixed.

This is the geometric/algebraic version of Schrödinger–Heisenberg duality.

A ray also defines an algebraic vector state,

$$\omega_\psi(A) = \langle\psi\vert A\vert\psi\rangle.$$

The expectation function is simply evaluation of that state:

$$f_A([\psi])=\omega_\psi(A).$$

For the full matrix algebra these vector states are pure. Restricting a vector state to a smaller observable algebra can make it mixed; an entangled bipartite vector restricted to one subsystem is an elementary example. The projective formulation studies the geometry of the rays themselves. The algebraic formulation includes them but also admits mixed, thermal, and representation-dependent states that need not lie in one projective Hilbert space.

Reading Path

| Question | Start here | Main idea |

|---|---|---|

| What changes in the algebraic viewpoint? | Algebraic Formulation Overview | Algebra, state, and automorphism are taken as primary. |

| Which elements and operations form the observable algebra? | Observables as an Algebra | Self-adjoint observables sit inside a full complex *-algebra. |

| How is a state defined without choosing a vector? | States as Positive Linear Functionals | Positivity and normalization produce expectations and probabilities. |

| How is time evolution stated abstractly? | Dynamics as Automorphisms | A C*-dynamical system is a continuous automorphism group. |

| What changes in the geometric viewpoint? | Geometric Quantum Mechanics Overview | Rays carry metric, symplectic, and phase-bundle structures. |

| Why are rays the physical pure states? | Projective Hilbert Space | Projectors and horizontal tangents remove global phase. |

| How are pure-state distances measured? | Fubini–Study Geometry | Transition probabilities define distance and quantum speed. |

| Why is Schrödinger evolution Hamiltonian? | Hamiltonian Flow on Projective Hilbert Space | Energy expectation generates the projective symplectic flow. |

| How does projective geometry become Berry geometry? | Relation to Berry Geometry | Eigenstate families pull the universal connection and curvature back to parameter space. |

Choosing a Formulation

Use algebraic language when the main objects are observables, expectation functionals, thermal states, local algebras, thermodynamic limits, superselection sectors, or inequivalent representations.

Use geometric language when the question concerns pure-state distance, phase-free state motion, energy uncertainty as speed, variational state manifolds, unitary control, or geometric phase.

Use density operators and quantum channels for mixed-state and open-system calculations. Pure projective geometry does not replace the convex geometry of density matrices, and closed automorphisms do not replace completely positive maps.

Use the standard Hilbert-space formulation whenever it is the shortest route. These formulations are structural lenses, not requirements that every finite matrix calculation be rewritten.

Exercises

Consider the normalized qubit states $|0\rangle$ and $|+\rangle=(|0\rangle+|1\rangle)/\sqrt2$. Find their Fubini–Study distance. Compare the result with the angle between their Bloch vectors.

Solution

The overlap has magnitude $1/\sqrt2$, so $d_{\rm FS}=\pi/4$. Their Bloch vectors point along $z$ and $x$, with angle $\pi/2$. With the stated metric convention the pure-qubit sphere has radius $1/2$: projective distance is half the Bloch-vector angle. The transition probability is $\cos^2d_{\rm FS}=1/2$.

Why does projective Hilbert space not contain every algebraic state?

Solution

Projective Hilbert space consists of pure vector rays in a chosen Hilbert-space representation. Algebraic states are all normalized positive functionals on the observable algebra. Even in finite dimensions, mixed density matrices define algebraic states that are not points of projective Hilbert space.

In infinite dimensions, the distinction is stronger. Algebraic states may be nonnormal in a chosen representation or may generate inequivalent GNS representations. No single preselected projective Hilbert space need contain all of them as vector rays.

References

• A. Ashtekar and T. A. Schilling, “Geometrical formulation of quantum mechanics,” in On Einstein’s Path, Springer, 1999.

• I. Bengtsson and K. Życzkowski, Geometry of Quantum States, 2nd ed., Cambridge University Press, 2017.

• O. Bratteli and D. W. Robinson, Operator Algebras and Quantum Statistical Mechanics 1, 2nd ed., Springer, 1987.

• R. Haag, Local Quantum Physics: Fields, Particles, Algebras, 2nd ed., Springer, 1996.

• T. W. B. Kibble, “Geometrization of quantum mechanics,” Communications in Mathematical Physics 65, 189–201, 1979.

• F. Strocchi, An Introduction to the Mathematical Structure of Quantum Mechanics, 2nd ed., World Scientific, 2008.