The Architecture of the Theory

The architecture of quantum mechanics is best understood as a set of questions and the mathematical structures that answer them. This prevents a common mistake: treating Hilbert spaces, operators, measurements, and Hamiltonians as separate inventions rather than parts of one framework.

The Organizing Questions

Question	Quantum structure
What is the state?	vector, ray, or density operator
What can be measured?	observable, PVM, POVM, or instrument
How does it evolve?	Hamiltonian, unitary evolution, channel
How do systems combine?	tensor product
What is symmetry?	unitary or antiunitary representation
What creates uncertainty?	noncommutativity and state-dependent variances
What creates classical behavior?	limits, approximation, decoherence, and coarse graining

The details are developed in Core Formalism. This page gives the architecture before the machinery.

What Is the State?

A quantum state is the object used to compute probabilities for future measurements given a preparation. In the simplest pure-state language, a state is represented by a vector $\lvert\psi\rangle$, with physically equivalent vectors differing by a nonzero complex scale. More carefully, pure physical states are rays in projective Hilbert space.

Mixed states and subsystem states are represented by density operators $\rho$. They are essential for statistical mixtures, entanglement with unobserved degrees of freedom, thermal states, and open systems.

Canonical pages:

• Quantum States
• Pure States
• Density Operators

What Can Be Measured?

In the standard sharp case, an observable is represented by a self-adjoint operator with a spectral decomposition. The eigenvalues label possible outcomes, while the spectral projectors determine probabilities and ideal state updates.

More general measurements use positive operator-valued measures and instruments. That distinction matters: a POVM can specify outcome probabilities without specifying the post-measurement state.

Canonical pages:

• Observables
• Projectors
• Projective Measurement
• POVMs: First Encounter

How Does It Evolve?

For a closed system with time-independent Hamiltonian $H$, evolution is generated by

$$U(t)=\exp(-iHt/\hbar).$$

The state evolves by

$$\lvert\psi(t)\rangle = U(t)\lvert\psi(0)\rangle.$$

The same idea becomes more elaborate for time-dependent Hamiltonians, interaction pictures, propagators, path integrals, and open systems. The architectural point remains: dynamics is not an add-on. It is generated by the Hamiltonian or, for open systems, by a map or generator that describes the effective evolution of a subsystem.

Canonical pages:

• Hamiltonians
• Unitary Time Evolution
• Schrodinger Equation

How Do Systems Combine?

Composite systems use tensor products. If system $A$ has Hilbert space $\mathcal H_A$ and system $B$ has Hilbert space $\mathcal H_B$, the composite system is represented on

$$\mathcal H_A\otimes\mathcal H_B.$$

This is not a Cartesian product. It contains product states, but also entangled states that cannot be written as products. The tensor-product rule is therefore the architectural source of entanglement, subsystem states, reduced density matrices, quantum information, and many-body Hilbert spaces.

Canonical pages:

• Tensor Products
• Composite Systems
• Entangled States

What Is Symmetry?

A symmetry is represented by a transformation that preserves physical transition probabilities. In standard quantum mechanics, this leads to unitary or antiunitary transformations. Continuous symmetries have generators, and those generators are tied to conservation laws.

Symmetry is why angular momentum, spin, degeneracy, selection rules, and representation theory are central rather than decorative.

For the main route, use Symmetry, Angular Momentum, and Spin.

What Creates Uncertainty?

Uncertainty is not merely experimental clumsiness. It is tied to the state and to the noncommuting structure of observables. If two observables do not commute, they generally cannot be assigned simultaneously sharp values in arbitrary states.

The commutator controls the standard uncertainty relation, while the anticommutator and covariance refine the statement. The canonical starting point is Commutators.

What Creates Classical Behavior?

Classical behavior is not produced by a single switch. Several mechanisms can contribute:

• actions large compared with $\hbar$;
• narrow wave packets and Ehrenfest-type behavior;
• stationary-phase or WKB approximations;
• decoherence from environmental entanglement;
• coarse graining and limited measurement resolution;
• high occupation numbers or collective variables.

These mechanisms explain why classical models work in many regimes without making quantum mechanics disappear. Start with Classical Limit and Decoherence Preview.

Why This Architecture Matters

The architecture helps prevent category errors:

• A state is not an observable.
• A basis change is not a measurement.
• A measurement model is not automatically a detector model.
• A density matrix can describe ignorance, entanglement with an environment, or both depending on preparation.
• A formula without its assumptions is not a usable result.

Common Mistakes

• Starting with wavefunctions and forgetting the abstract state they represent.
• Treating eigenvalues as the whole measurement structure.
• Treating tensor products as ordinary pairs of states.
• Explaining uncertainty as only a measurement disturbance effect.
• Saying decoherence is the same thing as collapse.

References

• P. A. M. Dirac, The Principles of Quantum Mechanics, 4th ed., Oxford University Press, 1958.
• J. von Neumann, Mathematical Foundations of Quantum Mechanics, Princeton University Press, 1955.
• R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994.
• A. Peres, Quantum Theory: Concepts and Methods, Kluwer, 1995.

Exercises

1. Classify each object as a state, observable, evolution operator, or measurement effect: $\rho$, $H$, $U(t)$, and $E_i$.

Solution

$\rho$ is a density-operator state. $H$ is a Hamiltonian observable and generator of time evolution. $U(t)$ is a unitary evolution operator. $E_i$ is a POVM effect associated with a measurement outcome.

2. Why is a tensor product architecturally different from keeping a pair of separate subsystem states?

Solution

A pair of separate subsystem states can only describe product assignments. The tensor product also contains entangled states, which have correlations that cannot be represented by assigning each subsystem its own pure state independently.