Equivalence of Matrix and Wave Mechanics

Matrix mechanics and wave mechanics first looked like rival theories. Matrix mechanics emphasized transition quantities, noncommuting products, and spectral data. Wave mechanics emphasized wavefunctions, differential equations, and boundary conditions. Their equivalence was one of the decisive clarifications of 1926: the two languages give the same physical predictions when states, observables, and transformations are translated correctly.

This page closes the historical arc of this chapter. Dirac’s Transformation Theory explains the representation viewpoint in more detail; Equivalent Formulations gives the modern Core Formalism orientation.

Two Formulations

Matrix mechanics began from spectral transition data. A quantity such as position was represented by an array of transition elements,

$$X_{mn},$$

with multiplication

$$(XY)_{mn} = \sum_k X_{mk}Y_{kn}.$$

This multiplication is generally noncommutative. It fit the transition structure of atomic spectroscopy but looked unfamiliar because it did not picture an electron moving along a classical orbit.

Wave mechanics began from matter waves. A spinless particle was represented by a wavefunction $\psi(x,t)$ obeying a differential equation. For a time-independent Hamiltonian, the stationary problem has the form

$$\hat H\varphi_n(x) = E_n\varphi_n(x).$$

This language looked more visual. It suggested waves, nodes, boundary conditions, and differential equations. But the wavefunction also required a probability interpretation, and for many-particle systems it lives on configuration space rather than ordinary three-dimensional space.

The historical problem was not simply choosing the prettier notation. Both approaches worked. The deeper question was why they worked together.

Shared Spectra

The first sign of equivalence was spectral agreement. In wave mechanics, allowed energies appear as eigenvalues of a differential operator:

$$\hat H\varphi_n = E_n\varphi_n.$$

In matrix mechanics, choose the energy basis. The Hamiltonian is diagonal:

$$H_{mn} = \langle m\vert H\vert n\rangle = E_n\delta_{mn}.$$

The same energy numbers $E_n$ control stationary states and transition frequencies:

$$\omega_{mn} = \frac{E_m-E_n}{\hbar}.$$

This shared spectrum mattered historically because spectra were empirical facts. If wave mechanics and matrix mechanics predicted the same line positions, they were not merely aesthetically different. They were answering the same experimental questions.

But matching spectra alone is not the whole equivalence. A theory must also match transition amplitudes, expectation values, probabilities, and time evolution. That required a more systematic representation language.

Transformations and Representations

Let $\{\varphi_n(x)\}$ be an orthonormal set of energy eigenfunctions in wave mechanics. A state can be expanded as

$$\psi(x) = \sum_n c_n\varphi_n(x).$$

The coefficients are components of the same state in the energy basis:

$$c_n = \langle n\vert\psi\rangle.$$

Thus a wavefunction and a column of coefficients are not two different physical states. They are two representations of one state.

Operators translate the same way. If $A$ is an operator, its matrix elements in the energy basis are

$$A_{mn} = \langle m\vert A\vert n\rangle.$$

In coordinate representation, the same matrix element can be written schematically as

$$A_{mn} = \int \varphi_m(x)^* \bigl(A\varphi_n\bigr)(x)\,dx,$$

with the obvious modifications for higher dimensions, spin, degeneracy, and domains. This formula is the bridge: the matrix entry is computed from wavefunctions, but it represents the same operator information.

Changing basis changes the components and matrices, not the underlying state or operator. If $U$ is a unitary change of orthonormal basis, then state components and operator matrices transform consistently:

$$c' = U^\dagger c, \qquad A' = U^\dagger A U.$$

All probabilities and expectation values are preserved when the whole description is transformed together.

Hilbert-Space Unification

The later Hilbert-space viewpoint made the equivalence conceptually clean. A pure state is an abstract vector or ray. A wavefunction is a coordinate representation:

$$\psi(x) = \langle x\vert\psi\rangle.$$

A matrix is an operator representation in a chosen basis:

$$A_{mn} = \langle e_m\vert A\vert e_n\rangle.$$

The physical predictions come from representation-independent pairings such as amplitudes and expectation values:

$$\langle \phi\vert\psi\rangle, \qquad \langle\psi\vert A\vert\psi\rangle.$$

This unification did not erase the usefulness of the old languages. It explained them. Matrix mechanics is especially natural for finite-dimensional systems, angular momentum, perturbation theory, and transitions. Wave mechanics is especially natural for spatial motion, bound-state differential equations, scattering, and semiclassical intuition. Both are windows into the same state-operator framework.

Von Neumann’s later mathematical formalization clarified Hilbert spaces, self-adjoint operators, spectral theory, and measurement postulates. Dirac’s transformation theory supplied a powerful physical notation and conceptual grammar. Together they made it clear why the early formulations were not separate theories.

Modern Significance

The equivalence of matrix and wave mechanics remains more than a historical curiosity.

It teaches several habits that still matter:

• do not identify a state with one representation of the state;
• do not identify an operator with one matrix representation;
• choose the representation that makes the problem transparent;
• translate states and observables together when changing representation;
• check domains, boundary conditions, and spectra before claiming equivalence in infinite-dimensional systems.

For example, solving a bound-state problem in coordinate space may produce eigenfunctions $\varphi_n(x)$. Computing transition amplitudes then often means forming matrix elements such as $\langle m\vert A\vert n\rangle$. The calculation moves freely between wave mechanics and matrix mechanics because both are representations of the same formal structure.

The lesson also prepares readers for later formulations. The Schrödinger picture, Heisenberg picture, interaction picture, path integrals, phase-space methods, and density-operator language can all be equivalent in their proper domains. Equivalence means preservation of predictions after a disciplined translation, not superficial similarity of formulas.

Historical Cautions

The equivalence was not a single moment when all interpretive questions disappeared. Wave mechanics still needed Born’s probability interpretation. Matrix mechanics still needed a clearer state language. Continuous spectra and unbounded operators required mathematical care. Spin, identical particles, relativistic quantum theory, and quantum fields extended the framework beyond the first matrix-wave comparison.

It is also misleading to say wave mechanics “proved” matrix mechanics wrong because wavefunctions are easier to visualize. Wave mechanics made many calculations more accessible, but the noncommutative operator structure discovered in matrix mechanics survived as the backbone of the theory.

The historically stable conclusion is more precise: matrix mechanics and wave mechanics were different representations of one emerging quantum mechanics, not competing final ontologies.

Common Mistakes

• Treating wave mechanics as the real theory and matrix mechanics as obsolete notation.
• Treating matrix mechanics as a finite-dimensional approximation to wave mechanics. Matrix elements and operator algebra also organize infinite-dimensional systems.
• Saying two formulations are equivalent just because they use the same symbol $H$. The states, observables, inner products, domains, and boundary conditions must also match.
• Confusing equality of spectra with full equivalence of predictions.
• Thinking a wavefunction is basis-independent. It is a representation in a continuous basis.
• Thinking a matrix is basis-independent. It is a representation of an operator in a chosen basis.
• Forgetting that equivalence can fail if approximations, truncations, or boundary conditions are changed silently.

Cross-Links

• Matrix Mechanics and Operator Ideas
• Heisenberg’s Matrix Mechanics
• Born and Jordan’s Matrix Formulation
• Dirac’s Transformation Theory
• Observables Before Operators
• Commutation Relations in Historical Context
• Historical Origin of Uncertainty
• Schrödinger’s Wave Mechanics
• Interpreting the Wavefunction
• Equivalent Formulations
• Bases and Representations
• Wavefunctions as Representations
• Operator Representations
• Change of Basis
• Representation Translation Table

References

• E. Schrödinger, “Über das Verhältnis der Heisenberg-Born-Jordanschen Quantenmechanik zu der meinen,” Annalen der Physik 79, 734-756, 1926.
• E. Schrödinger, “Quantisierung als Eigenwertproblem,” Annalen der Physik 79, 361-376, 1926, DOI: 10.1002/andp.19263840404.
• W. Heisenberg, “Über quantentheoretische Umdeutung kinematischer und mechanischer Beziehungen,” Zeitschrift für Physik 33, 879-893, 1925, DOI: 10.1007/BF01328377.
• M. Born and P. Jordan, “Zur Quantenmechanik,” Zeitschrift für Physik 34, 858-888, 1925, DOI: 10.1007/BF01328531.
• P. A. M. Dirac, “The physical interpretation of the quantum dynamics,” Proceedings of the Royal Society A 113, 621-641, 1927, DOI: 10.1098/rspa.1927.0012.
• J. von Neumann, Mathematical Foundations of Quantum Mechanics, Princeton University Press, 1955.
• B. L. van der Waerden, ed., Sources of Quantum Mechanics, Dover, 1968.
• M. Jammer, The Conceptual Development of Quantum Mechanics, 2nd ed., American Institute of Physics, 1989.
• J. Mehra and H. Rechenberg, The Historical Development of Quantum Theory, Springer, 1982-2001.

Exercises

1. Suppose $\{\varphi_n(x)\}$ is an orthonormal basis and

$$\psi(x) = \sum_n c_n\varphi_n(x).$$

Show that normalization of $\psi$ implies $\sum_n\lvert c_n\rvert^2=1$.

Solution

Using orthonormality,

$$\int \varphi_m(x)^*\varphi_n(x)\,dx = \delta_{mn}.$$

Then

$$\int \lvert\psi(x)\rvert^2\,dx = \sum_{m,n} c_m^*c_n \int \varphi_m(x)^*\varphi_n(x)\,dx = \sum_n \lvert c_n\rvert^2.$$

If $\psi$ is normalized, the integral equals $1$, so $\sum_n\lvert c_n\rvert^2=1$.

2. Write the matrix element $A_{mn}$ of an operator $A$ in terms of coordinate-space wavefunctions.

Solution

If $\varphi_m$ and $\varphi_n$ are the coordinate representations of the basis states, then

$$A_{mn} = \langle m\vert A\vert n\rangle = \int \varphi_m(x)^* \bigl(A\varphi_n\bigr)(x)\,dx,$$

assuming the operator and functions lie in the needed domains.

3. Why does equality of energy spectra alone not prove full equivalence of two formulations?

Solution

Energy spectra test only one part of the theory. Full equivalence also requires matching states, observables, transition amplitudes, probabilities, time evolution, inner products, and boundary or domain assumptions. Two descriptions might agree on energy eigenvalues but disagree on matrix elements or measurement predictions if the translation is incomplete.

4. Explain why $\psi(x)$ and the component list $c_n$ can represent the same quantum state.

Solution

$\psi(x)=\langle x\vert\psi\rangle$ is the position representation of the abstract state. The components $c_n=\langle n\vert\psi\rangle$ are the representation of the same state in another basis. They are related by a basis transformation, for example

$$\psi(x) = \sum_n c_n\varphi_n(x),$$

when $\varphi_n(x)=\langle x\vert n\rangle$. The physical state is the object being represented, not either coordinate list alone.