Probability and the Born Rule

Quantum theory predicts probability distributions for specified measurements on specified states. The Born rule connects these two inputs: a state supplies amplitudes or a density operator, a measurement supplies projectors or more general effects, and their pairing gives normalized probabilities.

This chapter is the canonical home for that probability rule and its immediate statistical consequences. Its central discipline is to keep amplitudes, probabilities, probability densities, expectation values, and state-update rules distinct. Each answers a different question.

Required background. State Vectors supplies normalized states and basis amplitudes; Projectors supplies the operator for a sharp event.

Helpful background. Probability Spaces: A Light Introduction supplies events, normalized measures, means, and variances.

What This Chapter Owns

This chapter develops

• complex probability amplitudes and interference;
• the Born rule for pure states, density operators, and spectral projectors;
• practical calculations for discrete spectra, including degeneracy;
• probability densities and interval probabilities for continuous spectra;
• expectation values, moments, variance, and standard deviation;
• correlations, symmetrized covariance, and connected correlations;
• static and dynamical transition probabilities;
• probability calculations after a change of basis.

Classical probability spaces, conditional probability, random variables, and information measures belong to Probability and Information. Conditioned post-measurement states belong to Measurement and State Update. The rule’s development belongs to Born Rule: Historical Context. Interpretations and proposed derivations of the Born rule belong to Foundations and Interpretations; they are distinct from the probability rule used in this chapter.

State, Event, and Probability

A probability statement is incomplete until both the preparation and the measured event are specified. For a normalized pure state $\lvert\psi\rangle$ and a projective outcome represented by $P_a$,

$$p(a\mid\psi) = \langle\psi\rvert P_a\lvert\psi\rangle.$$

For a density operator $\rho$,

$$p(a\mid\rho) = \operatorname{Tr}(\rho P_a).$$

If the outcome is nondegenerate, $P_a=\lvert a\rangle\langle a\rvert$, so

$$p(a\mid\psi) = \bigl\lvert\langle a\rvert\psi\rangle\bigr\rvert^2.$$

The state alone does not determine a list of probabilities independent of measurement context. The same state gives different distributions for spin measured along different axes or for position and momentum. The measurement event is equally part of the probability assignment.

For a complete discrete projective measurement,

$$P_aP_b = \delta_{ab}P_a, \qquad \sum_aP_a=I.$$

These relations imply normalization:

$$\sum_a p(a\mid\rho) = \operatorname{Tr}\!\left(\rho\sum_aP_a\right) = \operatorname{Tr}\rho = 1.$$

The broader generalized-measurement rule replaces the projectors by positive effects $E_a$ satisfying $\sum_aE_a=I$, giving $p(a)=\operatorname{Tr}(\rho E_a)$. The detailed theory of POVMs and instruments belongs to Measurement and Open Quantum Systems. The Born rule remains the state-measurement pairing that produces probabilities.

Amplitudes and Interference

An amplitude is a complex number, not a probability. For a transition from $\lvert\psi\rangle$ to $\lvert\phi\rangle$, the amplitude is

$$\mathcal A_{\psi\to\phi} = \langle\phi\rvert\psi\rangle,$$

and the corresponding probability is its squared modulus:

$$P_{\psi\to\phi} = \bigl\lvert\mathcal A_{\psi\to\phi}\bigr\rvert^2.$$

When indistinguishable alternatives lead coherently to the same outcome, their amplitudes add before the modulus is squared. For two alternatives,

$$\bigl\lvert\mathcal A_1+\mathcal A_2\bigr\rvert^2 = \lvert\mathcal A_1\rvert^2 + \lvert\mathcal A_2\rvert^2 + 2\operatorname{Re} \!\left(\mathcal A_1^*\mathcal A_2\right).$$

The cross term carries relative-phase information. Probabilities add directly when the alternatives are operationally distinguished, incoherent, or represented as mutually exclusive events in the relevant measurement description. Deciding which rule applies requires a physical account of the alternatives, not merely algebraic preference.

Discrete and Continuous Outcomes

For a discrete nondegenerate eigenbasis $\{\lvert a_n\rangle\}$, expand

$$\lvert\psi\rangle = \sum_n c_n\lvert a_n\rangle, \qquad c_n=\langle a_n\rvert\psi\rangle.$$

Then $p_n=\lvert c_n\rvert^2$. For a degenerate value, sum probabilities over an orthonormal basis of the eigenspace or, more cleanly, use its spectral projector. The result cannot depend on which basis is chosen inside that eigenspace.

For an absolutely continuous position distribution,

$$p_\psi(x) = \lvert\psi(x)\rvert^2$$

is a probability density. The probability for an interval is

$$\Pr_\psi(x\in[a,b]) = \int_a^b dx\, \lvert\psi(x)\rvert^2.$$

A density has units inverse to its integration variable and need not be less than one at every point. For a purely continuous distribution, an exact point has probability zero even though the density there may be nonzero. Mixed spectra require both discrete masses and continuous contributions, most cleanly expressed with spectral projectors for measurable sets.

Probability densities transform with the measure. If $y=f(x)$ is one-to-one on the region of interest, then

$$p_y(y) = p_x(x(y)) \left\lvert\frac{dx}{dy}\right\rvert.$$

This Jacobian is why a wavefunction component and a probability density must always be interpreted together with their coordinate convention and measure.

Moments and Statistical Summaries

For a sharp observable $A$, the expectation value in a pure state is

$$\langle A\rangle_\psi = \langle\psi\rvert A\lvert\psi\rangle,$$

and in a density operator it is

$$\langle A\rangle_\rho = \operatorname{Tr}(\rho A).$$

An expectation value is the probability-weighted ensemble mean, not generally a possible single-shot result and not necessarily the most probable result. Higher moments summarize more of the distribution. In particular,

$$(\Delta_\rho A)^2 = \langle A^2\rangle_\rho - \langle A\rangle_\rho^2.$$

Zero variance means the state has support in a single eigenspace of $A$. It does not require that eigenspace to be one-dimensional.

For two self-adjoint observables, a real symmetrized covariance is

$$\operatorname{Cov}_\rho(A,B) = \frac12 \left\langle \Delta A\,\Delta B + \Delta B\,\Delta A \right\rangle_\rho,$$

where $\Delta A=A-\langle A\rangle_\rho I$. If $A$ and $B$ do not commute, ordered products can differ and need not define an ordinary classical joint moment. The experimental protocol and operator ordering must therefore be stated.

Transitions and Basis Choice

After unitary evolution from $t_0$ to $t$, the probability of finding a final state $\lvert\phi\rangle$ is

$$P_{\psi\to\phi}(t,t_0) = \bigl\lvert \langle\phi\rvert U(t,t_0) \lvert\psi\rangle \bigr\rvert^2.$$

This is a Born-rule probability after dynamics, not by itself a rule for measurement backaction. Transition rates, perturbative approximations, and scattering observables add further limiting procedures beyond this finite-time probability statement.

A change of coordinate basis cannot change a physical probability. Changing which observable is measured can. If state components are transformed while the measurement projectors are transformed consistently, the trace pairing remains invariant. If the state is fixed but a different measurement basis is selected, the experiment and generally the outcome distribution change.

Page Map

Question	Canonical page	Main distinction
What is a probability amplitude?	Probability Amplitudes	complex amplitude versus real probability
What is the general probability rule?	Born Rule	state-measurement pairing versus interpretation
How are finite or countable outcomes handled?	Born Rule for Discrete Spectra	coefficients versus degenerate projectors
How are densities and intervals handled?	Born Rule for Continuous Spectra	density versus point probability
What does an average predict?	Expectation Values	ensemble mean versus single outcome
How is outcome spread quantified?	Variance and Standard Deviation	distribution width versus measurement error
How are two-observable statistics described?	Correlations and Covariance	symmetrized covariance versus ordered correlator
How likely is a final state or subspace?	Transition Probabilities	overlap after evolution versus update rule
How is the measurement basis used?	Probability in Different Bases	coordinate change versus changed measurement

These nine articles form the planned chapter.

Suggested Routes

First systematic pass

Read Probability Amplitudes, Born Rule, and Born Rule for Discrete Spectra. Then study Expectation Values and Variance and Standard Deviation. The planned next step is General Uncertainty Relations.

The specialist routes below assume this first systematic pass through variance, in addition to the chapter’s Required background. They name where to branch, not substitutes for those prerequisites.

Wave-mechanics route

Read Born Rule for Continuous Spectra, Expectation Values, Variance and Standard Deviation, and Probability in Different Bases. The planned wave-mechanics extension pairs them with Normalization, Wavefunctions as Representations, and Momentum-Space Representation before canonical systems.

Quantum-information route

Read Probability in Different Bases, Correlations and Covariance, and Transition Probabilities. Continue to Density Operators, Entangled States, Partial Trace, and Quantum Operations. For generalized measurement theory, continue in Measurement and Open Quantum Systems.

Dynamics and scattering route

Read Transition Probabilities after Born Rule. The planned dynamics and scattering extension then continues to Unitary Time Evolution and First-Order Transition Probability.

Probability Sanity Checks

• Specify the state, measurement, and outcome event before writing a probability.
• Check positivity and total normalization.
• For a degenerate outcome, verify that the answer is invariant under basis rotations within its eigenspace.
• Attach the correct measure and units to every continuous density.
• Distinguish coherent addition of amplitudes from addition of exclusive probabilities.
• Confirm that expectation values and variances are real for self-adjoint observables.
• Test basis invariance by transforming states and measurement operators together.
• In numerical work, monitor normalization, positivity, grid measure, cutoff dependence, and convergence.

Common Mistakes

• Squaring an amplitude without specifying the outcome it refers to. Amplitudes are conditional on initial and final alternatives.
• Adding probabilities for coherent indistinguishable alternatives. Their amplitudes must be combined first.
• Treating a density as a point probability. Continuous probabilities require integration over a set.
• Forgetting degeneracy. A measurement outcome can correspond to a multidimensional eigenspace.
• Calling an expectation value the most likely result. Mean, mode, and allowed outcomes are different concepts.
• Calling quantum variance apparatus error. It describes spread in the state-defined outcome distribution; detector noise is additional.
• Assuming noncommuting observables possess an ordinary joint distribution. Ordering and protocol matter.
• Confusing transition probability with state update. Unitary evolution plus a final event gives a probability, not the conditioned post-measurement map.
• Changing basis components inconsistently. States and measurement operators must use the same representation.

Cross-Links

• States and Representations
• Observables and Operators
• Compatibility, Commutators, and Uncertainty
• Measurement and State Update
• Density Operators and Mixed States
• Probability and Information
• Born Rule glossary entry
• Born Rule formula entry

References

• L. E. Ballentine, Quantum Mechanics: A Modern Development, 2nd ed., World Scientific, 2014.
• M. Born, “Zur Quantenmechanik der Stoßvorgänge,” Zeitschrift für Physik 37, 863–867, 1926, doi:10.1007/BF01397477.
• P. A. M. Dirac, The Principles of Quantum Mechanics, 4th ed., Oxford University Press, 1958.
• A. Peres, Quantum Theory: Concepts and Methods, Kluwer, 1995.
• R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994.
• J. von Neumann, Mathematical Foundations of Quantum Mechanics, Princeton University Press, 1955.