Purification Overview
Purification represents a mixed state as part of a larger pure state. It is a mathematical construction for tracking possible correlations, not a claim that a particular auxiliary system was present in the experiment. Purification develops the equivalence theorems and applications.
Helpful background. Reduced density matrices explains tracing out a factor, and spectral decomposition supplies the eigenvalues used below.
Building a Purifying State
Section titled “Building a Purifying State”A purification of is a normalized vector such that
In finite dimensions, write the nonzero spectral terms as . Choose an orthonormal set of reference kets and form
Its squared norm is . Tracing over kills the terms because the reference kets are orthogonal, leaving . The minimum reference dimension is the rank ; a smaller reference cannot supply enough Schmidt terms.
Nonuniqueness
Section titled “Nonuniqueness”For , one choice is
Applying any unitary to produces another purification with the same reduction. For example, flipping the reference basis gives . The reference’s labels have changed, while Alice’s local statistics have not.
More generally, purifications are related by an isometry on the occupied reference subspaces; if the reference spaces have the same finite dimension, this relation extends to a unitary. No particular purification singles out the actual preparation ensemble.
Quick Check
Section titled “Quick Check”Why would fail to purify ?
Solution
Both terms carry the same reference ket . The state factors as , so tracing out leaves a pure state with nonzero off-diagonal coherence. Distinct orthogonal reference kets are what remove those cross terms in the desired construction.
References
Section titled “References”- M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information, 10th anniversary ed., Cambridge University Press, 2010, sec. 2.5.
- J. Watrous, The Theory of Quantum Information, Cambridge University Press, 2018, ch. 2. Author’s manuscript.