---
title: Pusz–Woronowicz Interpolation Theory
url: https://www.emergentmind.com/topics/pusz-woronowicz-interpolation-theory
type: topic
---

# Pusz–Woronowicz Interpolation Theory

The Pusz–Woronowicz interpolation theory provides a foundational framework for constructing and analyzing interpolations (or "means") of positive sesquilinear (and, in particular, quadratic) forms on finite-dimensional vector spaces. Originating from functional analysis and operator algebras, this theory underpins modern approaches to the monotonicity of quantum relative entropy—a result central to quantum information theory—by providing a structured and order-theoretic machinery for positive forms and their interpolations. Uhlmann's proof of the data-processing inequality for quantum relative entropy is the canonical application, leveraging the Pusz–Woronowicz interpolation to circumvent analytic subtleties of more direct operator approaches [2509.11221], [2210.13093].

## 1. Foundations of the Pusz–Woronowicz Interpolation Theory

Let $V$ be a finite-dimensional complex vector space. A sesquilinear form $\alpha: V \times V \to \mathbb{C}$ is positive if $\alpha(v, v) \geq 0$ for all $v \in V$. The set $\mathscr{F}_+(V)$ denotes the cone of positive sesquilinear forms on $V$, partially ordered by $\beta \leq \alpha$ if and only if $\alpha - \beta$ is positive. 

Given two such forms, the Pusz–Woronowicz theory constructs a family of "means"—interpolating forms—between them, parameterized by $t \in [0,1]$. Fixing a basis, any $\alpha \in \mathscr{F}_+(V)$ corresponds uniquely to a positive semidefinite operator $T$ via $\alpha(v, w) = v^\dagger T w$. For two forms $\alpha$ and $\beta$, the theory guarantees the existence of a Hilbert space $H$, a surjection $h: V \to H$, and a commuting pair of positive operators $A, B$ on $H$ such that $\alpha(v, w) = \langle h(v), A h(w) \rangle$ and $\beta(v, w) = \langle h(v), B h(w) \rangle$ [2509.11221].

The functional calculus on such commuting pairs enables the definition of new forms via operator functions $f(A, B)$. The central interpolation is given by $f_t(x, y) = x^{1-t} y^t$ for $t \in [0,1]$, yielding the interpolated form
$$
\gamma^t_{\alpha \to \beta}(v, w) = \langle h(v), A^{1-t} B^t h(w) \rangle.
$$

## 2. Structural Properties and Inequalities

Key order-theoretic and extremal properties of the interpolated forms include:

- **Ordering of Forms:** If $\alpha' \leq \alpha$ and $\beta' \leq \beta$, then for all $t \in [0,1]$, 
  $$
  \gamma^t_{\alpha' \to \beta'} \leq \gamma^t_{\alpha \to \beta} 
  $$
  [2509.11221].
- **Pull-back Monotonicity:** For a linear map $\psi: U \to V$ and forms $\alpha, \beta \in \mathscr{F}_+(V)$,
  $$
  \psi^*(\gamma^t_{\alpha \to \beta}) \leq \gamma^t_{\psi^* \alpha \to \psi^* \beta}
  $$
  where $\psi^* \alpha (u, u') := \alpha(\psi(u), \psi(u'))$ [2509.11221].
- **Geometric Mean Extremality:** The mid-point form $\sqrt{\alpha \beta} := \gamma^{1/2}_{\alpha \to \beta}$ satisfies
  $$
  |\sqrt{\alpha \beta}(v, w)|^2 \leq \alpha(v, v) \beta(w, w)
  $$
  and is extremal with respect to this inequality [2509.11221].

These properties establish a strong compatibility between algebraic, order, and analytic structures on the cone $\mathscr{F}_+(V)$.

## 3. Application to Uhlmann’s Monotonicity Theorem

Quantum relative entropy between density operators $\rho$ and $\sigma$ is defined as $D(\rho \| \sigma) := \operatorname{Tr}[\rho \log \rho - \rho \log \sigma]$ when $\mathrm{supp}\,\rho \subseteq \mathrm{supp}\,\sigma$, and $+\infty$ otherwise [2210.13093]. Uhlmann’s proof of the monotonicity relation
$$
D(\Phi(\rho) \| \Phi(\sigma)) \leq D(\rho \| \sigma)
$$
for completely positive trace-preserving (CPTP) maps $\Phi$ constructs two positive forms on $V = \mathcal{B}(H_{ab})$:
- $\rho_L(A, B) = \operatorname{Tr}[\rho\, B\, A^\dagger]$,
- $\sigma_R(A, B) = \operatorname{Tr}[\sigma\, A^\dagger B]$,

with associated left and right multiplication operators $L_\rho$ and $R_\sigma$ that commute. The $t$-interpolation is then 
$$
\gamma^t_{\rho_L \to \sigma_R}(A, B) = \operatorname{Tr}[\rho^{1-t} A \sigma^t B^\dagger].
$$

Relative entropy is recovered via differentiation at $t=0$ of the interpolated forms:
$$
S_{\rho \| \sigma}(A, B) = -\liminf_{t \to 0^+} \frac{\gamma^t_{\rho_L \to \sigma_R}(A, B) - \rho_L(A, B)}{t}
$$
with $S_{\rho \| \sigma}(I, I) = D(\rho \| \sigma)$ [2509.11221].

Pull-back and ordering monotonicities yield
$$
D(\Phi(\rho) \| \Phi(\sigma)) \leq D(\rho \| \sigma)
$$
establishing the data-processing inequality for quantum relative entropy [2210.13093].

## 4. Extensions and Regularization

Pusz–Woronowicz interpolation naturally extends to non-invertible density operators. When $\mathrm{supp}\,\rho \subseteq \mathrm{supp}\,\sigma$ but either $\rho$ or $\sigma$ is not full-rank, the $t$-interpolated form may lose regularity at $t=0$. The definition then involves a lower Dini derivative (liminf), and equivalently, one may regularize by setting $\rho_\epsilon = \rho + \epsilon I$ and $\sigma_\epsilon = \sigma + \epsilon I$, applying the invertible case, and taking $\epsilon \to 0$ [2509.11221]. All pull-back and ordering arguments are stable under this limiting process.

A support-based definition is also used: $D(\rho \| \sigma) = +\infty$ if $\mathrm{supp}\,\rho \not\subseteq \mathrm{supp}\,\sigma$, making monotonicity trivial in this case [2509.11221].

## 5. Comparison: Pusz–Woronowicz Interpolation vs. Modular Theory Approaches

Petz’s proof of monotonicity relies on modular operators $\Delta_{\rho, \sigma} = L_\sigma R_\rho^{-1}$ and Jensen’s contractive operator inequality in combination with an integral formula for $-\log x$. The method requires careful treatment (notably the Petz–Nielsen repair) for operator convexity, and is seen as more direct though analytic [2509.11221]. In contrast, the Uhlmann/Pusz–Woronowicz approach functions on the order-theoretic and functional calculus structure of sesquilinear forms, thus generalizing more transparently to non-invertible cases, positive unital Schwarz maps, and arbitrary $C^*$-algebraic contexts. No use of operator convexity beyond geometric mean extremality is needed, and the argument exploits only basic order and pull-back properties of interpolated forms.

Conceptually, the abstraction of the Pusz–Woronowicz interpolation theory provides a lattice-theoretic and functional-analytic viewpoint that unifies and simplifies proofs of monotonicity and related inequalities in quantum information theory [2509.11221], [2210.13093].

## 6. Broader Context and Applications

The Pusz–Woronowicz interpolation is central to modern proofs and extensions of quantum data-processing inequalities, strong subadditivity, and the analysis of quantum quasi-entropies. It is fundamental to understanding why coarse-graining, quantum channels, or subalgebra restriction cannot increase the distinguishability of quantum states as measured by relative entropy [2210.13093]. Its abstract structure underpins stability properties of statistical inference in the quantum regime and has implications for the second law of thermodynamics and limits on information transmission.

Extensions of the Pusz–Woronowicz framework have been used to establish monotonicity for other divergences, such as quasi-entropies and Rényi relative entropies. The order-theoretic formulation supports generalizations to infinite-dimensional settings and to the full setting of $C^*$-algebras when suitable technical regularizations are made [2210.13093].

### Summary Table: Key Components of the Pusz–Woronowicz Interpolation Framework

| Component                | Description                                              | Reference              |
|--------------------------|---------------------------------------------------------|------------------------|
| Sesquilinear forms       | Positive forms on $V \times V$                          | [2509.11221]           |
| Interpolated forms       | $\gamma^t_{\alpha \to \beta}$ using operator calculus   | [2509.11221]           |
| Order/pull-back monotonicity | Fundamental inequality tools                         | [2509.11221]           |
| Geometric mean extremality | Extremal structure for the geometric mean              | [2509.11221]           |

The Pusz–Woronowicz interpolation theory is thus a central pillar in the rigorous theory of quantum information, providing a versatile toolkit for monotonicity proofs and a deeper understanding of the structural aspects of positive operator-valued maps and quantum divergences.

Source: https://www.emergentmind.com/topics/pusz-woronowicz-interpolation-theory