---
title: Fermionic Partial Transpose (FPT) Explained
url: https://www.emergentmind.com/topics/fermionic-partial-transpose-fpt
type: topic
---

# Fermionic Partial Transpose (FPT) Explained

A fermionic partial transpose (FPT) is a linear operation on fermionic many-body density matrices and operators, designed to generalize the bosonic partial transpose for use in the context of indistinguishable fermions with anti-commuting operator algebra. The FPT incorporates fundamental sign conventions that arise from the graded tensor structure of fermionic Fock space, rendering it compatible with observables such as quantum entanglement negativity, and endowing it with a central role in the construction and classification of symmetry-protected topological (SPT) invariants—including homotopy invariants and division superalgebra indices in matrix product state (MPS) formalism. Two prominent variants of the FPT—standard and twisted—reflect the presence or absence of boundary Majorana zero modes. The operation underlies recent advances in fermionic entanglement measures and topological characterization in both free and interacting systems [2206.03260][2503.07742][1710.01886][1906.04211].

## 1. Algebraic Definition and Structural Properties

The fermionic partial transpose acts on an operator or density matrix $\rho$ by exchanging specific subsystems and applying sign factors tracking the Koszul grading of fermion operators. For a bipartition of the system into subsystems $A_1$ and $A_2$, and a decomposition of a parity-even operator $A$,
\[
A = \sum_{k_1+k_2\, \mathrm{even}} A_{p_1\cdots p_{k_1},\,q_1\cdots q_{k_2}}\, a_{p_1}\cdots a_{p_{k_1}}\, b_{q_1}\cdots b_{q_{k_2}},
\]
where $a_{p_j}$ and $b_{q_j}$ are Majorana operators in $A_1$ and $A_2$, the FPT with respect to $A_1$ is defined as
\[
A^{\top_1} = \sum_{k_1+k_2\,\mathrm{even}} A_{p_1\cdots p_{k_1},\,q_1\cdots q_{k_2}}\, i^{k_1}\, a_{p_1}\cdots a_{p_{k_1}}\, b_{q_1}\cdots b_{q_{k_2}},
\]
where the $i^{k_1}$ phase arises from graded transposition, reflecting the number of Majoranas being transposed and their statistics [1710.01886][1906.04211].

In the occupation-number basis, for basis vectors $|n^1, n^2\rangle$, the action is 
\[
|n^1, n^2\rangle\langle m^1, m^2| \mapsto i^{n^1 + m^1} (-1)^{(n^1 + m^1)(n^2 + m^2)} |m^1, n^2\rangle\langle n^1, m^2|.
\]
The standard FPT is “untwisted”; a twisted variant incorporates an additional $(-1)^{|n_A| + |n_A'|}$, reflecting the presence of boundary Majorana zero modes [2403.12121].

Key properties include:
- Linearity;
- Preservation of trace and Hermiticity (up to parity);
- Involution up to a parity flip: $(A^{\top_1})^{\top_1} = (-1)^{F_1} A (-1)^{F_1}$;
- Well-definedness only on the fermion parity-even algebra.

## 2. Physical Motivation and Comparison with the Bosonic Partial Transpose

The foundational challenge in the fermionic case is that the Hilbert space is not a simple tensor product due to anticommutation relations between creation/annihilation operators. The bosonic partial transpose acts by reordering labels in a product basis, an operation incompatible with the superselection rules of Fock space and the necessity for antisymmetrization in fermionic systems. The FPT implements a “partial” time-reversal or partial transpose that respects fermionic statistics by embedding the appropriate sign structure at the operator level [1611.07536][1906.04211].

For two-fermion antisymmetric states, the FPT can be implemented as antisymmetrization after ordinary partial transpose, 
\[
\rho^{\text{FPT}} = A\, [T_2(\rho)]\, A,
\]
where $A$ is the antisymmetrizer [1407.3088]. More generally, the sign structure can be traced to the representation of the fermionic algebra as a superalgebra and the necessity to preserve Koszul signs when factors are exchanged under a partial operation [2206.03260][1710.01886].

## 3. Variants: Standard and Twisted Fermionic Partial Transpose

The FPT admits two inequivalent constructions distinguished by spin structure at the A–B boundary:
- **Standard (untwisted) FPT**: appropriate when the A–B boundary does not carry an unpaired Majorana zero mode, the sign structure arises only from the number of reordered fermions.
- **Twisted FPT**: introduced when there is a boundary Majorana, further supplementing with a parity-dependent sign $(-1)^{|n_A| + |n_A'|}$. Algebraically, for a density matrix $\rho$ written in the occupation basis,
  \[
  \rho^{T_A}_{\rm std}: \left\langle n_A, m_B \left| \rho^{T_A}_{\rm std} \right| n_A', m_B' \right\rangle = (-1)^{\varphi(n_A, n_A')} \left\langle n_A', m_B | \rho | n_A, m_B' \right\rangle,
  \]
  \[
  \rho^{T_A}_{\rm tw}: \left\langle n_A, m_B \left| \rho^{T_A}_{\rm tw} \right| n_A', m_B' \right\rangle = (-1)^{\varphi(n_A, n_A') + |n_A| + |n_A'|} \left\langle n_A', m_B | \rho | n_A, m_B' \right\rangle,
  \]
  where $\varphi$ counts the necessary swaps [2403.12121][1906.04211].
  
The untwisted FPT is pseudo-Hermitian; its spectrum lies on six rays in the complex plane. The twisted FPT is Hermitian with a spectrum that is real, featuring both positive and negative branches [1906.04211]. The choice of variant encodes global topological information such as boundary conditions or the presence of unpaired Majorana zero modes.

## 4. Applications to Entanglement and Topological Invariants

The FPT is central to the definition and computation of the fermionic logarithmic negativity:
\[
\mathcal{E} = \ln \lVert \rho^{T_{\mathrm{F}}} \rVert_1,
\]
where $T_{\mathrm{F}}$ denotes the fermionic partial transpose [2503.07742][1502.01369]. For free or Gaussian states, the FPT does not yield a single Gaussian but, in general, a linear combination of two (for a bipartition) or four (for disjoint intervals) Gaussian operators, allowing exact computation via covariance matrices and determinant formulas [1502.01369][1503.09114].

In topological systems, particularly in one-dimensional fermionic SPT phases protected by anti-unitary symmetries, the FPT-derived string order parameters become quantized invariants:
- For particle–hole symmetry $C$, define 
  \[
  Z_{k,\ell}^C = \operatorname{Tr} \left( \sigma_{k+\ell} (\sigma_{k+\ell})^{\mathrm{t}_k} \right),
  \]
  where $\mathrm{t}_k$ is the partial graded transpose over the first $k$ sites [2206.03260].
- The invariant
  \[
  \exp\left( \frac{2\pi i}{8}\, \eta_C \right) = \lim_{k,\ell \to \infty} e^{\frac{3}{2}\Delta_{k+\ell}} Z_{k,\ell}^C
  \]
  labels a real division superalgebra $\mathbb{D}$ characterizing the MPS bond algebra structure of the phase.

For time-reversal symmetry, an analogous invariant
\[
\exp\left( \frac{2\pi i}{2}\, \eta_T \right)
\]
classifies distinct topological phases associated with $\mathbb{Z}_2$ indices.

## 5. Computational Methods and Free Fermion Implementations

In free-fermion systems, the FPT, when formulated in the overlap-matrix framework, enables an explicit calculation of the logarithmic negativity. For a bipartite pure state of $M$ modes with single-particle overlap matrix $M_A$ with eigenvalues $P_\gamma$, the FPT leads to a block-diagonal structure where each block’s trace norm is
\[
\lVert \rho_\gamma^{T^f_B} \rVert_1 = 1 + 2\sqrt{P_\gamma (1-P_\gamma)}.
\]
Thus,
\[
\mathcal{E} = \sum_{\gamma=1}^M \ln \left[1 + 2 \sqrt{P_\gamma (1-P_\gamma)} \right].
\]
This reproduces the universal scaling $\mathcal{E}(l) \sim \frac{1}{2} \ln l$ in critical 1D chains and $\mathcal{E} \sim L \ln L$ in 2D systems [2503.07742].

For mixed (disjoint interval) states, the overlap-matrix methodology fails in general, as the correct spectrum of the FPT is not globally preserved. In this regime, one must resort to Green’s function or path-integral calculations [1503.09114][1906.04211].

## 6. Non-local Order Parameters and Bulk Invariance

The FPT gives rise to non-local order parameters which become homotopy invariants under $G$-symmetric continuous deformations. In the presence of anti-unitary symmetries, the phase of these invariants is quantized and cannot change continuously, providing a bulk characterization of topological phases irrespective of microscopic details. Specifically, for a translation-invariant ground state with short-range correlations, the FPT-based order parameter converges exponentially to its quantized invariant, robust under any $G$-symmetric perturbation [2206.03260]. The invariants extracted via FPT are precisely those labeling classes in the group cohomology or superalgebraic classification of 1D fermionic SPT phases.

## 7. Connections, Limitations, and Outlook

The FPT is algebraically equivalent, in certain cases, to the matrix realignment criterion, and thus serves as a general separability and entanglement detector for indistinguishable particles [1407.3088]. However, for pure fermionic states with definite global parity, certain entangled states can appear PPT under the standard FPT, indicating that only twisted variants may act as full entanglement monotones [2403.12121]. The FPT generalizes within the framework of anyonic partial transpose, further extending its reach to non-Abelian anyons.

Open directions include analytic continuation of moments for complex spectra, generalizations to higher dimensions, and rigorous treatment of interacting systems and disjoint intervals. For critical systems and SPT phases, the FPT underpins a unified covariance-matrix/DFA approach to topological invariants, scaling laws, and bulk-boundary correspondence. The method sets a firm foundation for both operational entanglement diagnostics and algebraic/topological classification in fermionic many-body theory [2206.03260][1710.01886][2403.12121][1906.04211][2503.07742].

Source: https://www.emergentmind.com/topics/fermionic-partial-transpose-fpt