Parastatistics and Anyons

Bosons pile into a single level without limit; fermions admit at most one. Are those the only choices? The symmetrization postulate says yes in the three-dimensional world we inhabit — identical-particle states must be either totally symmetric or totally antisymmetric — but that postulate is an input, not a theorem. Relax it and two genuinely different generalizations appear, each with its own statistical mechanics: parastatistics, where up to $p$ particles may share a mode, and anyons, two-dimensional excitations whose exchange multiplies the wavefunction by a continuous phase $e^{i\theta}$ anywhere between the bosonic $+1$ and the fermionic $-1$ . We take the microscopic picture from the quantum-mechanics track. In aq-qm-anyons-fractional-statistics the reason a continuous phase becomes possible is topological: in two dimensions the permutation group $S_N$ , "whose only one-dimensional pictures are the two we know, gives way to the braid group $B_N$ , which admits a continuous circle of exchange phases $e^{i\theta}$ ." Particles carrying such a phase "are anyons, and the angle $\theta$ is their statistics." That article owns the braiding origin — why two dimensions unlocks the phase, and why three dimensions forbids it. This article owns the thermodynamic consequence: the grand-canonical occupation numbers that interpolate between Bose-Einstein and Fermi-Dirac, and the finite occupancy cap $p$ that a para-Fermi gas of order $p$ enforces. The unifying statement is that every one of these statistics is a rule for how many particles a single mode accepts, encoded in the single-level grand partition function $\Xi_k=\sum_n (ze^{-\beta\varepsilon})^n$ . Fermions stop the sum at $n=1$ ; bosons run it to infinity; parafermions of order $p$ stop it at $n=p$ ; Haldane's exclusion parameter $g$ tunes the depletion rate continuously. One knob, three named limits, and a continuum in between — that is intermediate statistics.

Key takeaways

The choice "boson or fermion" reflects the symmetrization postulate, a physical input; relaxing it admits parastatistics (integer order $p$ ) and, in two dimensions, anyons (continuous phase $\theta$ ).
Para-Fermi statistics of order $p$ caps single-mode occupancy at $n\in\{0,1,\dots,p\}$ ; the single-level grand sum is a finite geometric series $\Xi_k=(1-w^{p+1})/(1-w)$ with $w=ze^{-\beta\varepsilon}$ .
The order- $p$ mean occupation interpolates: $p=1$ reproduces Fermi-Dirac $\bar n=1/(e^{\beta(\varepsilon-\mu)}+1)$ , and $p\to\infty$ reproduces Bose-Einstein $\bar n=1/(e^{\beta(\varepsilon-\mu)}-1)$ .
The microscopic braid-group origin of the anyon phase $e^{i\theta}$ is owned by aq-qm-anyons-fractional-statistics; this article develops the thermodynamics that phase implies.
Haldane's exclusion statistics with parameter $g$ gives an occupation that reduces to Bose-Einstein at $g=0$ and Fermi-Dirac at $g=1$ , with fractional $g$ between.
In $d\ge3$ a double exchange is contractible, so the exchange phase squares to one and only $\pm1$ survive — the reason anyons are strictly two-dimensional.
Key idea: every statistics is a rule for how many particles a mode accepts. Fermi stops the single-level sum at $n=1$ , Bose runs it to $\infty$ , para-Fermi of order $p$ stops it at $n=p$ , and $g$ tunes the cap continuously — the occupation number interpolates smoothly between Fermi-Dirac and Bose-Einstein.

The core insight

Intermediate statistics is a continuous dial on the single-mode occupancy cap. The grand-canonical single-level partition function $\Xi_k=\sum_{n}(ze^{-\beta\varepsilon})^n$ carries the entire content: truncating at $n=1$ gives Fermi-Dirac, running to $n=\infty$ gives Bose-Einstein, truncating at $n=p$ gives para-Fermi order $p$ , and Haldane's parameter $g$ interpolates the two endpoints for fractional statistics. The braid-group reason a continuous exchange phase is even allowed is quantum-mechanics-owned; the occupation numbers it produces are statistical mechanics.

Where this lives today

Anyons stopped being a theorist's curiosity when the fractional quantum Hall effect gave them a laboratory. Quasiparticles of the $\nu=1/3$ Laughlin state carry charge $e/3$ — confirmed by shot-noise measurements — and 2020 interferometry and anyon-collider experiments measured their fractional exchange statistics directly, closing a four-decade loop from Wilczek's coinage to a braiding phase read off a current. Non-Abelian anyons, whose braids implement unitary gates on a degenerate ground-state manifold, are the physical substrate of topological quantum computation: information stored in the fusion channels is protected by the same topology that forbids anyons in three dimensions. Haldane's exclusion statistics, meanwhile, has become a working tool far beyond two dimensions — it describes the low-temperature thermodynamics of spinon excitations in one-dimensional antiferromagnets and quasiparticles in integrable models, wherever the state-counting deviates from the clean Bose or Fermi rule.

The physical picture

Take two identical particles, drag one slowly around the other, and return them to their starting configuration. In three dimensions any such loop can be lifted "over the top" and shrunk to a point, so exchanging twice is the same as doing nothing; the single-exchange phase must square to $1$ , leaving only $+1$ (bosons) or $-1$ (fermions). In two dimensions the particles are confined to a plane and cannot pass over one another — the loop genuinely encircles a forbidden point — so a double exchange is topologically nontrivial and the phase is unconstrained. That is the whole reason a continuous $e^{i\theta}$ is possible in the plane and nowhere else. The first figure is the hero: a dial from $\theta=0$ (boson) through the anyonic middle to $\theta=\pi$ (fermion). Ordinary matter lives at the two poles; the fractional quantum Hall quasiparticles sit at intermediate angles, "the real, measured excitations of two-dimensional electron gases in the fractional quantum Hall regime."

Figure

The statistics dial from boson through anyon to fermion

The punchline: the endpoints are the familiar $\pm1$ , but between them lies a continuum of legitimate exchange phases — accessible only because a two-dimensional braid cannot be undone.

Parastatistics answers a different question in any dimension: what if the many-body wavefunction is neither fully symmetric nor fully antisymmetric, but carries a hidden internal label? Antisymmetrizing over the combined space-and-label coordinates lets the spatial part be partially symmetric, so up to $p$ particles can share a spatial mode before the labels run out. The order- $p$ para-Fermi gas therefore looks like a fermion gas whose Pauli cap has been raised from $1$ to $p$ . The second figure shows that occupancy ladder tightening from Bose ( $p=\infty$ , no cap) down through para-Fermi to Fermi ( $p=1$ ).

Figure

Occupancy caps: Bose runs to infinity, para-Fermi stops at p, Fermi stops at one

The single-level grand sum is a finite geometric series when the cap is finite; the cap is the only thing that changes across the whole family.

The core equations

The engine for all of these statistics is the single-level grand partition function, the sum over how many particles occupy one mode. Its range of summation is the statistics.

\Xi_k=\sum_{n=0}^{n_{\max}}\big(z\,e^{-\beta\varepsilon_k}\big)^{n},\qquad z=e^{\beta\mu},\ \ w\equiv z\,e^{-\beta\varepsilon_k} \qquad (1)

where: $\Xi_k$ — grand partition function of one single-particle level; $z$ — fugacity; $w=ze^{-\beta\varepsilon_k}$ the per-mode Boltzmann-fugacity weight; $n_{\max}$ the occupancy cap. Physical reading: $n_{\max}=1$ is Fermi, $n_{\max}=\infty$ is Bose, $n_{\max}=p$ is para-Fermi of order $p$ — the statistics is entirely the upper limit of this sum. For a finite cap $p$ the sum is a truncated geometric series, which closes in elementary form. This is the para-Fermi single-level partition function.

\Xi_k^{(p)}=\sum_{n=0}^{p}w^{n}=\frac{1-w^{\,p+1}}{1-w} \qquad (2)

where: $p$ — the para-Fermi order (maximum occupancy of a mode). Physical reading: setting $p=1$ gives $1+w$ (Fermi) and letting $p\to\infty$ gives $1/(1-w)$ (Bose); the finite- $p$ form is the smooth bridge between them, valid for any $w\ge0$ because the numerator makes it a finite sum with no convergence condition. The mean occupation follows from the standard grand-canonical relation, the logarithmic fugacity derivative of $\Xi_k$ .

\bar n_k=z\left(\frac{\partial\ln\Xi_k}{\partial z}\right)_{\!T}=\frac{w}{1-w}-\frac{(p+1)\,w^{\,p+1}}{1-w^{\,p+1}} \qquad (3)

where: $\bar n_k$ — mean occupation of the level, bounded $0\le\bar n_k\le p$ . Physical reading: the first term is the bosonic occupation, and the second is the correction that enforces the cap; as $p\to\infty$ the second term vanishes and Eq. (3) collapses to the Bose form, while $p=1$ collapses it to Fermi-Dirac. For the two-dimensional anyon gas the natural language is Haldane's exclusion statistics: a parameter $g$ measures how fast the available single-particle states are depleted as particles are added. The mean occupation is given implicitly through an auxiliary variable $w(\zeta)$ .

\bar n=\frac{1}{w(\zeta)+g},\qquad w^{g}\,(1+w)^{1-g}=\frac{1}{\zeta},\qquad \zeta=e^{-\beta(\varepsilon-\mu)} \qquad (4)

where: $g$ — the Haldane exclusion parameter ( $g=0$ bosonic, $g=1$ fermionic); $\zeta$ — the single-level fugacity factor. Physical reading: at $g=0$ the functional equation gives $1+w=1/\zeta$ , so $\bar n=1/(1/\zeta-1)=1/(e^{\beta(\varepsilon-\mu)}-1)$ , the Bose distribution; at $g=1$ it gives $w=1/\zeta$ , so $\bar n=1/(1/\zeta+1)=1/(e^{\beta(\varepsilon-\mu)}+1)$ , the Fermi distribution — fractional $g$ interpolates. The exchange phase itself, borrowed from the quantum-mechanics owner, is the topological label that makes all this possible in the plane.

\psi(\dots,\mathbf r_i,\dots,\mathbf r_j,\dots)=e^{i\theta}\,\psi(\dots,\mathbf r_j,\dots,\mathbf r_i,\dots),\qquad \theta\in[0,\pi] \qquad (5)

where: $\theta$ — the statistical angle owned by aq-qm-anyons-fractional-statistics. Physical reading: $\theta=0$ is symmetric (boson), $\theta=\pi$ is antisymmetric (fermion), and a double exchange multiplies by $e^{2i\theta}$ — nontrivial in the plane, but forced to unity in three dimensions where it must equal the identity.

Derivations

Derivation AThe para-Fermi occupation interpolates Fermi-Dirac and Bose-Einstein

The goal is to derive the mean occupation of a single mode for a para-Fermi gas of order $p$ and show it is a genuine bridge between the two familiar statistics: it must reduce exactly to the Fermi-Dirac function $1/(e^{\beta(\varepsilon-\mu)}+1)$ when $p=1$ and to the Bose-Einstein function $1/(e^{\beta(\varepsilon-\mu)}-1)$ as $p\to\infty$ . The strategy is to close the finite geometric series for $\Xi_k$ , take the logarithmic fugacity derivative, and check both endpoints. We take the microscopic occupancy cap $n\le p$ as given (the group-theoretic content is quantum-mechanics-owned); this card owns the resulting thermodynamic distribution.

Step 1. Write the single-level grand sum with cap $p$ . With $w=ze^{-\beta\varepsilon}$ , $\Xi_k=\sum_{n=0}^{p}w^n=(1-w^{p+1})/(1-w)$ , a finite geometric series. Why: the para-Fermi rule allows $n=0,1,\dots,p$ particles in the mode and no more, so the sum truncates after the $n=p$ term — exactly where the fermion sum truncates after $n=1$ and the boson sum never truncates.

Step 2. Take the occupation as the logarithmic derivative. $\bar n=z\,\partial_z\ln\Xi_k=w\,\partial_w\ln\Xi_k$ since $w\propto z$ . Differentiating $\ln[(1-w^{p+1})/(1-w)]$ gives $\bar n=\dfrac{w}{1-w}-\dfrac{(p+1)w^{p+1}}{1-w^{p+1}}$ . Why: the first term is the ordinary bosonic result from $\ln(1-w)^{-1}$ ; the second, from $\ln(1-w^{p+1})$ , is the penalty that enforces the finite cap.

Step 3. Recover Fermi-Dirac at $p=1$ . Set $p=1$ : $\bar n=\dfrac{w}{1-w}-\dfrac{2w^{2}}{1-w^{2}}=\dfrac{w}{1-w}-\dfrac{2w^{2}}{(1-w)(1+w)}=\dfrac{w(1+w)-2w^2}{(1-w)(1+w)}=\dfrac{w(1-w)}{(1-w)(1+w)}=\dfrac{w}{1+w}$ . With $w=e^{-\beta(\varepsilon-\mu)}$ this is $1/(e^{\beta(\varepsilon-\mu)}+1)$ . Why: the cap-penalty exactly cancels the bosonic pole, leaving the bounded Fermi function — the $p=1$ para-Fermi gas is an ordinary fermion gas.

Step 4. Recover Bose-Einstein as $p\to\infty$ . For $0\le w<1$ , $w^{p+1}\to0$ , so the second term vanishes and $\bar n\to w/(1-w)=1/(e^{\beta(\varepsilon-\mu)}-1)$ . Why: an unbounded cap is precisely the boson rule, and the geometric series recovers the Bose pole as $\varepsilon\to\mu^{+}$ ; the finite- $p$ gas smoothly removes that pole by capping occupancy at $p$ . Verified: the finite geometric sum, the derivative, and both limits ( $p=1\to$ FD, $p\to\infty\to$ BE) are checked symbolically for $p=1,2,3,5,10$ in verify_parastatistics-anyons.py.

Go deeper · Derivation B · 1 minHaldane exclusion statistics reproduces Bose and Fermi at g=0 and g=1

Derivation BHaldane exclusion statistics reproduces Bose and Fermi at g=0 and g=1

The goal is to show that Haldane's continuous exclusion parameter $g$ furnishes a single occupation formula whose two integer endpoints are exactly the Bose-Einstein ( $g=0$ ) and Fermi-Dirac ( $g=1$ ) distributions, so that fractional $g$ is a legitimate thermodynamic interpolation rather than an ad hoc guess. The route is to solve the defining functional equation for the auxiliary variable $w(\zeta)$ at the two endpoints and substitute into $\bar n=1/(w+g)$ . This is the thermodynamic face of the anyon phase; the braiding origin of a fractional $g$ is owned by the quantum-mechanics article and is not re-derived here.

Step 1. State the two ingredients. The mean occupation is $\bar n=1/[w(\zeta)+g]$ where $w$ solves $w^{g}(1+w)^{1-g}=1/\zeta$ , with $\zeta=e^{-\beta(\varepsilon-\mu)}$ . Why: $g$ counts how many single-particle states each added particle removes from the available pool; $g=1$ is one-for-one exclusion (Pauli), $g=0$ is no exclusion (Bose).

Step 2. Solve the endpoint $g=0$ . The functional equation becomes $(1+w)^{1}=1/\zeta$ , i.e. $w=1/\zeta-1$ . Then $\bar n=1/(w+0)=1/(1/\zeta-1)=\zeta/(1-\zeta)$ . Why: substituting $\zeta=e^{-\beta(\varepsilon-\mu)}$ gives $\bar n=1/(e^{\beta(\varepsilon-\mu)}-1)$ — the Bose-Einstein distribution, complete with its pole at $\varepsilon=\mu$ .

Step 3. Solve the endpoint $g=1$ . Now $w^{1}(1+w)^{0}=1/\zeta$ , i.e. $w=1/\zeta$ . Then $\bar n=1/(w+1)=1/(1/\zeta+1)=\zeta/(1+\zeta)=1/(e^{\beta(\varepsilon-\mu)}+1)$ . Why: the $+g=+1$ in the denominator plays exactly the role of the $+1$ that makes the Fermi function bounded by one — exclusion statistics at $g=1$ is Fermi-Dirac.

Step 4. Read off the interpolation. For $0<g<1$ the occupation is bounded by $1/g$ (the pole of $1/(w+g)$ sits at $w=-g<0$ , unreachable for $w\ge0$ ), so fractional-statistics gases carry more than one particle per state but less than a boson's unbounded pile. Why: $g$ is a continuous cap-softness knob; the maximum occupation $1/g$ diverges as $g\to0$ (Bose) and equals one at $g=1$ (Fermi), matching the two named limits. Verified: the $g=0$ and $g=1$ solutions of $w^g(1+w)^{1-g}=1/\zeta$ and the resulting Bose/Fermi occupations are checked symbolically in verify_parastatistics-anyons.py.

Limiting cases

The intermediate-statistics family collapses to familiar physics in five diagnostic limits.

Regime	Result	Meaning
$p=1$ (para-Fermi)	$\bar n=1/(e^{\beta(\varepsilon-\mu)}+1)$	Ordinary fermions — Pauli cap of one
$p\to\infty$ (para-Fermi)	$\bar n=1/(e^{\beta(\varepsilon-\mu)}-1)$	Ordinary bosons — no cap, Bose pole returns
$g=0$ (Haldane)	Bose-Einstein	No state depletion on adding a particle
$g=1$ (Haldane)	Fermi-Dirac	One-for-one state depletion (exclusion)
$d\ge3$	$e^{i\theta}=\pm1$ only	Double exchange contractible: only bosons/fermions

Common trapParastatistics of order p is not the same thing as an anyon with a fractional phase

The two generalizations answer different questions and live in different settings. Parastatistics keeps the permutation group $S_N$ in any dimension but enlarges the allowed occupancy to an integer $p$ ; it is characterized by a discrete order and, in $d\ge3$ , is physically equivalent to ordinary particles carrying a hidden internal symmetry. Anyons exist only in $d=2$ , replace $S_N$ by the braid group $B_N$ , and carry a genuinely continuous statistical angle $\theta$ (or, for non-Abelian anyons, a braiding matrix). Confusing "order $p$ " with "phase $\theta$ " conflates an integer cap with a continuous angle, and conflates a three-dimensional group-theory refinement with a two-dimensional topological one. Haldane's $g$ is a third, thermodynamic, parameter — continuous like $\theta$ but defined by state-counting rather than by braiding.

Worked examples

Example 1 — Enumerate a para-Fermi mode of order two.

Step 1. Setup. A single mode of a para-Fermi gas of order $p=2$ accepts $n\in\{0,1,2\}$ . With $w=e^{-\beta(\varepsilon-\mu)}$ , the grand-canonical weights are $1,\ w,\ w^2$ , so $\Xi_k=1+w+w^2$ .

Step 2. Occupation at $\varepsilon=\mu$ . Here $w=1$ , weights $1,1,1$ , and $\bar n=(0\cdot1+1\cdot1+2\cdot1)/(1+1+1)=3/3=\mathbf{1}$ . The mode holds one particle on average — exactly the middle of its range $[0,2]$ , as symmetry demands.

Step 3. Deep below the chemical potential. Take $\varepsilon-\mu=-5k_BT$ , so $w=e^{5}=148.4$ . Then $\bar n=(0+148.4+2\cdot148.4^2)/(1+148.4+148.4^2)=(148.4+44{,}046)/(1+148.4+22{,}023)=44{,}195/22{,}172=\mathbf{1.993}$ . The mode is nearly filled to its cap of $2$ .

Step 4. Far above the chemical potential. Take $\varepsilon-\mu=+5k_BT$ , $w=e^{-5}=6.74\times10^{-3}$ . Then $\bar n\approx w=\mathbf{6.7\times10^{-3}}$ , essentially empty — the classical Boltzmann tail, identical to what a fermion or boson mode gives when $w\ll1$ .

Example 2 — Fractional charge and braid angle of a Laughlin quasiparticle.

Step 1. Setup. The $\nu=1/3$ fractional quantum Hall state hosts quasiparticles that, per aq-qm-anyons-fractional-statistics, "carry a fraction $e/m$ of the electron's charge and braid with angle $\pi/m$ ." Here $m=3$ .

Step 2. Fractional charge. The quasiparticle charge is $q=e/m=e/3=\mathbf{0.333\,e}$ — one-third of the electron charge, a directly measured quantity in shot-noise experiments.

Step 3. Statistical angle. The braid angle is $\theta=\pi/m=\pi/3=\mathbf{60^\circ}$ , strictly between the bosonic $0$ and the fermionic $\pi$ — a genuine anyon.

Step 4. Interpret via Haldane. Mapping to exclusion statistics, the $\nu=1/m$ Laughlin quasiholes carry $g=1/m=1/3$ : adding a quasiparticle removes a fraction $1/3$ of a state from the pool, intermediate between the bosonic $g=0$ and fermionic $g=1$ . The three descriptions — phase $\pi/3$ , charge $e/3$ , exclusion $g=1/3$ — are three faces of the same fractional statistics.

Origins · 1 minfrank wilczek and how this was discovered

frank wilczek

born 1951, Nobel Prize 2004 · Frank Wilczek named "anyons" in 1982, arguing that particles in two dimensions can carry any exchange phase, building on the 1977 analysis of Leinaas and Myrheim. The idea grew from his work on flux-charge composites and matured into a cornerstone of the fractional quantum Hall effect and topological quantum computation. Wilczek's Nobel Prize was for asymptotic freedom in the strong interaction, but "anyon" is the coinage that made intermediate statistics a laboratory reality.

History

1953

— H. S. Green introduces parastatistics through trilinear commutation relations, generalizing bosons and fermions

1971

— Doplicher, Haag, and Roberts show parastatistics in d≥3 is ordinary statistics with hidden internal symmetry

1977

— Leinaas and Myrheim analyze identical-particle statistics in two dimensions, finding a continuous exchange phase

1982

— Wilczek coins "anyon" for two-dimensional particles with fractional statistics

1984

— Halperin identifies fractional statistics in the fractional quantum Hall quasiparticles

1991

— Haldane formulates exclusion statistics with a continuous parameter g, giving a thermodynamic interpolation

Going further

Texts: Pathria & Beale, Statistical Mechanics (3rd ed.), Ch. 6 (quantum statistics) and supplementary discussion of intermediate statistics; Khare, Fractional Statistics and Quantum Theory; Wilczek, Fractional Statistics and Anyon Superconductivity; Stern, "Anyons and the quantum Hall effect," Ann. Phys. 323 (2008).
Microscopic owner: the braid-group origin of the continuous exchange phase $e^{i\theta}$ , and why two dimensions is special, are owned by aq-qm-anyons-fractional-statistics. This article develops the statistical mechanics — the interpolating occupation numbers and the finite occupancy cap.
Builds on: Indistinguishability of Identical Particles · Symmetrization Postulate · Bosons, Fermions, and the Spin-Statistics Connection · Pauli Exclusion Principle · Occupation Number Representation.
Leads to: Fermi-Dirac Distribution · Bose-Einstein Distribution · Grand Canonical Ensemble for Ideal Quantum Gases; and outward to the Fractional Quantum Hall Effect and Topological Quantum Computation.
Provenance: AllQFT `sm` export (Pathria & Beale, Ch. 6, intermediate statistics), aq-sm-parastatistics-anyons, ref STAT.QSTAT.PARAS.1216. All para-Fermi sums, the $p=1\to$ FD and $p\to\infty\to$ BE limits, the Haldane $g=0,1$ endpoints, the exchange-phase reductions, and the Laughlin $e/3$ , $\pi/3$ numbers are python-verified (26 checks, all PASS).

If bosons and fermions are the only options in our world, why study anyons at all?

Because "our world" here means three space dimensions, and a great deal of interesting physics happens on effectively two-dimensional surfaces — the interface of a semiconductor heterostructure, a graphene sheet, the surface of a topological material. On such a plane the topological obstruction that forces $e^{i\theta}=\pm1$ disappears, and quasiparticle excitations really do braid with fractional phase. The fractional quantum Hall effect is the flagship case: its quasiparticles carry charge $e/3$ and anyonic statistics, both now measured. Anyons are not hypothetical particles awaiting an accelerator; they are emergent excitations of ordinary electrons confined to a plane.

What is the difference between the order p and the Haldane parameter g?

Both interpolate Fermi and Bose, but they are different constructions. The order $p$ is an integer occupancy cap: a mode accepts $0,1,\dots,p$ particles, so $p=1$ is Fermi, $p=\infty$ is Bose, and intermediate integers are para-Fermi statistics defined through Green's trilinear commutation relations. Haldane's $g$ is a continuous real parameter defined thermodynamically, by the rate at which adding a particle depletes the pool of available states; $g=0$ is Bose, $g=1$ is Fermi, and fractional $g$ (like $g=1/3$ for $\nu=1/3$ quasiparticles) describes genuine anyons. In short: $p$ is a discrete cap from group theory, $g$ is a continuous cap-softness from state-counting.

Does this article re-derive the anyon exchange phase?

No. The reason a continuous phase $e^{i\theta}$ is even allowed — that the two-dimensional braid group $B_N$ replaces the permutation group $S_N$ and "admits a continuous circle of exchange phases $e^{i\theta}$ " — is owned by aq-qm-anyons-fractional-statistics. This article takes that phase as input and asks the statistical-mechanics question: given a family of exchange statistics, what are the equilibrium occupation numbers? The answer is the interpolating $\bar n$ of Eqs. (3) and (4). We translate the topological label into thermodynamics; we do not re-derive the topology.

Why does the para-Fermi occupation have that awkward second term?

The finite cap is what makes it awkward, and also what makes it correct. A boson mode has occupation $w/(1-w)$ , which diverges as $w\to1$ (the Bose pole). A para-Fermi mode must never exceed $p$ particles, so something has to tame that divergence. The second term, $-(p+1)w^{p+1}/(1-w^{p+1})$ , is exactly the subtraction that caps the occupation at $p$ : as $w\to1$ it grows to cancel the bosonic blow-up, leaving $\bar n\to p/2$ at $\varepsilon=\mu$ and $\bar n\to p$ deep below. For $p=1$ it cancels the pole entirely and leaves the bounded Fermi function; for $p\to\infty$ it vanishes and the Bose form returns.

Is a para-Fermi gas of order 2 the same as a spin-1/2 fermion gas, which also holds two per level?

No, though the confusion is natural. A spin- $\tfrac12$ fermion level really consists of two distinct modes (spin up and spin down), each obeying Pauli exclusion with cap one; the "two per level" is two separate one-particle boxes. A para-Fermi mode of order two is a single mode that intrinsically accepts up to two identical particles with no extra label. Thermodynamically they differ: two independent Fermi modes give occupation $2/(e^{\beta(\varepsilon-\mu)}+1)$ (twice a Fermi function, still capped by two but shaped like FD), whereas the para-Fermi mode gives the single-mode formula of Eq. (3) with $p=2$ , a different function of energy.

Where does parastatistics actually show up if d≥3 makes it trivial?

The Doplicher–Haag–Roberts result says a para-particle of order $p$ in three dimensions is physically equivalent to an ordinary boson or fermion carrying a hidden $SU(p)$ internal symmetry — so parastatistics is not a new option for fundamental particles in our world. Its value is twofold. First, it clarifies that the symmetrization postulate is genuinely a postulate, forced only up to internal-symmetry relabelings. Second, the same finite-cap counting reappears as an effective description: quasiparticles with a bounded occupancy, or systems where an internal color label limits how many excitations share a mode, are handled by the para-Fermi machinery even when the underlying particles are ordinary fermions.