When a one-form anomaly orders thermal transitions

Key claim

An anomaly can exclude a trivially gapped confined phase without deciding how deconfinement occurs.

It may order two transition temperatures while leaving their separation, transition orders, and critical exponents undetermined.

Theme

The CP–center anomaly as an ordering constraint on thermal transitions.

Guiding question

How can a four-dimensional anomaly involving a one-form symmetry constrain the ordinary ${\mathbb{Z}}_2$ symmetry of a Polyakov loop?

Logical scope

The anomaly is an exact obstruction. A thermal phase diagram follows only after specifying which infrared mechanisms are available to match it.

For pure $\mathrm{SU}(2)$ Yang–Mills at $\theta =\pi$, CP and the electric ${\mathbb{Z}}_2^{(1)}$ center symmetry have a mixed ‘t Hooft anomaly.

Thermal compactification splits the four-dimensional center background into two pieces. One still probes spatial Wilson loops. The other becomes the background for the zero-form center symmetry acting on the Polyakov loop.

The anomaly survives this split. Its consequences for phases, however, require infrared assumptions.

Setup

Put pure $\mathrm{SU}(2)$ Yang–Mills on

with Euclidean action

The trace convention is chosen so that $Q\in \mathbb{Z}$ for an ordinary $\mathrm{SU}(2)$ bundle.

At $\theta =\pi$, CP sends

which returns to $\pi$ after a $2\pi$ shift. Without a center background, this makes CP a symmetry.

Now couple the electric one-form center symmetry to

In this background, the instanton number can become fractional. The $2\pi$ shift needed by CP is no longer invisible.

In a standard choice of local counterterm, the anomalous transformation is represented by

where

is the Pontryagin square.

Counterterms

A local counterterm can change the displayed phase, but not the anomaly class. The invariant content is the obstruction to preserving all relevant symmetries in a trivial infrared phase.

Changing the counterterm convention may change the representative phase. It cannot remove the mixed anomaly while preserving the required symmetries.

This transformation law is the exact ultraviolet statement. It contains no claim yet about a mass gap, spontaneous symmetry breaking, or a transition temperature.

Analysis

Derivation: reduce the anomaly

Let

For backgrounds pulled back from $M_3$, decompose

with

The two fields have different three-dimensional jobs:

Background	Three-dimensional symmetry	Charged probe
$B_s$	Spatial ${\mathbb{Z}}_2^{(1)}$ center	Spatial fundamental Wilson loop
$A$	Thermal ${\mathbb{Z}}_2^{(0)}$ center	Fundamental Polyakov loop

Write the Polyakov loop using $\mathcal{A}$ for the dynamical Yang–Mills connection:

The thermal center acts as

The Pontryagin square is a quadratic refinement:

Therefore the part of $\mathcal{P}(B)$ that depends on both $A$ and $B_s$ is

Integrating over the thermal circle gives

The four-dimensional anomaly thus contains the three-dimensional response

A one-form symmetry has not turned wholesale into a zero-form symmetry. Its background has split into a spatial two-form field and a one-form field obtained by placing one leg on the thermal circle.

Scope and assumptions: isolate the infrared assumptions

Anomaly matching forbids a unique, symmetry-preserving, trivially gapped infrared phase.

Assumption hinge

The temperature inequality depends here. Topological order, additional gapless modes, or a different realization of spatial center would reopen the apparently excluded window.

To translate that statement into an ordering of thermal transitions, assume that the relevant three-dimensional thermal regime has:

1. locality and a mass gap;
2. unbroken spatial ${\mathbb{Z}}_2^{(1)}$ center symmetry;
3. no intrinsic topological order or other nontrivial long-range sector.

In a center-symmetric confined phase,

so the thermal ${\mathbb{Z}}_2^{(0)}$ center is unbroken.

Suppose CP restored at a lower temperature than deconfinement:

Then the interval

would preserve CP, thermal center, and, by assumption, spatial center.

If that interval were also gapped and topologically trivial, all anomalous symmetries would be preserved in a forbidden infrared phase. Hence

under the stated assumptions.

The loopholes are physical, not semantic. A gapless phase, intrinsic topological order, or broken spatial center can match the anomaly without the conventional ordering argument.

Physical interpretation: what the inequality does and does not know

The reduced anomaly connects generalized symmetry to a Landau order parameter:

This explains why the Polyakov loop enters an argument that began with a one-form symmetry.

It does not turn the Yang–Mills theory into a unique three-dimensional Ising model. A potential $V(P)$ can describe thermal-center breaking, but $P$ alone cannot reproduce the response

The spatial one-form sector must remain visible in any effective description that claims to match the full anomaly.

The inequality also allows a deconfined but CP-broken window:

In that phase, thermal center is broken, so the anomaly is already matched even though CP remains broken.

Recent lattice work using analytic continuation from imaginary $\theta$ reports evidence for precisely this strict ordering in pure $\mathrm{SU}(2)$ Yang–Mills. That is dynamical evidence, not part of the anomaly theorem.

False claim to diagnose

Because the anomaly is invariant under renormalization-group flow, deconfinement and CP restoration must occur at the same temperature and share a universality class.

Anomaly matching constrains symmetry realization. It does not select a unique path through the space of phases.

Equality,

is allowed but not required. A strict inequality is also allowed, as is a first-order transition. If a continuous transition occurs, its critical exponents require further dynamical information.

A defensible replacement is:

Given the infrared assumptions above, CP cannot be restored inside a center-symmetric confined phase.

What follows — and what does not

Keep the logical layers separate:

Claim	Status
CP at $\theta =\pi$ has a mixed anomaly with ${\mathbb{Z}}_2^{(1)}$ center.	Exact symmetry statement
Thermal compactification yields a response $(-1{)}^{\int A⌣B_s}$.	Exact background-field reduction
A unique, symmetric, trivially gapped phase is impossible.	Anomaly-matching consequence
$T_{\mathrm{deconf}}\le T_{\mathrm{CP}}$.	Consequence with explicit infrared assumptions
The two transitions coincide or share critical exponents.	Not fixed by the anomaly
Pure $\mathrm{SU}(2)$ has a CP-broken deconfined window.	Dynamical question with recent numerical evidence

The anomaly removes regions from the phase diagram. It does not draw the remaining phase boundaries for us.

Exercise

On $S_{\beta }^1\times M_3$, take

and use

1. Extract the part of

   $$\frac{1}{2}{\int }_{S_{\beta }^1\times M_3}\mathcal{P}(B)$$Copy LaTeX

   that depends on both $A$ and $B_s$.

2. Show that CP produces

   $$(-1{)}^{{\int }_{M_3}A⌣B_s}.$$Copy LaTeX

3. Assume a gap, unbroken spatial center, and no intrinsic topological order. Prove by contradiction that

   $$T_{\mathrm{CP}}<T_{\mathrm{deconf}}$$Copy LaTeX

   is impossible.

Hint 1

Only a term containing one factor of $u$ can survive integration over the thermal circle and still depend on both three-dimensional backgrounds.

Hint 2

Translate “confined” into unbroken thermal center:

$$⟨P⟩=0.$$Copy LaTeX

Then inspect the hypothetical interval between the two transition temperatures.

Oral check 1. Why does one component of a four-dimensional two-form background become a three-dimensional one-form background?

Oral check 2. Name one infrared mechanism that can match the anomaly while preserving CP and thermal center.

Solution

The quadratic property gives the mixed term

$$2B_s⌣A⌣u.$$Copy LaTeX

Thus

$$\frac{1}{2}{\int }_{S_{\beta }^1\times M_3}2B_s⌣A⌣u={\int }_{M_3}A⌣B_s(\mathrm{mod}2).$$Copy LaTeX

Substitution into the CP phase yields

$$\exp [\mathrm{i}\pi {\int }_{M_3}A⌣B_s]=(-1{)}^{{\int }_{M_3}A⌣B_s}.$$Copy LaTeX

Now suppose

$$T_{\mathrm{CP}}<T_{\mathrm{deconf}}.$$Copy LaTeX

Immediately above $T_{\mathrm{CP}}$ but below $T_{\mathrm{deconf}}$, CP is restored while thermal center remains unbroken.

If spatial center also remains unbroken, the phase preserves all three symmetries appearing in the reduced anomaly.

A local, gapped, topologically trivial phase cannot do so. Under these assumptions, the proposed interval is impossible, and therefore

$$T_{\mathrm{deconf}}\le T_{\mathrm{CP}}.$$Copy LaTeX

No contradiction arises when the inequality is strict in the opposite direction. Between $T_{\mathrm{deconf}}$ and $T_{\mathrm{CP}}$, thermal center is broken and can match the anomaly.

Check your understanding

Why does anomaly matching support

under stated assumptions, but not

Answer by naming one forbidden phase and one allowed intermediate phase.

You may also reply with “deeper,” “too easy,” “too hard,” or your derivation.

Further Reading

• Theta, Time Reversal, and Temperature
• Deconfinement and CP-breaking at theta=pi in Yang–Mills theories and a novel phase for SU(2)
• Numerical evidence for a CP-broken deconfined phase in four-dimensional SU(2) Yang–Mills

Connections and next step

• Conceptual primitive: ‘t Hooft anomaly matching under thermal compactification.
• Fields connected: generalized symmetries, confinement thermodynamics, and topological response.
• Toy system: four-dimensional pure $\mathrm{SU}(2)$ Yang–Mills at $\theta =\pi$.
• Decisive structure: the Pontryagin-square cross term reducing to $A⌣B_s$.
• Assumption challenged: an anomaly fixes a unique phase transition.
• Exercise type: cohomological reduction followed by a phase-ordering contradiction.
• Next step: anomaly matching on CP domain walls at $\theta =\pi$.
• Avoid repeating soon: isolated instanton saddles and static-versus-dynamic critical exponents.