BPS Spectra, Chambers, and Wall Crossing

Seiberg–Witten periods determine the central charge of any allowed electromagnetic vector, but a BPS spectrum is additional chamber-dependent information. For pure $SU(2)$, a strong-coupling chamber contains two hypermultiplets, while a weak-coupling chamber contains a vector multiplet and two infinite dyon towers. A phase-ordered wall-crossing identity relates their protected indices.

Required background. Curves and periods supplies $Z_\gamma(u)$, while marginal stability and wall crossing fixes the index and ordering conventions. Helpful background. Framed BPS states gives the halo interpretation.

Spectrum data at a point

To state a BPS spectrum, specify

$$(u,\Gamma,\langle\ ,\ \rangle,Z,\mathcal H,\Omega),$$

where $u$ is the vacuum, $\Gamma$ the transported charge lattice, $\mathcal H$ a chosen half-plane of central-charge phases distinguishing particles from antiparticles, and $\Omega(\gamma;u)$ the protected second helicity supertrace.

The mass formula

$$M_\gamma=|Z_\gamma|$$

applies if a BPS state exists. It does not decide $\Omega(\gamma;u)$.

A chamber is a connected region in which no relevant central-charge rays align and no BPS state crosses the boundary of $\mathcal H$. Within it, protected indices are locally constant after charges are parallel transported.

Walls of marginal stability

For a possible decay $\gamma=\gamma_1+\gamma_2$, the wall is

$$\operatorname{Im}\!\left( Z_{\gamma_1}\overline{Z_{\gamma_2}} \right)=0, \qquad \operatorname{Re}\!\left( Z_{\gamma_1}\overline{Z_{\gamma_2}} \right)>0.$$

At the wall,

$$|Z_\gamma|=|Z_{\gamma_1}|+|Z_{\gamma_2}|,$$

so a bound state can reach infinite separation. Which side is stable depends on the Dirac pairing and the oriented phase order.

For pure $SU(2)$ with constituent charges associated with the two finite singularities, the curve of marginal stability can be written as the locus where the relevant period ratio is real, for example

$$\operatorname{Im}\frac{a_D}{a}=0$$

away from zeros of $a$, with a further sign condition selecting alignment. Its precise drawing depends on branch cuts; it passes through the singular points and separates strong- and weak-coupling chambers.

A pure-SU(2) charge convention

Use period charges $\gamma=(p,q)$ with

$$Z_\gamma=pa_D+qa.$$

For the particle lattice convention used in the monodromy formulas, normalize the physical pairing as

$$\langle(p,q),(p',q')\rangle =2(pq'-qp').$$

The factor two is the physical root–coroot polarization of the adjoint particle lattice and is the same factor appearing in the local monodromies. The displayed labels are root-unit labels: $W=(0,1)$. In the common GMN coordinates,

$$(n_e^{\mathrm{GMN}},n_m^{\mathrm{GMN}})=(2q,p),$$

so the same $W$ is $(2,0)$ and the two strong states below are $(0,1)$ and $(2,-1)$. This translation reverses the chapter’s determinant orientation; pairing signs and factor order must therefore be translated with the labels.

Choose the two strong-chamber particle charges

$$\gamma_1=(1,0), \qquad \gamma_2=(-1,1).$$

$\gamma_1$ is the monopole. $\gamma_2$ is the antiparticle orientation of the dyon charge $(1,-1)$ used to label the second singularity. They satisfy

$$\langle\gamma_1,\gamma_2\rangle=2.$$

Choose the phase half-plane so both are particles. In the strong chamber,

$$\Omega(\gamma_1)=\Omega(\gamma_2)=+1,$$

and, for the standard pure theory, these are the only stable particle hypermultiplets in that half-plane.

The weak-coupling spectrum

Crossing to weak coupling produces:

• hypermultiplets of charges

  $$\gamma_n^+=(n+1)\gamma_1+n\gamma_2=(1,n), \qquad n\ge0,$$

• hypermultiplets of charges

  $$\gamma_n^-=n\gamma_1+(n+1)\gamma_2=(-1,n+1), \qquad n\ge0,$$

• a vector multiplet of charge

  $$\gamma_W=\gamma_1+\gamma_2=(0,1).$$

In the second-helicity convention,

$$\Omega(\gamma_n^\pm)=+1, \qquad \Omega(\gamma_W)=-2.$$

Antiparticles carry the opposite charges and lie outside the chosen half-plane. This is the same spectrum in which the coordinate determinant of the two seed charges is one but their physical pairing is two; the chapter crosswalk separates curve homology, particles, GMN labels, and genuine lines.

The tower is a statement about the protected one-particle spectrum in the weak chamber. Multi-particle states are not additional primitive $\Omega$ entries. The strong- and weak-coupling spectra and their change across the marginal-stability curve are derived in Bilal and Ferrari 1996, §§3–6.

For a frozen numerical weak-chamber datum, use the chapter base point $u_b=2\Lambda^2$ and half-plane

$$-\frac{\pi}{4}<\arg Z<\frac{3\pi}{4}.$$

There

$$\frac{a}{\Lambda}=1.966685301550330\ldots, \qquad \frac{a_D}{i\Lambda}=0.473434436535537\ldots,$$

so the displayed $\gamma_n^\pm$ and $\gamma_W$ lie in that half-plane and have the stated weak-chamber indices. The structured figure record fixes the numerical tolerance before classifying the rays.

The ordered wall-crossing factorization

Use twisted-torus variables

$$X_\gamma X_{\gamma'} =(-1)^{\langle\gamma,\gamma'\rangle} X_{\gamma+\gamma'}$$

and define the pullback automorphism

$$\mathcal K_\gamma: X_\beta\longmapsto X_\beta(1-X_\gamma)^{\langle\beta,\gamma\rangle}.$$

A state contributes $\mathcal K_\gamma^{\Omega(\gamma)}$. A sector product also needs a boundary and an orientation. We choose them so that the strong-chamber factors are written $\gamma_2$ then $\gamma_1$ from left to right. Their action on coordinate functions is composed rightmost first, so $\mathcal K_{\gamma_1}$ acts first. For the 2-Kronecker pairing above, the strong-chamber product

$$\mathcal K_{\gamma_2}\mathcal K_{\gamma_1}$$

is equal to the weak-chamber phase-ordered product

$$\left( \prod_{n=0}^{\nearrow\infty} \mathcal K_{(n+1)\gamma_1+n\gamma_2} \right) \mathcal K_{\gamma_1+\gamma_2}^{-2} \left( \prod_{n=\infty}^{\searrow0} \mathcal K_{n\gamma_1+(n+1)\gamma_2} \right).$$

The arrows on the limits record the opposite phase order of the two towers. Under the chapter crosswalk, $(n_e^{\mathrm{GMN}},n_m^{\mathrm{GMN}})=(2q,p)$, this is exactly the pure-$SU(2)$ identity with strong charges $(2,-1)$ and $(0,1)$ and weak charges $(2n,1)$ for $n\ge0$, $(2,0)$, and $(2n,-1)$ for $n\ge1$. No additional factor reversal is made: the cited definition already gives the action on the coordinate functions. If point transformations rather than their coordinate actions are used, the entire identity must be reversed consistently. The vector exponent $-2$ is essential; omitting it spoils the symplectomorphism identity. The underlying stability-data transformation is formulated in Kontsevich and Soibelman 2008, §2.3, and the pure-$SU(2)$ identity is given in Gaiotto, Moore, and Neitzke 2010, §2.2, Eq. (2.26).

The identity can be checked on formal variables to any bounded charge order. Such a truncation verifies the declared finite sector, not the entire infinite product unless a formal convergence or algebraic argument is supplied.

The shared rank-one figure and structured chamber record place this corrected pullback order beside the same periods, based loops, particle polarization, and later small-$\mathcal N=1$ deformation. Its finite formal-series fixture is a bounded check, while the cited wall-crossing theorem supplies the all-orders statement.

Determining a chamber numerically

Given numerical periods at $u$:

1. transport the charge basis from a fixed base point;

2. compute $Z_{\gamma_1}$ and $Z_{\gamma_2}$ with error bounds;

3. evaluate

   $$\operatorname{Im}(Z_{\gamma_1}\overline{Z_{\gamma_2}});$$

4. verify that its magnitude exceeds numerical uncertainty away from a wall;

5. order all included rays by a continuously unwrapped phase;

6. apply the appropriate chamber spectrum and test the ordered product.

Using a principal-value phase independently at each point can create artificial order jumps at $\arg Z=\pi$.

Higher-rank methods

In general $\mathcal N=2$ theories, BPS quivers, spectral networks, semiclassical quantization, string webs, and protected indices can determine spectra in selected chambers. Each method has a domain:

• a BPS quiver requires a finite positive basis in a chosen half-plane;
• a spectral network requires a suitable curve/differential presentation;
• semiclassics applies where solitons are large and weakly coupled;
• wall crossing propagates seed data but does not create those seeds.

Completeness must be argued separately. A finite chamber does not imply every chamber is finite.

Protected versus unprotected information

$\Omega$ is stable under deformations away from walls because long multiplets cancel. It can jump when a bound state meets the continuum. It does not determine:

• the full spin-refined spectrum unless a refined index is used;
• long-multiplet degeneracies;
• BPS wavefunctions or sizes;
• scattering amplitudes;
• whether an allowed charge has states with net zero index.

State these omissions whenever exporting a spectrum.

Common pitfalls

Mixing charge normalizations. The factor two can appear in the pairing, in electric charge labels, or in the central-charge convention; it cannot be changed in only one place.

Calling the weak tower chamber-independent. Most of its states decay on the marginal-stability curve.

Using wall crossing without seed states. The identity constrains changes given a spectrum in one chamber; it does not determine that input from periods alone.

Exercises

Using $\gamma_1=(1,0)$ and $\gamma_2=(-1,1)$:

1. verify their pairing is two;
2. compute $\gamma_0^\pm$, $\gamma_1^\pm$, and $\gamma_W$;
3. identify which strong-chamber singular charge is represented by $-\gamma_2$.

Solution

$$\langle\gamma_1,\gamma_2\rangle =2(1\cdot1-0\cdot(-1))=2.$$

The first tower entries are

$$\gamma_0^+=(1,0), \quad \gamma_1^+=(1,1), \quad \gamma_0^-=(-1,1), \quad \gamma_1^-=(-1,2),$$

and $\gamma_W=(0,1)$. Finally, $-\gamma_2=(1,-1)$ is the dyon charge whose central charge vanishes at the second finite singularity in the pure-solution convention.

Check the wall-crossing identity in the quotient that discards every monomial $x^iy^j$ with $i+j>4$, where $x=X_{\gamma_1}$ and $y=X_{\gamma_2}$. Which tower factors can contribute, and what common action do the strong and weak products have on $x$ and $y$?

Solution

The charge of either $\gamma_n^+$ or $\gamma_n^-$ has total degree $2n+1$ in the positive basis $(\gamma_1,\gamma_2)$. Its first correction to $x$ or $y$ therefore has degree at least $2n+2$, so only $n=0,1$ can matter in the degree-four quotient. The weak product truncates to

$$\mathcal K_{\gamma_0^+} \mathcal K_{\gamma_1^+} \mathcal K_{\gamma_W}^{-2} \mathcal K_{\gamma_1^-} \mathcal K_{\gamma_0^-}.$$

For the strong product, rightmost-first composition gives the exact expressions

$$x\longmapsto x(1-y)^2, \qquad y\longmapsto \frac{y}{\left[1-x(1-y)^2\right]^2}.$$

Expanding through total degree four yields

$$x\longmapsto x-2xy+xy^2,$$

and

$$y\longmapsto y+2xy+3x^2y-4xy^2 +4x^3y-12x^2y^2+2xy^3.$$

Applying the five retained weak factors gives the same two polynomials. This proves equality only in the stated finite quotient; the all-orders identity still requires the wall-crossing theorem or a compatible formal-limit argument.

References

• Bilal, Adel, and Frank Ferrari. “The Strong-Coupling Spectrum of the Seiberg–Witten Theory.” Nuclear Physics B 469 (1996): 387–402. arXiv:hep-th/9602082.
• Gaiotto, Davide, Gregory W. Moore, and Andrew Neitzke. “Four-Dimensional Wall-Crossing via Three-Dimensional Field Theory.” Communications in Mathematical Physics 299 (2010): 163–224. arXiv:0807.4723.
• Kontsevich, Maxim, and Yan Soibelman. “Stability Structures, Motivic Donaldson–Thomas Invariants and Cluster Transformations.” 2008. arXiv:0811.2435.