Reading the original ABJM paper

Why does a theory with two gauge nodes and four chiral fields describe a vacuum space with four complex coordinates per brane? Why does that calculation support an M2-brane interpretation without making every rank and level accessible to classical gravity? Keep those questions separate while reading this paper.

The source is Ofer Aharony, Oren Bergman, Daniel Louis Jafferis, and Juan Maldacena, N=6 superconformal Chern-Simons-matter theories, M2-branes and their gravity duals, arXiv:0806.1218v4, 23 October 2008, JHEP 10 (2008) 091. Use this version throughout. Its PDF has an unnumbered title page, so PDF page numbers are one greater than printed page numbers.

Start with the theory’s definition, two-sided gauge coupling, and quiver representations. The later sections below identify their additional preparation. You need not first learn every brane construction: the action can be read and checked on its own.

One version detail matters immediately. In footnote 6, printed p. 11/PDF p. 12, the authors retract an earlier non-Abelian description of their monopoles as endpoints of unobservable Wilson lines in the presence of charged matter. Read the corrected disorder-operator argument. An Abelian shortcut is not a general replacement for a non-Abelian monopole boundary condition.

Keep the theory and the notation fixed

Our main theory is compact $U(N)_k\times U(N)_{-k}$, equal $N\geq1$, integer $k>0$, with no further quotient of the centers. Each factor has rank $N$. The flat parent generically has twelve real Poincaré supercharges; the chiral-superfield notation makes four manifest. The paper’s full-action symmetry argument concerns the generic $\mathcal N=6$ theory. The exceptional levels one and two require monopole information, not just a new reading of the letter $N$.

Source notation	Meaning in the canonical lessons
$A_1,A_2$ and $B_1,B_2$	Independent chirals in opposite bifundamental representations; $B$ is not $A^\dagger$
$C_I=(A_1,A_2,B_1^*,B_2^*)$	The scalar tuple $Y^I=(A_1,A_2,B_1^\dagger,B_2^\dagger)$, all in the same color representation; it is not a tuple of four chiral superfields
$\operatorname{Tr}$ in (2.5)	Ordinary fundamental trace, with no division by $N$; the kinetic terms fix the field normalization
$\varphi_{(1)},\varphi_{(2)}$ in (2.4)	Algebraic adjoint chiral variables used to generate the quartic interaction, not additional propagating matter
$\sigma_{(1)},\sigma_{(2)}$ in (A.3)	The vector scalars $\sigma,\widehat\sigma$ fixed by the signed matter-density matrices
$A_b=A_{(1)}-A_{(2)}$, $A_{\widetilde b}=A_{(1)}+A_{(2)}$	Difference and sum of compact connections; their flux integers are correlated
$\tau$ near (2.9)	A compact dual angle, not the axio-dilaton also called $\tau$ in (3.1) and Appendix B
$R$ in (4.2)	The seven-sphere radius; the AdS radius is $R/2$

For the Abelian gauge transformations in section 2.3, $a=-A_{(1)}$ and $\widehat a=-A_{(2)}$ translate the source’s $D=d+iA_{(1)}-iA_{(2)}$ to our $D=d-ia+i\widehat a$. This statement alone does not translate a Lorentzian action, a Euclidean weight, or a monopole charge. The adjacent action-and-flux dictionary fixes those together. In particular, our Euclidean first-node term is $+ik\,\mathrm{CS}(a)/(4\pi)$ with weight $e^{-S_E}$.

Read the action and its symmetry argument

For this passage, complete the quartic interaction, auxiliary elimination, and full-action symmetry argument. Then read the original in this order:

Verified passage	Question to answer	Canonical calculation
(2.1)–(2.2), printed pp. 3–4/PDF pp. 4–5	Which vector fields have algebraic equations rather than Yang–Mills kinetic terms?	Opposite levels and complete multiplets
(2.4)–(2.6), printed pp. 5–6/PDF pp. 6–7	How does eliminating the two adjoint chirals fix the relative sign and coefficient?	Recover the coefficient
(A.1)–(A.4), printed p. 30/PDF p. 31	Why is the F-term potential only part of the answer?	Eliminate the vector auxiliaries
(A.5)–(A.7), printed p. 31/PDF p. 32	Which ordered sextic invariants combine into an $SU(4)$ expression?	A finite R-symmetry rotation
Section 2.2 after (2.6), printed p. 6/PDF p. 7	What acts on the supercharges, rather than only on scalar polynomials?	Full-action enhancement

Your action checkpoint is

$$W=\frac{4\pi}{k}\operatorname{Tr} (A_1B_1A_2B_2-A_1B_2A_2B_1). \tag{PC13.1}$$

Its coefficient is not arbitrary once the complete theory is fixed. Its derivatives have three fields, so their squared norms are sextic. The vector scalars supply other sextic terms. On the physical bosonic slice, (A.3) gives $\sigma=(2\pi/k)\sum_i(A_iA_i^\dagger-B_i^\dagger B_i)$ and $\widehat\sigma=(2\pi/k)\sum_i(A_i^\dagger A_i-B_iB_i^\dagger)$. Use these particular equations when comparing the potential; a generic sign convention from the review preceding (2.2) is not a replacement for the Appendix A dictionary.

The symmetry check is stronger than observing an invariant scalar potential. The source begins with a complete $\mathcal N=3$ action and combines its R-symmetry with flavor symmetries of the full multiplets. The canonical lesson supplies that logic. Neither a quartic charge count nor an $SU(4)$-invariant polynomial alone proves twelve supercharges.

Read the vacuum and operator arguments

Prepare with the one-brane quotient, many-brane global data, monopole dressing, and exceptional-level enhancement.

Verified passage	Question to answer	Canonical calculation
Section 2.3, (2.7)–(2.9), printed pp. 7–8/PDF pp. 8–9	Why does the phase quotient remain finite?	Compact dualization
Paragraph after (2.9), printed p. 8/PDF p. 9	What is permuted on the diagonal branch?	Permute whole tuples
Section 2.4, (2.10) and footnote 6, printed pp. 10–11/PDF pp. 11–12	Which flux cancels which matter charge?	Derive the CS gauge weight
Section 2.4, printed pp. 11–12/PDF pp. 12–13	Which further input makes the special operators into symmetry currents?	Protection and nonzero descendants
Sections 2.5–2.6, printed pp. 13–15/PDF pp. 14–16	Why are the $SU(N)$ variants separate global theories?	The global group matters

Here is the first calculation to reproduce while reading. For rank one, all scalar entries commute, and both contributions to the potential vanish without imposing a moment-map equation. In our integral gauge basis $b=a-\widehat a$, $c=\widehat a$, compact dualization gives

$$kb+d\tau=0,\qquad \tau\sim\tau+2\pi,\qquad (Y,\tau)\longmapsto(e^{i\alpha}Y,\tau-k\alpha). \tag{PC13.2}$$

The gauge choice $\tau=0$ leaves $\alpha=2\pi j/k$, modulo $2\pi$. Equivalently, the locally invariant lift $X=e^{i\tau/k}Y$ changes by $e^{2\pi i/k}$ when the angle’s lift changes by $2\pi$. Thus $\mathcal M_1=\mathbb C^4/\mathbb Z_k$, with one common phase on all four coordinates. There are still eight real coordinates; a finite identification does not remove a dimension.

This is the same quotient as AB (2.7)–(2.9). To follow the paper’s sum/difference variables, retain $n_+=n_1+n_2$ and $n_-=n_1-n_2$, so $n_+\equiv n_-\pmod2$. Do not turn them into two independent unit-flux circles. The canonical derivation also retains the self-CS term and the global definition on nontrivial bundles.

At higher rank the standard diagonal branch is an unordered collection of these four-component points, $\operatorname{Sym}^N(\mathbb C^4/\mathbb Z_k)$. The paper argues for the moduli space using generic massive off-diagonal modes; the lesson explicitly identifies what the diagonal-sector calculation proves. It is not a classification of every degenerate matrix solution by inspection.

There are two further reading cautions. The original paper motivates the special-level enhancement through operators and branes; the later field-theory protection argument is identified separately in the linked enhancement lesson. Also read footnote 8, printed p. 12/PDF p. 13: a spectrum inferred from the proposed moduli space is not an independent test of that same moduli-space proposal.

Branes and holographic regimes

The additional preparation for this section is the ABJM holographic dictionary. It defines the string and Planck lengths, flux normalization, and radius/coupling map used here.

Read section 3.1 through figure 4, printed pp. 15–18/PDF pp. 16–19, as an ultraviolet construction. The D3-branes wrap a circle cut into two intervals by fivebranes; the intervals explain the two gauge factors. Rotating one fivebrane produces a Yang–Mills–Chern–Simons theory with massive vector partners. Taking energies below their mass gives the pure CS-matter theory. Do not put these massive Yang–Mills modes back into the conformal action.

Sections 3.2–3.3, printed pp. 19–20/PDF pp. 20–21, change the description: dualities turn the fivebrane data into circles that shrink in an eight-dimensional geometry, while the D3-branes become M2-branes. Appendix B, especially (B.9)–(B.12), printed pp. 34–35/PDF pp. 35–36, tracks the angular periods that produce the same finite quotient near the intersection. This is a brane/duality argument for the interpretation of the field theory, not a consequence of the quartic polynomial alone.

For the gravity parameters, the exact landmarks are section 4.1, (4.1)–(4.7), printed pp. 21–23/PDF pp. 22–24. With $R$ the seven-sphere radius, $\ell_p$ the eleven-dimensional Planck length, $\ell_s^2=\alpha'$ and $g_s=e^\phi$, the classical map gives

$$\frac{R^6}{\ell_p^6}=32\pi^2Nk,\qquad \frac{L_{\rm fiber}}{2\pi\ell_p}=\frac{R}{k\ell_p}, \qquad \frac{R_{\rm str}^2}{\alpha'}=2^{5/2}\pi\sqrt{\lambda}, \qquad g_s^2=2^{5/2}\pi\frac{N^{1/2}}{k^{5/2}}, \quad \lambda=\frac Nk . \tag{PC13.3}$$

$R_{\rm str}$ is the overall internal radius in the string-frame metric; the AdS radius has an extra factor $1/2$. The source sets $\alpha'=1$ in (4.5). These are the displayed leading supergravity relations, not an assertion that finite-charge corrections have vanished.

Small curvature and a large eleven-dimensional fiber require $Nk\gg1$ and $N\gg k^5$, up to the fixed numerical constants. The weakly coupled, weakly curved type-IIA window instead requires $1\ll\lambda$ and $\lambda^{5/2}/N^2\ll1$. A large level is not by itself either condition. The exact one-brane quotient did not require any of them.

Section 4.1 then discusses thermal free energy in (4.8)–(4.9), printed p. 24/PDF p. 25. It is not the later localized three-sphere partition function. Sections 4.2–4.3 address wrapped objects and high-spin operators; they are subsequent applications, not needed to reconstruct the defining action. In particular, the global boundary-condition questions in section 4.2 cannot be settled by deleting a central gauge field locally.

Restore the circle patching in the quotient

The metric in AB (4.3)–(4.4) compresses a useful global step. A Hopf fiber is the common phase of a unit vector in $\mathbb C^4$; its base is the space of complex lines, $\mathbb{CP}^3$. Restrict to $z_3=z_4=0$ to inspect one $\mathbb{CP}^1$, an ordinary two-sphere. This restriction tests the integer twisting without requiring the full base metric.

On the northern patch write $z_1=e^{i\psi_N}\cos(\theta/2)$ and $z_2=e^{i(\psi_N+\phi)}\sin(\theta/2)$. Both $\psi_N$ and $\phi$ have period $2\pi$, with $0\leq\theta\leq\pi$. The real connection of the common phase is $-i\sum\overline z_i\,dz_i=d\psi_N+\omega_N$. On the southern patch use $\psi_S=\psi_N+\phi$. Direct differentiation gives

$$\omega_N=\frac{1-\cos\theta}{2}\,d\phi,\qquad \omega_S=-\frac{1+\cos\theta}{2}\,d\phi,\qquad d\omega=\frac12\sin\theta\,d\theta\wedge d\phi . \tag{PC13.4}$$

The northern potential is regular at the north pole and the southern one at the south pole. They differ by $d\phi$, while $d\psi+\omega$ is unchanged.

Now quotient by the common phase of order $k$. A normalized fiber angle is $\varphi=k\psi$, of period $2\pi$. It obeys $\varphi_S=\varphi_N+k\phi$; the global normalized connection is $\Theta=d\varphi+k\omega$. Hence, with the indicated positive base orientation,

$$\frac1{2\pi}\int_{\mathbb{CP}^1}d(k\omega)=k,\qquad ds^2_{\rm fiber}=\frac{R^2}{k^2}\Theta^2,\qquad L_{\rm fiber}=\frac{2\pi R}{k}. \tag{PC13.5}$$

The integer is the circle bundle’s first Chern number in this orientation: its transition function winds $k$ times. The quotient leaves a circle with a shorter period and different twisting. Quotienting by the entire continuous phase would instead remove that direction and produce $\mathbb{CP}^3$, which is not $S^7/\mathbb Z_k$. This explains why a finite vacuum quotient and a later dimensional reduction are different operations.

Exit calculations

1. Check a function across the two patches

Set $k=3$. On the northern patch of the restricted $S^3/\mathbb Z_3$, take $f_N=e^{i\varphi_N}\cos^3(\theta/2)$. Find its southern expression. Does it define a smooth scalar? Does $e^{i\varphi_N/3}\cos(\theta/2)$ define one? Use the metric below to check the allowed function’s kinetic eigenvalue:

$$ds^2=R^2\left[ \frac{(d\varphi+3\omega)^2}{9} +\frac14(d\theta^2+\sin^2\theta\,d\phi^2)\right].$$

Hint 1

Use $\varphi_N=\varphi_S-3\phi$. Near the south pole, $\cos(\theta/2)e^{-i\phi}$ is a smooth complex local coordinate.

Hint 2

For a function independent of $\phi$ in the northern gauge, the needed dimensionless Laplacian is $R^2\Delta f=4(f_{\theta\theta}+\cot\theta f_\theta)+[9+36\omega_{N,\phi}^2/\sin^2\theta]f_{\varphi\varphi}$.

Solution

The southern expression is $f_S=e^{i\varphi_S}e^{-3i\phi}\cos^3(\theta/2)$. It is smooth: the last two factors form the cube of a smooth local coordinate. The northern expression is smooth at the other pole and periodic under $\varphi_N\to\varphi_N+2\pi$. Thus it descends from the invariant polynomial $z_1^3$ on the covering sphere.

The proposed cube root changes by $e^{2\pi i/3}$ around the fiber, so it is not a scalar function on the quotient, although it is a good local expression. For the allowed function, put $v=\cos^3(\theta/2)$. Then $v'/v=-3\tan(\theta/2)/2$ and $v''/v=9\tan^2(\theta/2)/4-3\sec^2(\theta/2)/4$. Substitution into the displayed Laplacian, using $\omega_{N,\phi}=(1-\cos\theta)/2$ and $f_{\varphi\varphi}=-f$, gives $-\Delta f=15f/R^2$. The kinetic energy is positive, and the eigenvalue matches degree-three scalar harmonics on the round covering $S^3$. This is a geometric mode check, not an identification with an AdS particle dimension.

2. Translate a flux insertion with its action

Keep the site’s compact rank-one theory at $k=4$. A positive matched unit flux is imposed by $df_c=2\pi\delta_x$ in the first-order path integral with multiplier term $-i\int\tau\,df_c/(2\pi)$. A $B$ scalar has relative gauge charge $-1$. Which of $e^{i\tau}B_1^4$ and $e^{-i\tau}B_1^4$ has the required flux and is gauge invariant? Why can the source’s printed positive exponent near (2.10) not be copied in isolation?

Hint 1

The Euclidean weight is $e^{-S_E}$. Write the Fourier delta imposing $df_c-2\pi\delta_x=0$ before reading the insertion’s sign.

Hint 2

Under a relative gauge transformation, $\tau\to\tau-4\alpha$ and $B_1\to e^{-i\alpha}B_1$. Check both factors.

Solution

The delta representation contains $\exp[i\int\tau\,df_c/(2\pi)-i\tau(x)]$. Thus $e^{-i\tau}$ creates the stated positive flux. Its weight is $+4$, canceled by the $-4$ of $B_1^4$. The other exponent creates the opposite flux and has weight $-4$, so its product with $B_1^4$ has weight $-8$.

The original paper’s integrated equation uses its own connection, Lorentzian level, orientation, and dualization convention. The canonical dictionary translates these together: with fixed Wick orientation, $A_{\rm AB}=-a$, $n_{\rm AB}=-m$ and $k_{\rm AB}=-k$ in that gauge-sector comparison. A bare exponent is not an invariant convention. This calculation checks gauge neutrality and flux; identifying a protected chiral dimension additionally requires the monopole boundary condition, vacuum contribution, and shortening assumptions.

3. Change the integer sequence, not the quotient argument

Compare two sequences of standard theories as integer $n\to\infty$: $(N,k)=(n^6,n)$ and $(N,k)=(n^6,n^2)$. Find the scaling of the M-theory circle in Planck units, the string curvature parameter $\alpha'/R_{\rm str}^2$, and $g_s^2$. Which classical description is parametrically appropriate? Does either answer alter the exact rank-one phase identification at a fixed $k$?

Hint 1

Use $R/(k\ell_p)\propto(N/k^5)^{1/6}$, $\alpha'/R_{\rm str}^2\propto(N/k)^{-1/2}$ and $g_s^2\propto N^{1/2}/k^{5/2}$.

Hint 2

Small string curvature does not imply small string coupling. Keep the two tests separate.

Solution

In the first sequence, $\lambda=n^5$, the M-theory circle grows as $n^{1/6}$, the string curvature parameter falls as $n^{-5/2}$, and $g_s^2$ grows as $n^{1/2}$. The eleven-dimensional radii are large, while weakly coupled type IIA fails.

In the second, $\lambda=n^4$, the M-theory circle shrinks as $n^{-2/3}$, the string curvature parameter falls as $n^{-2}$, and $g_s^2$ falls as $n^{-2}$. The weakly curved, weakly coupled IIA description is appropriate; treating the shrinking circle as a large eleven-dimensional direction is not.

These are parametric statements in the displayed classical dictionary. Both sequences change $N$ and $k$ and are neither a fixed-level nor a fixed-ratio limit. Neither calculation changes the finite quotient derived from the compact gauge theory. In particular, an exact one-brane vacuum space at a chosen $k$ is not a large-rank gravity approximation.

Return to the introduction after these checks. You should now be able to identify which conclusions follow from the action, which use compact gauge data, which invoke a brane or protected-multiplet argument, and which require a separate parametric limit.