1: \chapter{The Basu-Harvey Equation and M2-M5 Intersections}\label{BasuHarvey}
2:
3: This section is primarily concerned with the work of Basu and Harvey\cite{BasuH}, who put forward a proposal for the generalisation of the Nahm equation (\ref{Nahmeq}) that should arise in M2-M5 system. They also presented a solution to this equation representing coincident membranes ending on a five-brane.
4:
5: As the action for multiple M2-branes is not known, it is a case of using this work to gain insight into what this action should be, rather than having an action from which to derive it. We refer to the equation they put forward as the Basu-Harvey equation, and we will see that it is deduced by requiring properties we would expect from generalising the D1-D3 case to a fuzzy three-sphere cross-section, rather than a fuzzy two-sphere. As Appendix \ref{app:FS3} describes, the fuzzy three-sphere is more complicated than the fuzzy two-sphere. In fact it is more complicated than the fuzzy four-sphere as odd spheres in general are harder to work with.
6:
7: The solution can be compared with the self-dual string of the previous section, as well as the energies expected for such a configuration and the properties of fluctuations around the solution. After these checks one can feel a bit more confident and look at some terms we might expect in multiple M2-brane action if the Basu-Harvey equations is to appear as a Bogomol'nyi equation.
8:
9: The Basu-Harvey equation, and its solution in terms of the fuzzy three-sphere construction given in Appendix \ref{app:FS3}, is not without its drawbacks, not least the fact that the degrees of freedom are $N\times N$ matrices when the expectation was for $N^{3/2}$ degrees of freedom for $N$ coincident membranes.
10:
11:
12: \section{Properties Expected for the Solution}
13:
14: We know that the M5 picture of the intersection is in terms of the self-dual string, described in Section \ref{sec:sds} and pictured in Figure \ref{pic:M2-M5}. Thus we expect a relationship between the radius, $R$, in the directions transverse to the membrane yet tangent to the five-brane (i.e. $R=\sqrt{(X^1)^2+(X^2)^2+(X^3)^2+(X^4)^2}\,$), and $s$, the M2 worldvolume direction away from the five-brane, of the form
15: \beq\label{ridge}
16: s\sim \frac{Q}{R^2},
17: \eeq
18: as implied by (\ref{sds}) (Q is the quantised charge, given by the number of membranes). As the self-dual string was a static solution with no dependence on $\sigma$, the co-ordinate along the string, the active scalars should only depend on $s$ for the membrane description as well.
19:
20: In the D1-D3 case there were three active scalars transverse to the string in the directions of the D3-brane worldvolume. This meant the cross-section was a (fuzzy) two-sphere, as expected for a spike. In our M-theory system we have an extra transverse scalar and the cross-section should be a (fuzzy) three-sphere, giving the $SO(4)$ symmetry we would expect between the scalars. As well as this $SO(4)$ invariance we also demand translation invariance.
21:
22: The fuzzy $S^3$ co-ordinates, $G^i$, obey the equation
23: \beq\label{FS3}
24: G^i +\frac{1}{2(n+2)}\epsilon_{ijkl}G_5G^jG^kG^l =0,
25: \eeq
26: where $G_5$ is a difference of projection operators, it obeys $\{G_5,G^i\}=0$ and $(G_5)^2=1$ (full details can be found in Appendix \ref{app:FS3}). This equation is a quantised version of a higher Poisson bracket equation obeyed by the classical 3-sphere (as $[\alpha^a,\alpha^b]=2i\epsilon_{abc}\alpha^c$ is for the two sphere; again more details are contained in Appendix \ref{app:PN}). The fuzzy three-sphere relation (\ref{FS3}) was first obtained in \cite{BasuH} and we will derive it in full in Appendix \ref{app:G4b}. It is around this relation that the Basu-Harvey equation is built.
27:
28: \section{The Basu-Harvey Equation}
29:
30: The Basu-Harvey equation is given by
31: \beq\label{BaH}
32: \frac{dX^i}{d s}+\frac{M_{11}^3}{8\pi\sqrt{2N}}\frac{1}{4!}\epsilon_{ijkl}[G_5,X^j,X^k,X^l]=0.
33: \eeq
34: The anti-symmetric 4-bracket is a sum over permutations with sign, e.g.
35: \beq\label{N4b}
36: [X^1,X^2,X^3,X^4]=\sum_{\mbox{perms}\ \sigma} \mbox{sign}(\sigma)X^{\sigma(1)}X^{\sigma(2)}X^{\sigma(3)}X^{\sigma(4)}\, ,
37: \eeq
38: and it can be thought of as a quantum Nambu bracket\cite{Nambu,CZ}. We could have placed such a bracket on the second term of (\ref{FS3}) if we had included the correct combinatorial factor. The equation is translation invariant under $X^i\rightarrow X^i+v^iI$, where $I$ is the identity.
39:
40: $G_5$ and the scalars will belong to an algebra containing the fuzzy three-sphere. There are three main possibilities for what this algebra is. The first is \mnc, the algebra of $N \times N$ matrices, where $N$ is the dimension of the representation of $SO(4)$ in which the the fuzzy three-sphere we work with lies. For the fuzzy three-sphere only, this dimension coincides with the square of the radius in terms of the $G^i$ (i.e. $\sum_i G^iG^i=N$). $N$ is what we will identify with the number of membranes, and this is the $N$ that appears in the Basu-Harvey equation (\ref{BaH}) above\footnote{The original version of the Basu-Harvey equation did not contain this factor due to an error in calculating (\ref{FS3}). In an updated version of their paper they included a parameter $\lambda$ in the numerator rather than the $1/\sqrt{2N}$ factor we have included following \cite{Nogradi, bermancopland}. $\lambda$ was treated as a coupling with $\lambda^2 N$ required to be fixed in the large-$N$ limit for consistency.}. A second possibility for the algebra is that generated by the $G^i$, which is a sub-algebra of \mnc. The third option is the algebra which in the large-$N$ limit agrees with classical algebra of functions on $S^3$, namely the spherical harmonics. Though this third algebra is important in later sections, we will assume our fields are in \mncs at present.
41:
42: By analogy with the D1-D3 system the Basu-Harvey equation should appear as a Bogomol'nyi equation for minimising the energy, and should also follow from the vanishing of the supersymmetry variation of the Fermions on the membrane.
43:
44:
45: \section{The Membrane Fuzzy Funnel Solution}
46: Similar to the D1-D3 case we expect a static solution with scalars proportional to the fuzzy 3-sphere co-ordinates $G^i$ and only depending on $s$. An ansatz
47: \beq
48: X^i(s)=f(s)G^i
49: \eeq
50: leads quickly to the solution
51: \beq\label{soln}
52: X^i(s)=\frac{i\sqrt{2\pi}}{M_{11}^\frac{3}{2}}\frac{1}{\sqrt{s}}G^i.
53: \eeq
54: The physical radius is given by
55: \beq\label{traces2}
56: R=\sqrt{\left|\frac{\mbox{Tr} \sum(X^i)^2}{\mbox{Tr} \openone}\right|}.
57: \eeq
58: Substitution and rearranging results in
59: \beq
60: s\sim \frac{N}{R^2} \, ,
61: \eeq
62: which is the self dual string behaviour we expect when we identify $N$ with the number of membranes (c.f (\ref{ridge})).
63:
64: If this is the Bogomol'nyi equation we would expect that we could also minimise the energy by satisfying (\ref{BaH}) with a sign difference between the terms, leading to the opposite topological charge. In this case we can solve by removing the factor of $i$ from (\ref{soln}) or by multiplying each co-ordinate by $G_5$.
65:
66: \section{Energy and Action for Multiple Membranes}\label{enact}
67:
68: Given that the Basu-Harvey equation should arise as a Bogomol'nyi equation, as the Nahm equation did in the D1-D3 system (and as the generalised version does in (\ref{D1bog})), it is not unreasonable to define the energy of our static configuration with four non-zero scalars as in \cite{BasuH}
69: \bea \label{m2energy}
70: E = T_2 \int d^2 \sigma {\rm Tr} \bigg[ \left( \frac{d X^i}{ds} +
71: \frac{M_{11}^3}{8\pi\sqrt{2N}}\frac{1}{4!} \epsilon_{ijkl}
72: [G_5, X^j, X^k, X^l] \right)^2 \cr
73: + \left(1- \frac{M_{11}^3}{16\pi \sqrt{2N}} \frac{1}{4!}\epsilon_{ijkl} \left\{ \frac{d X^i}{ds} ,
74: [G_5, X^j, X^k, X^l] \right\} \right)^2 \bigg]^{1/2} .
75: \eea
76: The membrane tension $T_2$ is given by $T_2=M_{11}^3/(2\pi)^2$ and the integral is over the two spatial worldvolume directions $\sigma$ and $s$. In what follows we will consider the $X^i$ to obey $\{G_5,X^i\}=0$. In terms of the fuzzy three-sphere algebra this means restricting the $X^i$ to lie in $Hom(\cRp,\cRm)$ or $Hom(\cRm,\cRp)$. This is of course obeyed by the co-ordinates $G^i$. It means, for instance, that by multiplying out the squares in (\ref{m2energy}) $G_5$ can be eliminated. We can also remove the 4-bracket and combinatorial factor in (\ref{m2energy}) as the $\epsilon$ provides the anti-symmetrisation (though the bracket will be required in the action for translation invariance).
77:
78: If the scalars obey the Basu-Harvey equation then Bogomol'nyi bound is satisfied and the energy density linearises:
79: \beq
80: E = T_2 \int d^2 \sigma {\rm Tr} \left( 1- \frac{M_{11}^3}{8\pi \sqrt{2N}} \epsilon_{ijkl}
81: \frac{d X^i}{ds} G_5 X^j X^k X^l\right).
82: \eeq
83: The second term is a boundary term, and the boundary is the self dual string. Using the solution, the energy can be rewritten for large $N$ as
84: \beq\label{BHenergy}
85: E = N T_2 L \int ds + T_5 L \int 2\pi^2 dR R^3,
86: \eeq
87: where we have used $T_5=M_{11}^6/(2\pi)^5$ and $L$ is the length of the string. These two terms have the energy densities you would expect for $N$ membranes and a single five-brane respectively.
88:
89: As per the discussion of Section (\ref{validity}) we would expect this analysis to be only valid at the core ($s\rightarrow\infty$), though for large $N$ it agrees with the M5-brane picture, a description which should only be valid in the opposite limit. Examining (\ref{m2energy}) we can see that a Taylor expansion in terms of powers of $X^i$ is valid when $M_{11}^6R^6N^3\ll 1$, that is $R\ll \sqrt{N}M_{11}^{-1}$. Thus if $N$ is large, $R$ can be large as well.
90:
91: Given an expression for the energy such as (\ref{m2energy}) we can expand and deduce terms in the associated action. On generalising away from a static solution with only four non-zero scalars dependent only on $\sigma$, we expect terms like
92: \bea \label{m2act}
93: S &=& -T_2 \int d^3 \sigma {\rm Tr} \bigg[ 1 +
94: \left( \p_a X^M \right)^2
95: -\frac{1}{2N.3!}[X^M,X^N,X^P][X^M,X^N,X^P]
96: \non\\
97: &+&\frac{1}{2N.4.3!} \left[ \p_a X^L ,[X^M,X^N,X^P] \right] \times
98: \bigg( \left[\p^a X^L ,[X^M,X^N,X^P] \right]
99: \non\\&+&\left[ \p^a X^M
100: ,[X^L,X^P,X^N] \right] + \left[ \p^a X^N ,[X^L,X^M,X^P] \right]\non\\ &+&\left[
101: \p^a X^P ,[X^L,X^N,X^M] \right]
102: \bigg) +\ldots \bigg]^{1/2} \eea
103: in a multiple membrane action. $L,M,\ldots$ labels the 8 transverse directions and $a,b,\dots$ the 3 worldvolume directions. The three-bracket used is defined analogously to the quantum Nambu 4-bracket (\ref{N4b}) as a sum over the six permutation of the entries, with sign.
104:
105:
106: \section{Fluctuations on the Funnel}
107:
108: Basu and Harvey also performed a fluctuation analysis on the membrane fuzzy funnel similar to that performed on the D-string funnel in Section (\ref{furext}). The simplest fluctuations to consider are those in the four directions transverse to both the membrane and the five-brane. The analysis in the D-brane was linear, relying only on the dimensionally reduced super Yang-Mills action, which contained the kinetic term as well as a quartic potential. Basu and Harvey thus took a general kinetic term and the sextic coupling suggested in the last section as being sufficient for a linear analysis of the the membrane fluctuations.
109: In flat space and static gauge, the pull back of the metric is given by $P[G]_{ab}=\eta_{ab}+\partial_aX^M\partial_bX^M$, taking the determinant will lead to the first two terms of (\ref{m2act}). Therefore the action used for fluctuation analysis is
110: \beq\label{fluctact}
111: S=-T_2 \int d^3\sigma\mbox{Tr} \sqrt{-\det(P[G]_{ab})-\frac{1}{2N}\frac{1}{3!}[X^M,X^N,X^P][X^M,X^N,X^P]}\, .
112: \eeq
113: The fluctuations may depend on all three worldvolume co-ordinates and are proportional to the identity in the fuzzy sphere algebra, $\delta X^m(t,s,\sigma)=f^m(t,s,\sigma) \openone_N$. We keep terms up to quadratic order in the fluctuations, which gives
114: \beq
115: [X^M,X^N,X^P][X^M,X^N,X^P]=3(f^m)^2[X^i,X^j]^2\, ,
116: \eeq
117: where $M,N,P$ run over all indices, $m$ runs over the directions transverse to the both branes and $i,j$ run over the non-zero scalars of the solution. Evaluation of the commutator squared on the right-hand side proceeds via
118: \beq
119: [G^i, G^j]^2 = 2 G^i G^j G^i G^j -2 N^2\openone
120: \eeq
121: and
122: \beq\label{ijij}
123: G^i G^j G^i G^j\Prp=-(n+1)(n+3)\Prp=-2N\Prp.
124: \eeq
125: Finally this leads to
126: \beq\label{ijsq}
127: [G^i, G^j]^2=-2N(N+2)\, .
128: \eeq
129: Note that (\ref{ijij}) and (\ref{ijsq}) correct an error in \cite{BasuH}. However, the leading order in $n$ behaviour of (\ref{ijsq}) remains unchanged and so the conclusions below remain the same.
130:
131: Returning to the action (\ref{fluctact}) we now have
132: \beq
133: S = -T_2 \int d^3 \sigma {\rm Tr} \sqrt{ H -H (\p_t f^m)^2 +(\p_s f^m)^2
134: +H (\p_\sigma f^m)^2 +\frac{N+2}{2s^2} (f^m)^2 },
135: \eeq
136: where
137: \beq
138: H=1+\frac{\pi N}{2M_{11}^3s^3}.
139: \eeq
140: Here also the potential term differs from that of \cite{BasuH} by an additional factor of $1/2$. This is due to our use of the anti-symmetric 3-bracket squared $[X^M,X^N,X^P]^2$, as opposed to $Q^{MNP}H^{MNP}$, where $Q^{MNP}=\{[X^M,X^N],X^P\}$ and $H^{MNP}=Q^{MNP}+Q^{NPM}+Q^{PMN}$. With the correct factor these terms are the same for $M,N,P=i,j,k$, but differ when considering fluctuations in overall transverse directions.
141:
142: The equation of motion for the linearised fluctuations becomes
143: \beq
144: (H \p_t^2 -\p_s^2 -H \p_\sigma^2 ) f^m (t,s,\sigma)
145: +\frac{N+2}{2s^2} f^m (t,s,\sigma) =0.
146: \eeq
147: In the $s\rightarrow\infty$ limit (where we have a flat membrane) the equation of motion reduces to
148: \beq
149: (-\p_t^2 +\p_s^2 +\p_\sigma^2) f^m =0.
150: \eeq
151: The solutions to this equation are plane waves with $SO(2,1)$ symmetry in the worldvolume directions, as one would expect for a membrane.
152: Although in the opposite limit the analysis should not be valid, as per the earlier discussion we keep $N$ large and find agreement with what we would expect. As $s\rightarrow 0$, $H\sim s^{-3}$ and the equation of motion gives
153: \beq
154: (-\p_t^2 + \p_\sigma^2) f^m + R^{-3} \frac{\p }{\p R}
155: \left( R^3 \frac{\p f^m}{\p R} \right) =0.
156: \eeq
157: This has exactly the $SO(2,1)\times SO(4)$ symmetry that we would expect for the M5 worldvolume with string soliton.
158:
159: Analysis of more complicated fluctuations and in other directions has not been completed due to the increased complexity of the fuzzy three-sphere algebra compared with that of the fuzzy two-sphere.
160:
161: \section{Issues with the Basu-Harvey Picture}
162:
163: The main issue with the Basu-Harvey picture of M2-M5 intersections is that the membrane degrees of freedom for $N$ coincident membranes are represented by $N\times N$ matrices. This is also the case in the D1-D3 system, but here there is an understanding of how these degrees of freedom arise as light fundamental strings stretching between the branes. However in M-theory there are various indications \cite{KT,Kleb} that there should be $N^{3/2}$ degrees of freedom, but there is no picture of how this number (or $N^2$) degrees of freedom would arise.
164:
165: One suggested resolution of the apparent contradiction\cite{BasuH} is the Basu-Harvey picture is an ultra-violet description which flows to an infra-red description in terms of $N^{3/2}$ degrees of freedom - this is the case for D2-branes (as discussed in Section 6.1 of \cite{AGMOO}). In later sections we will pursue the idea that we should use a different algebra than \mncs to describe the fuzzy three-sphere correctly in the large-$N$ limit.
166:
167: Other weaknesses of the construction that needed to be investigated were a lack of dimensional reduction to the Nahm equation and D1-D3 system, and lack of a supersymmetry transformation which link satisfying the Basu-Harvey equation to preserving supersymmetry. Again, we will address these issues in later chapters.
168:
169: \newpage
170: