64 Chapter 4. <strong>Bethe</strong>/gaugerepresentation, while a hypermultiplet contains two N = 1 chiral multiplets transforming inconjugate representations of the gauge group.Consider an N = 2 supersymmetric U(N) gauge theory in four dimension with L hypermultipletsin the (anti)fundamental representation. When we reduce the theory to two dimensionwe precisely get the superfield content we’re after. Just like the four real supercharges ofN = 1 combine to give the two-dimensional N = (2, 2) supersymmetry algebra, the eight superchargesof N = 2 supersymmetry now give N = (4, 4) supersymmetry. Turning on the twistedmasses (4.6) breaks the supersymmetry down to N = (2, 2). Indeed, nonzero ˜m a = iu givesmass to the components of Φ a , lifting the symmetry between Φ a and the vector superfield V .This method, where we introduce terms to the Lagrangian that explicitly violate (a part of)supersymmetry, is called soft supersymmetry breaking.Superpotential. <strong>The</strong> inhomogeneities ν l are complex numbers. Thus, the identification (4.6)of the (anti)fundamental twisted masses with the ν l and local spins is vacuous if the s l wouldbe complex as well. In fact, the <strong>Bethe</strong> sides requires half-integer values s l ∈ 1 2N for each of thelocal spins. We have to limit the possible values of our twisted masses.Recall from Section 3.4.2 that, by the decoupling theorem, a superpotential term for thematter fields does not influence the vacuum equations. On the other hand, in Section 3.2.3 wehave seen that the superpotential usually breaks the global flavour symmetry group H max downto a subgroup H. Since the twisted mass terms are obtained by weakly gauging the flavoursymmetry group, we can use a superpotential to arrange that s l ∈ 1 2 N.Consider the family of ‘ ˜QΦQ’ superpotentials of the formW (¯Φ¯f, Φ a , Φ f ) =L∑l,l ′ =1∑s¯Φ¯f,l (m s ) l l ′ (Φ a) 2s Φ l′f (4.7)involving powers of the adjoint superfield. We can view this superpotential as a generalizationof the complex mass term (3.42), with Φ a -dependent complex massm(Φ a ) l l ′ = ∑ s(m s ) l l ′ (Φ a) 2s .As usual we suppress the colour indices of the fields; notice that (4.7) has the structure of a colourinner product, and is in particular gauge-invariant (use holomorphic functional calculus forgeneral s ∈ C). In Section 3.4.2 we saw that chiral superfields have mass dimension zero in twospacetime dimensions, so [(m s ) l l ′] = 1 and the superpotential (4.7) is super-renormalizable forany value of s ∈ C. Nevertheless, in order to avoid poles and branch cuts for the superpotentialit is natural to restrict to s ∈ 1 2 N.To get the values (4.6) for the twisted masses we take the complex mass matrices diagonalin flavour space: (m s ) l l ′ = ϖ l δl l ′ δs ls . (<strong>The</strong> coefficients ϖ l ∈ C have nothing to do with thefunction ϖ s used to write down the Yang-Yang function (4.2).) Plugging this into (4.7) yieldsthe following superpotentialW (¯Φ¯f, Φ a , Φ f ) =L∑ϖ l ¯Φ¯f,l (Φ a ) 2s lΦ l f . (4.8)l=1<strong>The</strong> residual flavour symmetry transformations are given byΦ l f ↦−→ e (ν l−is l )u Φ l f ,Φ a ↦−→ e iu Φ a ,¯Φ¯f,l ↦−→ ¯Φ¯f,l e (−ν l−is l )u .(4.9)Upon weakly gauging this residual symmetry group we get twisted masses (4.6) with ν l ∈ Cand s l ∈ 1 2N as desired.
4.2. Further topics 654.1.3 DictionaryWe can summarize everything we have seen so far in the dictionary that is provided by the<strong>Bethe</strong>/gauge correspondence:<strong>Bethe</strong><strong>Gauge</strong>N = (2, 2) sym withmassive matter,L a = 1, ˜m a = iu, andsuperpotential (4.7)low energy limitspin chain ←→vacuum structureon Coulomb brachbae ←→ vacuum equationsN-particle sector ←→ G = U(N)rapidities λ n ←→ susy vacua σ nlength L ←→ L f = L¯f = LY (λ) ←→ ˜Weff (σ)twist parameter ϑ ←→ τ = ir + ϑ/2πlocal spins s l andinhomogeneities ν l←→˜m l f = (ν l − is l ) u˜m l¯f = (−ν l − is l ) uWe can use this dictionary to translate things we know on one side to the other side. In thisway we may hope to obtaining interesting new statements and uncover new structures. We willmention some developments along these lines in Section 4.2.A still more elaborate version of the dictionary, including the Kirillov-Reshetikhin modulesthat we touched upon in Section 2.2.3, can be found in §3.3 of [50].4.2 Further topicsTo conclude this chapter we outline three further aspects of the <strong>Bethe</strong>/gauge correspondence.Anisotropy/higher dimensions. Consider an N = 1 supersymmetric gauge theory in fourdimensions. By dimensional reduction to two dimensions we get a N = (2, 2) theory thathas all the nice properties which we already listed in Section 1.3.3 and which we need for the<strong>Bethe</strong>/gauge correspondence to work. However, instead of getting rid of the two dimensions atonce, we can proceed in smaller steps.We start with a superspace (isomorphic to) R 3,1|4 as described in Section 1.2.2. Instead ofreducing the spacetime we can compactify one of the spatial dimensions to a circle SR 1 withradius R, resulting in the superspace SR 1 × R2,1|4 . <strong>The</strong> component of the four-vector fieldcorresponding to the compactified dimension yields Kaluza-Klein modes with masses that areinversely proportional to the radius R. At low energy the resulting theory is effectively threedimensional.When we let R become very small, the kk modes become very massive, and canbe integrated out.We are on the right track: if we decrease the dimension by one more we get a theory thatis effectively two-dimensional and has N = (2, 2) supersymmetry at low energies, so that wecan apply the machinery of the <strong>Bethe</strong>/gauge correspondence. <strong>The</strong>re are two ways to achievethis. Firstly we can compactify one more dimension and get T 2 × R 1,1|4 ∼ = SR 1 × S1 R × ′ R1,1|4 .Alternatively, we reduce by one dimension, giving SR 1 × R1,1|4 . In either case the low energylimit gives an effective two-dimensional theory with N = (2, 2) supersymmetry. It turns out