15.07.2014 Views

Exact Results for 't Hooft Loops in Gauge Theories ... - Solvay Institutes

Exact Results for 't Hooft Loops in Gauge Theories ... - Solvay Institutes

Exact Results for 't Hooft Loops in Gauge Theories ... - Solvay Institutes

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

<strong>Exact</strong> <strong>Results</strong> <strong>for</strong> ’t <strong>Hooft</strong> <strong>Loops</strong> <strong>in</strong> <strong>Gauge</strong><br />

<strong>Theories</strong> on S 4<br />

Vasily Pestun<br />

based on arXiv:1105.2568 with Jaume Gomis, Takuya Okuda<br />

May 19, 2011<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 1/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Supersymmetry is a powerful tool to address dynamics of<br />

gauge theories<br />

• The N = 2 effective low energy Seiberg-Witten theory<br />

• Maldacena’s conjecture on the AdS/CFT correspondence<br />

<strong>for</strong> N = 4 SYM<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 2/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Loop operators are important gauge theory observables.<br />

Wilson loop operator measures effective <strong>for</strong>ces between<br />

quarks.<br />

(locally) supersymmetric Wilson loop operator was the key<br />

object <strong>in</strong> several AdS/CFT studies <strong>for</strong> N = 4 SYM<br />

• [Erickson-Semenoff-Zarembo] and [Drukker-Gross] circular<br />

1/2 BPS Wilson loops <strong>in</strong> N = 4 SYM<br />

• [Alday-Maldacena] polygonal Wilson loops and amplitudes<br />

<strong>in</strong> N = 4 SYM<br />

At λ → ∞, Wilson loop C ⊂ R 4 is dual to the m<strong>in</strong>imal surface <strong>in</strong><br />

AdS land<strong>in</strong>g on C ⊂ R 4<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 3/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Motivation<br />

Why we look on supersymmetric loops <strong>in</strong> N = 2 theories?<br />

• What from N = 4 SYM can be pushed to N = 2?<br />

• How S-duality acts? Can we compute vevs of Wilson and ’t<br />

<strong>Hooft</strong> and test S-duality precisely?<br />

• The loop observables <strong>in</strong> N = 2 theories are novel - not<br />

computable from Seiberg-Witten theory - hence we’ll learn<br />

someth<strong>in</strong>g new about N = 2 dynamics if compute them<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 4/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Wilson operators<br />

1. Def<strong>in</strong>e N = 2 theory on S 4 of radius r.<br />

2. Localize the path <strong>in</strong>tegral and get [V.P.’07]<br />

〈W R (C)〉 S 4 = 1 ∫<br />

[da]|Z Ω (ia, ɛ 1 , ɛ 2 , m, τ)| 2 tr R e 2πiar<br />

Z S 4<br />

Z Ω – Nekrasov’s partition function of N = 2 twisted theory on<br />

R 4 ɛ 1 ,ɛ 2<br />

[Losev-Moore-Nekrasov-Shatashvili’95,Nekrasov’02]<br />

ɛ 1 = ɛ 2 = 1/r<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 5/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Problem<br />

Compute exact vev of supersymmetric ’t <strong>Hooft</strong> operator <strong>in</strong><br />

N = 2 gauge theories<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 6/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Def<strong>in</strong>itions<br />

Def<strong>in</strong>e the N = 2 supersymmetry on S 4 : OSp(2|4) ⊂ SL(1|2, H)<br />

— subgroup of the N = 2 supercon<strong>for</strong>mal group with<br />

• 8 fermionic generators<br />

• Sp(4) ≃ SO(5) bosonic subgroup – isometry of S 4<br />

• SO(2) R symmetry<br />

All zero modes <strong>in</strong> the OSp(2|4) theory on S 4 are lifted. We<br />

<strong>in</strong>tegrate over all fields <strong>in</strong> the path <strong>in</strong>tegral. No moduli on S 4 .<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 7/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

To <strong>in</strong>troduce hypermultiplet masses <strong>in</strong> OSp(2|4) theory on S 4 :<br />

1 gauge the flavor symmetry<br />

2 give expectation value to Φ 0 <strong>in</strong> the flavor vector multiplet<br />

Remark on our conventions<br />

In the OSp(2|4) theory on S 4<br />

• m is real <strong>for</strong> positivity of the action<br />

• <strong>in</strong> con<strong>for</strong>mal theory m = 0<br />

• under localization m relates to Ω-background m Ω of<br />

[Nekrasov’02] as<br />

m Ω = ɛ 1 + ɛ 2<br />

2<br />

+ im<br />

The usual complex scalar field <strong>in</strong> N = 2 vector multiplet on R 4<br />

corresponds to a pair of real fields on S 4 : Φ 0 and Φ 9 .<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 8/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Supersymmetric Wilson loop<br />

∮<br />

W R (C) = tr R Pexp (Adx + iΦ 0 ds)<br />

Supersymmetric ’t <strong>Hooft</strong> loop<br />

Specify asymptotics <strong>in</strong> local R 3 normal to the loop (x = 0):<br />

F A → − ⋆ B dx<br />

2 x 2 , x → 0<br />

Φ 9 → B 1<br />

2 x , x → 0<br />

The parameter is B ∈ g (weight of ∨ G) def<strong>in</strong><strong>in</strong>g homomorphism<br />

U(1) → G.<br />

The asymptotics locally satisfies Bogomolny equations<br />

C<br />

D A Φ 9 = ⋆F A<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 9/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Method: Localization Pr<strong>in</strong>ciple<br />

Localize the path <strong>in</strong>tegral:<br />

∫<br />

Z 0 = [Dφ] e −S 0[φ]<br />

S 0 – the physical action<br />

Q – global fermionic symmetry, QS 0 = 0<br />

Q 2 = R – global bosonic symmetry<br />

De<strong>for</strong>m the action<br />

assum<strong>in</strong>g {Q 2 , V } = 0<br />

S t [φ] = S 0 [φ] + t{Q, V }<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 10/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Now consider<br />

Integrat<strong>in</strong>g by parts we get<br />

∫<br />

Z t = [Dφ] e −S 0[φ]−t{Q,V [φ]}<br />

d<br />

dt Z t = 0 =⇒ Z 0 = Z ∞<br />

At t = ∞ the one-loop approximation to Z t is exact. But<br />

Z t = Z 0 . Hence we get exact Z 0 .<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 11/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

We often take V = (Ψ, QΨ), where Ψ are the fermions of the<br />

theory.<br />

The path <strong>in</strong>tegral with action<br />

S t = S 0 + tQV<br />

and V = (Ψ, QΨ) localizes to configurations<br />

QΨ = 0<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 12/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

∫<br />

Z =<br />

X<br />

Localization summary<br />

e −S = ∑ ∫<br />

e −S| Yα Z1−loop [N Yα ]<br />

α Y α⊂X<br />

where:<br />

Q is a fermionic symmetry<br />

Q 2 = R is a bosonic symmetry<br />

QS = 0<br />

X = {Φ} is a space of fields<br />

Y α = {Φ|QΨ = 0} are components of the localization locus<br />

N Yα - normal bundle to Y α ⊂ X.<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 13/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Solution strategy<br />

We need to localize the path <strong>in</strong>tegral of N = 2 gauge theory on<br />

S 4 <strong>in</strong> the presence of ’t <strong>Hooft</strong> s<strong>in</strong>gularity.<br />

0 Def<strong>in</strong>e the action S, the t’ <strong>Hooft</strong> operator T and the<br />

fermionic operator Q compatible with S and T<br />

1 F<strong>in</strong>d the localization loci Y α (solve QΨ = 0).<br />

2 Compute exp(−S| Yα )<br />

3 Compute the determ<strong>in</strong>ant Z 1−loop [N Yα ]<br />

4 Integrate over Y α and sum over α<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 14/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Step 0. The Q.<br />

The Q is def<strong>in</strong>ed by a con<strong>for</strong>mal Kill<strong>in</strong>g sp<strong>in</strong>or ε on S 4<br />

∇ µ ε = Γ µ˜ε<br />

In the R 4 con<strong>for</strong>mal frame generic ε is<br />

ε = ˆε s + x µ Γ µˆε c<br />

We choose certa<strong>in</strong> ε s (chiral right) and ε c (chiral right) and fix<br />

Q = Q ε such that<br />

Q 2 = J + R<br />

where J is a self-dual rotation of S 4 and R is R-symmery <strong>in</strong><br />

OSp(2|4)<br />

North pole x = 0, ε is chiral right<br />

South pole x = ∞, ε is chiral left<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 15/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Step 1<br />

The equations are QΨ = 0, where<br />

QΨ = 1 2 F mnΓ mn ε − 1 2 φ aΓ aµ ∇ µ ε + iK i Γ 8i+4 ε<br />

For our Q, the equations <strong>in</strong>terpolate between<br />

• north pole: <strong>in</strong>stanton equations F + = 0 and DΦ 9 = 0<br />

• equator: Bogomolny equations DΦ 9 = ∗F <strong>in</strong> the space<br />

transversal to the S 1 orbits and <strong>in</strong>variance along S 1<br />

• south pole: anti-<strong>in</strong>stanton equations F − = 0 and DΦ 9 = 0.<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 16/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Technical details:<br />

Convenient coord<strong>in</strong>ates<br />

S 4 = S 1 × B 3<br />

where B 3 is a 3d solid ball with a boundary B 3 : ∑ 3<br />

i=1 x2 i < 1<br />

and S 1 is a circle τ<br />

The round S 4 metric is<br />

ds 2 =<br />

dx 2 i<br />

(1 + x 2 ) 2 + (1 − x2 ) 2<br />

(1 + x 2 ) 2 dτ 2<br />

B 3 -flat rescaled metric<br />

ds 2 = dx 2 i + (1 − x 2 ) 2 dτ 2<br />

The ’t <strong>Hooft</strong> loop is the S 1 fiber at x i = 0.<br />

The North pole, the first fixed po<strong>in</strong>t of Q 2 is x = (0, 0, 1)<br />

he South pole, the second fixed of Q 2 is x = (0, 0, −1)<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 17/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

In coord<strong>in</strong>ates (x i , τ) the equations <strong>for</strong> vector multiplet are<br />

[<br />

F 1τ + D 1 , i (<br />

1 + |⃗x|<br />

2 ) ]<br />

Φ 0 − x 3 Φ 9 + x 1 F 12 = 0<br />

2<br />

F 2τ +<br />

[<br />

F 3τ + D 3 , i 2<br />

[<br />

[<br />

D 2 , i (<br />

1 + |⃗x|<br />

2 ) ]<br />

Φ 0 − x 3 Φ 9 − x 2 F 21 = 0<br />

2<br />

(<br />

1 + |⃗x|<br />

2 ) Φ 0 − x 3 Φ 9<br />

]<br />

+ x 1 F 32 − x 2 F 31 = 0<br />

(<br />

1 + |⃗x|<br />

2 ) Φ 0 − x 3 Φ 9<br />

]<br />

+ x 1 F τ2 − x 2 F τ1 = 0<br />

D τ , i 2<br />

[<br />

[Φ 9 , D τ ] + Φ 9 , i (<br />

1 + |⃗x|<br />

2 ) ]<br />

Φ 0 + x 1 [Φ 9 , D 2 ] − x 2 [Φ 9 , D 1 ] = 0<br />

2<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 18/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

−(1+x 2 1−x 2 2−x 2 3)[D 1 Φ 9 ]−2x 1 x 2 [D 2 Φ 9 ]−2x 1 x 3 [D 3 Φ 9 ]−2x 2 [Dˆ4 Φ 9]<br />

− 2x 1 Φ 9 − 2x 1 x 3 F 12 + 2x 1 x 2 F 13 + (1 − x 2 1 + x 2 2 + x 2 3)F 23 − 2x 3 F 1ˆ4<br />

+ i(1 + |⃗x| 2 )K 1 + 2x 1 F 3ˆ4 = 0<br />

−(1−x 2 1+x 2 2−x 2 3)[D 2 Φ 9 ]−2x 1 x 2 [D 1 Φ 9 ]−2x 2 x 3 [D 3 Φ 9 ]+2x 1 [Dˆ4 Φ 9]<br />

− 2x 2 Φ 9 − 2x 2 x 3 F 12 − 2x 1 x 2 F 23 − (2 2 + x 2 1 − x 2 2 + x 2 3)F 13 − 2x 3 F 2ˆ4<br />

+ i(1 + |⃗x| 2 )K 2 + 2x 2 F 3ˆ4 = 0<br />

2x 1 x 3 [D 1 Φ 9 ] + 2x 2 x 3 [D 2 Φ 9 ] − (1 + x 2 1 + x 2 2 − x 2 3)[D 3 Φ 9 ] + 2x 3 Φ 9<br />

+ 4irΦ 0 + (1 − x 2 1 − x 2 2 + x 2 3)F 12 + 2x 1 x 3 F 23 − 2x 2 x 3 F 13<br />

− 2x 2 F 2ˆ4 − 2x 3F 3ˆ4 − 2x 1F 1ˆ4 + i ( 1 + |⃗x| 2) K 3 = 0<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 19/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

More compact <strong>for</strong>m of equations<br />

Let<br />

a i = 1 2 (D iΦ 9 + 1 2 ɛ ijkF jk ), b i = 1 2 (D iΦ 9 − 1 2 ɛ ijkF jk ).<br />

Then the 3d equations on B 3 are<br />

b i − δ i x i Φ 9 + δ i x 2 T ij a j = 0<br />

where<br />

δ 1 = δ 2 = −δ 3 = 1<br />

T ij = δ ij − 2x ix j<br />

x 2<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 20/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Smooth solutions<br />

The only topologically trivial solution and smooth away from<br />

s<strong>in</strong>gularity is<br />

x i<br />

F jk = − B 2 ɛ ijk<br />

|⃗x| 3 , F iˆ4 = −ig2 θ B x i<br />

16π 2 |⃗x| 3 ,<br />

Φ 9 =<br />

B<br />

2|⃗x| , Φ 0 = −g 2 θ B 1<br />

16π 2 |⃗x| + a<br />

,<br />

1 + |⃗x|2<br />

4r 2<br />

a/r<br />

K 3 = −(<br />

) 2<br />

,<br />

1 + |⃗x|2<br />

4r 2<br />

The moduli space of solutions is parametrized by a ∈ g.<br />

The path <strong>in</strong>tegral localizes to<br />

Y 0 = g<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 21/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Step 2. The action S| Y0<br />

Evaluat<strong>in</strong>g S 0 at supersymmetric configurations we get<br />

(<br />

S cl [a] = − 8π2 2π<br />

2<br />

g 2 Tr a2 +<br />

g 2 + g2 θ 2 )<br />

32π 2 Tr B 2<br />

In terms of<br />

we can rewrite<br />

τ = θ<br />

2π + 4πi<br />

g 2<br />

where<br />

S cl [a] = −πiτ Tr â(N) 2 + πi¯τ Tr â(S) 2<br />

â(N) = iΦ 0 (N) − Φ 9 (N)<br />

â(S) = iΦ 0 (S) + Φ 9 (S)<br />

are the parameters of the gauge trans<strong>for</strong>mation generated by<br />

Q 2 at fixed po<strong>in</strong>ts N and S<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 22/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Step 3. The one-loop determ<strong>in</strong>ant<br />

Denot<strong>in</strong>g the fields of even and odd statistics with a sub<strong>in</strong>dex e<br />

and o respectively, the Q multiplets are<br />

ˆQ · ϕ e,o = ˆϕ o,e<br />

ˆQ · ˆϕ o,e = R · ϕ e,o .<br />

and then<br />

ˆQ 2 · ϕ e,o = R · ϕ e,o ,<br />

We can show that the one-loop determ<strong>in</strong>ant is<br />

det CokerD vmR| o<br />

det KerD vmR| e<br />

· det CokerD hmR| o<br />

det KerD hmR| e<br />

.<br />

where D is a certa<strong>in</strong> (tranversally elliptic) differential operator<br />

def<strong>in</strong>ed from our tQV term.<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 23/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

To f<strong>in</strong>d the ratio of determ<strong>in</strong>ants (equivariant Euler character)<br />

we consider the <strong>in</strong>dex of operator D (equivariant Chern<br />

character)<br />

<strong>in</strong>d D = tr KerD e R − tr CokerD e R<br />

read the weights of R and comb<strong>in</strong>e them <strong>in</strong>to determ<strong>in</strong>ant<br />

us<strong>in</strong>g the rule<br />

∑<br />

c j e w j(ε 1 ,ε 2 ,â, ˆm f ) → ∏ w j (ε 1 , ε 2 , â, ˆm f ) c j<br />

j<br />

j<br />

Recall that R = J + R + [Φ, ·]<br />

J is the SD spatial rotation<br />

R is the R-symmetry rotation<br />

[Φ, ·] is the gauge trans<strong>for</strong>mation.<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 24/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Excision property of the <strong>in</strong>dex/Atiya-S<strong>in</strong>ger theory<br />

To f<strong>in</strong>d <strong>in</strong>dex <strong>for</strong> transversally elliptic operator D on a manifold<br />

M under equivariant U(1) action R we can cut M <strong>in</strong>to three<br />

pieces<br />

• neighborhood of the north pole<br />

• neighborhood of the equator<br />

• neighborhood of the south pole<br />

and apply Atiyah-S<strong>in</strong>ger fixed po<strong>in</strong>t <strong>for</strong>mula <strong>for</strong> the <strong>in</strong>dex<br />

<strong>in</strong>d D(R) = ∑ p∈F<br />

tr E0 (p) R − tr E1 (p) R<br />

.<br />

det T Mp (1 − R)<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 25/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

North (South) pole<br />

Near the North pole the operator D, def<strong>in</strong><strong>in</strong>g the equations, is<br />

the self-dual operator<br />

The <strong>in</strong>dex is then<br />

D SD : Ω 1 d∗ ⊕d +<br />

−→ Ω 0 ⊕ Ω 2+<br />

<strong>in</strong>d(D SD,C )(t 1 , t 2 ) = (t 1t 2 + t −1<br />

1 t−1 2 + 2) − (t 1 + t −1<br />

1 + t 2 + t −1<br />

2 )<br />

(1 − t 1 )(1 − t −1<br />

1 )(1 − t 2)(1 − t −1<br />

2 )<br />

1 + t 1 t 2<br />

=<br />

(1 − t 1 )(1 − t 2 ) . (5.1)<br />

<strong>for</strong> the U(1) action z 1 → t 1 z 1 , z 2 → t 2 z 2 .<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 26/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

When we tensor product sections with the adjo<strong>in</strong>t<br />

representation of G the <strong>in</strong>dex becomes (at t i = e iε i<br />

)<br />

<strong>in</strong>d(D)(ε 1 , ε 2 , â) =<br />

To read the R weights we expand<br />

1 + e iε 1<br />

e iε 2<br />

(1 − e iε 1 )(1 − e iε 2)<br />

= ∑<br />

(1 + e iε 1+iε 2<br />

) ∑<br />

2(1 − e iε 1 )(1 − e iε 2)<br />

n 1 ,n 2 ≥0<br />

w∈adj<br />

e iw·â . (5.2)<br />

(1 + e iε 1<br />

e iε 2<br />

)e <strong>in</strong> 1ε 1<br />

e <strong>in</strong> 2ε 2<br />

(5.3)<br />

There<strong>for</strong>e, the one-loop contribution from the pole is<br />

∏<br />

[n 1 ε 1 + n 2 ε 2 + α · â] 1/2 [(n 1 + 1)ε 1 + (n 2 + 1)ε 2 + α · â] 1/2 .<br />

n 1 ,n 2 ≥0<br />

(5.4)<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 27/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Barnes G-function<br />

The double product can be regularized<br />

G(1 + z) = (2π) z/2 e −((1+γz2 )+z)/2<br />

∞∏<br />

n=1<br />

(<br />

1 + z ) n<br />

e<br />

−z+ z2<br />

2n . (5.5)<br />

n<br />

and then the one-loop contribution from each pole is<br />

Z 1-loop,pole (â) = ∏ ( ) (<br />

α · â<br />

G 1/2 G 1/2 2 + α · â )<br />

ε<br />

ε<br />

α<br />

and recall that the gauge parameter â at the poles is (at θ = 0)<br />

â(N) = ia − B 2r<br />

â(S) = ia + B 2r ,<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 28/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Contribution to Z 1−loop from equatior<br />

Skipp<strong>in</strong>g derivation, the answer (<strong>for</strong> vectormultiplet)<br />

Z 1-loop,eq (â, B) = ∏ α>0<br />

[ ( (â<br />

s<strong>in</strong> πα ·<br />

ε + B ))] −α·B<br />

(5.6)<br />

2<br />

Schematically, this comes from extra modes that appear near<br />

the equator <strong>in</strong> the s<strong>in</strong>gular monopole background.<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 29/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Step 5. Integrate over Y 0<br />

Comb<strong>in</strong>ation of exp(−S| Y0 ) and Z 1−loop factorizes nicely <strong>in</strong>to<br />

〈<br />

ZT (B)<br />

〉<br />

=<br />

∫<br />

where<br />

and<br />

Z 1-loop,pole =<br />

∫<br />

da Z north · Z south · Z equator =<br />

da |Z north | 2 · Z equator<br />

Z north =Z cl (â(N), q) Z 1-loop,pole (â(N), im f )<br />

Z south =Z cl (â(S), ¯q) Z 1-loop,pole (â(S), im f )<br />

Z equator =Z 1-loop,eq (â(E), im f , B) ,<br />

[ ]<br />

1<br />

Z cl (â, q) = exp 2πiτ Tr â 2 .<br />

2ε 1 ε 2<br />

∏<br />

∏ NF<br />

∏ [<br />

f=1 w∈R<br />

G<br />

[ (<br />

G<br />

α·â<br />

) (<br />

ε G 2 +<br />

α·â<br />

ε<br />

(<br />

)<br />

1 + w·â<br />

ε<br />

− ˆm f<br />

ε<br />

G<br />

α<br />

)] 1/2<br />

(<br />

1 − w·â<br />

ε<br />

)]<br />

+ ˆm 1/2<br />

f<br />

ε<br />

and Z 1-loop,eq Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 30/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Other components Y α : <strong>in</strong>stanton corrections<br />

In the limit x → 0, up to the terms O(x 2 ) our Q complex <strong>in</strong> the<br />

neighborhood of the north pole is exactly as the one of the<br />

gauge theory on R 4 ε,ε [Nekrasov’02] . Hence the other<br />

components of the localization locus Y α are the po<strong>in</strong>t <strong>in</strong>stantons<br />

F + = 0 sitt<strong>in</strong>g at the north pole as <strong>in</strong> [Nekrasov’02] .<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 31/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

Result<br />

Includ<strong>in</strong>g the po<strong>in</strong>t <strong>in</strong>stanton contributions at the north and<br />

south poles, we f<strong>in</strong>ally get the result<br />

〈<br />

ZT (B)<br />

〉<br />

=<br />

∫<br />

da<br />

∣ Z north(ia − B ∣ ∣∣∣<br />

2<br />

2 ) · Z equator (ia)<br />

where Z north is Nekrasov’s partition function <strong>in</strong>clud<strong>in</strong>g classical,<br />

(slightly modified) one-loop and the <strong>in</strong>stanton factors.<br />

Our gauge theory result agrees with the computation <strong>in</strong> the<br />

Liouville/Toda theories by [Alday-Gaiotto-Tajikawa] conjecture<br />

per<strong>for</strong>med by [Drukker,Gomis,Okuda,Teschner’09] and [Alday,<br />

Gaiotto, Gukov, Tachikawa, Verl<strong>in</strong>de’09] and [Gomis,Floch’10] .<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 32/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

At the m<strong>in</strong>imal B the result is exact. At higher B there are<br />

further corrections com<strong>in</strong>g from the screen<strong>in</strong>g of the ’t <strong>Hooft</strong><br />

loop by the monopoles condens<strong>in</strong>g on the loop.<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 33/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

S-duality<br />

F<strong>in</strong>ally, let us compare our <strong>for</strong>mulae <strong>for</strong> Wilson and ’t <strong>Hooft</strong><br />

loops and<br />

∫<br />

〈 〉<br />

ZW (R) τ = [da]|Z north (ia, τ)| ∑ 2 e 2πiaw<br />

w<br />

〈<br />

ZT (B)<br />

〉<br />

τ = ∫<br />

da<br />

∣ Z north(ia − B ∣ ∣∣∣<br />

2<br />

2 , τ) · Z equator (ia)<br />

Notice:<br />

Wilson loop of weight w <strong>in</strong>serts operator exp(2πiwa)<br />

t’ <strong>Hooft</strong> loop of coweight B <strong>in</strong>serts shift operator exp(B ∂<br />

∂a )<br />

Hence, the latter <strong>for</strong>mula (magnetic) at τ ∨ = − 1 τ<br />

is Fourier<br />

trans<strong>for</strong>m of the <strong>for</strong>mer (electric).<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 34/35


Introduction Problem Def<strong>in</strong>itions Method Solution Result<br />

The Nekrasov’s functions Z north (â) are too complicated to<br />

explicitly <strong>in</strong>tegrate analytically over â, but the <strong>in</strong>tegrand is well<br />

def<strong>in</strong>ed as a quickly convergent series and we were able to<br />

per<strong>for</strong>m numerical check of the S-duality <strong>in</strong>variance.<br />

Vasily Pestun Localization <strong>for</strong> ’t <strong>Hooft</strong> operators on S 4 35/35

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!