(Elliptic) multiple zeta values in open superstring amplitudes

17
(Elliptic) multiple zeta values in open superstring amplitudes Johannes Broedel Humboldt University Berlin based on joint work with Carlos Mafra, Nils Matthes and Oliver Schlotterer arXiv:1412.5535, arXiv:1507.02254 Selected Topics in Theoretical High Energy Physics September 21 st , 2015

Transcript of (Elliptic) multiple zeta values in open superstring amplitudes

Page 1: (Elliptic) multiple zeta values in open superstring amplitudes

(Elliptic) multiple zeta valuesin open superstring amplitudes

Johannes BroedelHumboldt University Berlin

based on joint work with Carlos Mafra, Nils Matthes and Oliver SchlottererarXiv:1412.5535, arXiv:1507.02254

Selected Topics in Theoretical High Energy Physicstbilisi, sakartvelo, September 21st, 2015

Page 2: (Elliptic) multiple zeta values in open superstring amplitudes

IntroductionGoal: scattering amplitudes/cross sections in a field or string theory• standard method: Feynman/worldsheet graphs, useful and cumbersome• alternative idea: add symmetry→ obtain a more symmetric/constrained theory

→ learn about structure→ remove symmetry→ what remains• typical results: new language: special functions for particular theory

(e.g. spinor-helicity for massless theories)recursion relations: relate N -point to (N − 1)-point

• best scenario: avoid Feynman calculations completely/S-matrix approach

This talk:Open string theory as a simple (and very symmetric) testing ground• tree-level: polylogarithms (language) and Drinfeld associator (recursion)• one-loop elliptic iterated integrals (language) and elliptic multiple zeta values• Outlook: link to number theory/algebraic geometry

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 1/16

Page 3: (Elliptic) multiple zeta values in open superstring amplitudes

State of the art: open string theoryTree-level:• Calculation of all amplitudes based on [Broedel, Schlotterer

Stieberger ]multiple polylogarithms and multiple zeta values [Goncharov][Brown][Zagier]and the algebraic structure of string corrections at tree-level. [Schlotterer

Stieberger ]• Drinfeld associator avoids necessity of solving integrals at all.[ Broedel, Schlotterer

Stieberger, Terasoma]Complete calculation boiled down to recursive application of linear algebra.

Loop-level:• calculation based on elliptic iterated integrals and elliptic multiple zeta

values [ Broedel, MafraMatthes, Schlotterer][ Broedel

Matthes, Schlotterer]• no analogue of the Drinfeld method so far: integrals can not be replaced com-

pletely yet . . .

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 2/16

Page 4: (Elliptic) multiple zeta values in open superstring amplitudes

Why open string theory?Iterated integrals are essential in calculations in field and string theory:• same building blocks

field theory: multiple polylogarithms at loop level. Divergences appear.string theory: multiple polylogarithms at tree level. No divergences.• field theory: elliptic iterated integrals make an appearance in particular Feynman

diagrams. [Adams, BognerWeinzierl ][Caron-Huot

Larsen ]string theory: one-loop amplitudes are natural for elliptic iterated integrals.• field theory results can be obtained from open string theory in the low-energy

limit.

After all, string theory is a heavily constrained theorywith an amazing degree of symmetry⇒ should produce simple answers.

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 3/16

Page 5: (Elliptic) multiple zeta values in open superstring amplitudes

OutlineTree-level

0

z1 z2 zN−2

1

zN−1

zN =∞

· · ·N−2∏i=2

∫ zi+1

0dzi

multiple polylogarithmsG(a1,a2, . . . , an; z)

=∫ z

0dt 1t− a1

G(a2, . . . , an; t)

partial fractionmultiple zeta values ζDrinfeld method - no integrals

One-loop

z1 z2 zN−1 zN

t

1· · ·∫ 1

0dzN

N−1∏i=1

∫ zi+1

0dzi δ(z1)

elliptic iterated integralsΓ ( n1 n2 ... nr

a1 a2 ... ar ; z)

=∫ z

0dt f (n1)(t− a1) Γ ( n2 ... nr

a2 ... ar ; t)

Fay-identitieselliptic multiple zeta values ωelliptic (KZB) associator?

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 4/16

Page 6: (Elliptic) multiple zeta values in open superstring amplitudes

Tree-level, basicsN -point tree-level open-string amplitude: [Veneziano]. . . [Mafra, Schlotterer

Stieberger ]

Aopenstring = F .AYM

• dependence on external states in AYM• F = F (sij), sij = α′(ki + kj)2

• coefficients are multiple zeta values (MZVs) 0

z1 z2 zN−2

1

zN−1

zN =∞

· · ·

F 1,2,...,N =N−2∏i=2

∫ zi+1

0dzi

N−1∏i<j

|zij |sij

s12

z12

(s13

z13+ s23

z23

). . .

(s1,N−2

z1,N−2+ . . .+ sN−3,N−2

zN−3,N−2

)

N−2∏i=2

∫ zi+1

0

dzi

zi − ai

N−1∏i<j

|zij |sij︸ ︷︷ ︸expand. . .

⇒N−1∏i<j

∞∑nij=0

(sij)nij(ln |zij |)nij

nij !︸ ︷︷ ︸multiple polylogsMultiple polylogarithms

G(a1, a2, . . . , an; z) =∫ z

0

dtt− a1

G(a2, . . . , an; t), G(; z) = 1, G(~a; 0) = G(; 0) = 0

G(0, 0, . . . , 0︸ ︷︷ ︸w

; z) = 1w! (ln z)w G(1, 1 . . . , 1︸ ︷︷ ︸

w

; z) = 1w! lnw(1− z)

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 5/16

Page 7: (Elliptic) multiple zeta values in open superstring amplitudes

N−2∏i=2

∫ zi+1

0

dzizi − ai

N−1∏i<j

∞∑nij=0

(sij)nijG(0, 1, zl, zk)︸ ︷︷ ︸integrate step by step. . .

N−1∏i<j

∞∑nij=0

(sij)nijG(0, 1, 1)

︸ ︷︷ ︸rewrite polylogs as multiple ζ’s

ζn1,...,nr =∑

0<k1<···<kr

1kn1

1 · · · knrr

= (−1)rG(0, 0, . . . , 0, 1︸ ︷︷ ︸nr

, . . . , 0, 0, . . . , 0, 1︸ ︷︷ ︸n1

; 1) = ζ(w)

5-point-example:

F (23) = 1− ζ2(s12s23 + s12s24 + s12s34 + s13s34 + s23s34)+ ζ3(s2

12s23 + s12s223 + s2

12s24 + 2s12s23s24 + s12s224 + · · · ) + · · ·

+ ζ3,5(. . .) + · · ·

Pretty cumbersome - isn’t there an easier way to obtain the result?

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 6/16

Page 8: (Elliptic) multiple zeta values in open superstring amplitudes

Drinfeld-method [ Broedel, SchlottererStieberger, Terasoma]

Knizhnik-Zamolodchikov equation [ KnizhnikZamolodchikov]

dF(z0)dz0

=(e0z0− e1z0 − 1

)F(z0) .

• z0 ∈ C\0, 1, Lie-algebra generators e0, e1 0

z1 z2 zN−2 z0

1

zN−1

zN =∞

· · ·Regularized boundary values

C0 ≡ limz0→0

z−e00 F(z0)

(N− 1)-point

0

z1z2

z0 z0

1

zN−1

zN =∞

C1 ≡ limz0→1

(1− z0)e1F(z0)

N-point

0

z1 z2 zN−2 z0 z0

1

zN−1

zN =∞

· · ·

are related by the Drinfeld associator Φ: [Drinfeld][ LeMurakami][Furusho][Drummond

Ragoucy ]

C1 = Φ(e0, e1)C0, Φ(e0, e1) =∑

w∈0,1w[e0, e1]ζ(w)

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 7/16

Page 9: (Elliptic) multiple zeta values in open superstring amplitudes

Collect tree-level resultsTree-level

0

z1 z2 zN−2

1

zN−1

zN =∞

· · ·N−2∏i=2

∫ zi+1

0dzi

multiple polylogarithmsG(a1,a2, . . . , an; z)

=∫ z

0dt 1t− a1

G(a2, . . . , an; t)

partial fractionmultiple zeta values ζDrinfeld method - no integrals

One-loop

z1 z2 zN−1 zN

t

1· · ·∫ 1

0dzN

N−1∏i=1

∫ zi+1

0dzi δ(z1)

elliptic iterated integralsΓ ( n1 n2 ... nr

a1 a2 ... ar ; z)

=∫ z

0dt f (n1)(t− a1) Γ ( n2 ... nr

a2 ... ar ; t)

Fay-identitieselliptic multiple zeta values ωelliptic (KZB) associator?

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 8/16

Page 10: (Elliptic) multiple zeta values in open superstring amplitudes

z1 z2 zN−1 zN

t

1· · ·

One-loop open string• topologies: all genus-one worldsheets

with boundaries.• here: cylinder with insertions on

one boundary only: Im(z) = 0• one imaginary parameter: τ

General form of the integral:

A1-loopstring (1, 2, 3, 4) = s12s23A

treeYM(1, 2, 3, 4)

∫ ∞0

dτ I4pt(1, 2, 3, 4)(τ)

I4pt(1, 2, 3, 4)(τ) ≡∫ 1

0dz4

∫ z4

0dz3

∫ z3

0dz2

∫ z2

0dz1 δ(z1)

4∏j<k

[χjk(τ)

]sjk

︸ ︷︷ ︸Koba-Nielsen

Green’s function of the free boson on a genus-one surface with modulus τ :

lnχij(τ)∣∣∣zij=τxij

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 9/16

Page 11: (Elliptic) multiple zeta values in open superstring amplitudes

Compare tree-level

N−2∏i=2

∫ zi+1

0

dzizi − ai

N−1∏i<j

∞∑nij=0

1nij !

(sij)nij (ln |zij |)nij︸ ︷︷ ︸multiple polylogs

with one-loop situation:

∫ 1

0dzN

N−1∏i=1

∫ zi+1

0dzi δ(z1)

N∏i<j

∞∑nij=0

1nij !

(sij)nij (lnχij(τ))nij︸ ︷︷ ︸???

Suitable (iterated) object:

lnχij(τ) =∫ zi

zj

dw f (1)(w − zj , τ)

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 10/16

Page 12: (Elliptic) multiple zeta values in open superstring amplitudes

Natural weights for differentials on an elliptic curve: [Enriquez][BrownLevin ]

f (n)(z, τ) = f (n)(z + 1, τ) and f (n)(z, τ) = f (n)(z + τ, τ) .

Explicitly: (simplification in our situation because Im(z) = 0)

f (0)(z, τ) ≡ 1 f (1)(z, τ) ≡ ∂ ln θ1(z, τ) + 2πi ImzImτ

f (2)(z, τ) ≡ 12[(∂ ln θ1(z, τ) + 2πi ImzImτ

)2+ ∂2 ln θ1(z, τ)− 1

3θ′′′1 (0, τ)θ′1(0, τ)

]Parity: f (n)(−z, τ) = (−1)nf (n)(z, τ)

Relation to Eisenstein–Kronecker-series: [Kronecker][BrownLevin ]

F (z, α, τ) ≡ θ′1(0, τ)θ1(z + α, τ)θ1(z, τ)θ1(α, τ) ,

αΩ(z, α, τ) ≡ α exp(

2πiα Im(z)Im(τ)

)F (z, α, τ) =

∞∑n=0

f (n)(z, τ)αn

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 11/16

Page 13: (Elliptic) multiple zeta values in open superstring amplitudes

Elliptic iterated integrals (suppress τ -dependence from here. . . )

Γ ( n1 n2 ... nra1 a2 ... ar ; z) ≡

∫ z

0dw f (n1)(w − a1) Γ ( n2 ... nr

a2 ... ar ;w)

⇒ can rewrite any integral∫

1234 . . . into an elliptic iterated integral .

Products of differential weights (tree-level) ⇒ partial fraction:∫ z

0dw 1

w − a1

1w − a2

· · · ⇒ 1(w−a1)(w − a2) = 1

(w−a1)(a1−a2)+ 1(w−a2)(a2−a1)

Products of differential weights (one-loop) ⇒ Fay identities∫ z

0dw f (n1)(w − x)f (n2)(w) · · · ⇒ f (1)(w−x)f (1)(w) =f (1)(w−x)f (1)(x)−f (1)(w)f (1)(x)

+ f (2)(w) + f (2)(x) + f (2)(w − x)

The Fay identity is a form of the trisecant equation for Eisenstein–Kronecker series:

F (z1, α1)F (z2, α2) = F (z1, α1 + α2)F (z2 − z1, α2)+ F (z2, α1 + α2)F (z1 − z2, α1)

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 12/16

Page 14: (Elliptic) multiple zeta values in open superstring amplitudes

Elliptic multiple zeta values (eMZV’s)

ω(n1, n2, . . . , nr) ≡∫

0≤zi≤zi+1≤1

f (n1)(z1)dz1 f(n2)(z2)dz2 . . . f

(nr)(zr)dzr

= Γ(nr, . . . , n2, n1; 1) = Γ ( nr nr−1 ... n10 0 ... 0 ; 1)

Four-point result

I4pt(1, 2, 3, 4)(τ) =ω(0, 0, 0) − 2ω(0, 1, 0, 0) (s12 + s23)+ 2ω(0, 1, 1, 0, 0)

(s2

12 + s223)− 2ω(0, 1, 0, 1, 0) s12s23

+ β5 (s312 + 2s2

12s23 + 2s12s223 + s3

23)+ β2,3 s12s23(s12 + s23) + O(α′4)

with

β5 = 43[ω(0, 0, 1, 0, 0, 2) + ω(0, 1, 1, 0, 1, 0)− ω(2, 0, 1, 0, 0, 0)− ζ2 ω(0, 1, 0, 0)

]β2,3 = 1

3 ω(0, 0, 1, 0, 2, 0)− 32 ω(0, 1, 0, 0, 0, 2)− 1

2 ω(0, 1, 1, 1, 0, 0)

− 2 ω(2, 0, 1, 0, 0, 0)− 43 ω(0, 0, 1, 0, 0, 2)− 10

3 ζ2 ω(0, 1, 0, 0)

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 13/16

Page 15: (Elliptic) multiple zeta values in open superstring amplitudes

ComparisonTree-level

0

z1 z2 zN−2

1

zN−1

zN =∞

· · ·N−2∏i=2

∫ zi+1

0dzi

multiple polylogarithmsG(a1,a2, . . . , an; z)

=∫ z

0dt 1t− a1

G(a2, . . . , an; t)

partial fractionmultiple zeta values ζDrinfeld method - no integrals

One-loop

z1 z2 zN−1 zN

t

1· · ·∫ 1

0dzN

N−1∏i=1

∫ zi+1

0dzi δ(z1)

elliptic iterated integralsΓ ( n1 n2 ... nr

a1 a2 ... ar ; z)

=∫ z

0dt f (n1)(t− a1) Γ ( n2 ... nr

a2 ... ar ; t)

Fay-identitieselliptic multiple zeta values ωelliptic associator? [Knizhnik, Bernard

Zamolodchikov ]Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 14/16

Page 16: (Elliptic) multiple zeta values in open superstring amplitudes

Summary• eMZVs are natural language for one-loop amplitudes in open string theory• full amplitude only after τ -integration and consideration of other topologies

eMZVs might not be the only ingredient: Euler sums?• open-string result does not contain divergent eMZVs

xWhat else?• eMZVs: can be represented as iterated Eisenstein integrals• iterated Eisenstein integrals nicely related to special derivation algebra, [Pollack]

available cusp forms on the elliptic curve ⇔ number of ”basis” eMZVs [Brown]• number of ”basis” eMZVs + canonical choice known [Hain][Broedel, Matthes

Schlotterer ]• using our formalism, one can derive new relations in the derivation algebra u,

which match the known pattern of cusp forms• numerous relations for eMZVs: https://tools.aei.mpg.de/emzv

xGoal• closed/recursive form of the integrand for the one-loop open-string amplitude

in terms of iterated Eisenstein integrals (analogue of Drinfeld-method)• relation to functions ELi occurring in [Adams, Bogner

Weinzierl ]Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 15/16

Page 17: (Elliptic) multiple zeta values in open superstring amplitudes

Thanks!

Johannes Broedel: (Elliptic) multiple zeta values in open superstring amplitudes 16/16