Darren Forde SPh T Saclay arxiv hepth0507292 arxiv

Darren Forde (SPh. T Saclay) arxiv: hep-th/0507292, arxiv: hep-ph/0509358, arxiv: hep-ph/0604195 & arxiv: hep-ph/0607014. with C. Berger, Z. Bern, L. Dixon & D. Kosower

Overview

The unitarity bootstrap technique Focus on this Cut construction techniques [Bern, Dixon, Dunbar, Kosower] [Britto, Cachazo, Feng] [Britto, Buchbinder, Cachazo, Feng] [Britto, Feng, Mastrolia] [Bern, Bjerrum-Bohr, Dunbar, Ita] [Bedford, Brandhuber, Spence, Travaglini] Unitarity cutting techniques, Generalized unitarity, MHV techniques, Direct integration via residues, Recursion relation of integral coefficients.

Rational part of the loop amplitude Cut constructable part Calculating the rational terms z Shift the momentum of the amplitude A=R+C by a complex parameter z, Contribution from circle at infinity Poles Branch cuts From on-shell From know From on-shell recurrence relation cut pieces, C. Integrate over a circle at infinity

Spurious singularities Cut-constructible piece, Cn(z), contains spurious singularities, e. g. physical pole when r=s 1/s 2→ 0. unphysical pole when r=s 1/s 2→ 1. z Cancel against singularities in Rn(z). Sum over all poles to calculate Rn(z), Subtract spurious singularities from Rn(z). Sum over physical poles only. Overlap terms

On-shell recurrence relations [Britto, Cachazo, Feng], [Britto, Cachazo, Feng, Witten], [Bern, Dixon, Kosower] Calculate the sum of residues from factorisation properties of amplitudes, Rn A<n i and j are the shifted legs Final result independent of this choice. For 1 -loop one of AL or AR is the rational part of a loop amplitude.

Do we understand the factorization? Requires the complex factorization properties of the amplitude, Different from real factorization properties at 1 -loop, We see the appearance of “double” poles and “unreal” poles, Factorisation into channels containing 1 -loop A 3(++; +), A 3(++; -), A 3(--; -) and A 3(--; +) vertices is in general unknown. Solution? Avoid shifts that give factorizations in these No, there’s a problem! Generally such shifts get a channels? contribution from the circle at infinity, [Inf An].

On-shell recurrence for [ Inf An] General form of a shifted amplitude in z Given by C(0) An can be constructed from a shift Given by a recurrence relation of a second shift. Now have all the terms to construct the rational piece.

General strategy

1 -loop example Calculate the A 1 -loop(1 -, 2 -, 3 -, 4+, 5+, 6+) gluon amplitude, [Berger, Bern, Dixon, DF, Kosower] Sum other helicities Complete amplitude, all pieces known [Bern, Dixon, Kosower] [Berger, Bern, Dixon, DF, Kosower] [Xiao, Yang, Zhu] [Bedford, Brandhuber, Spence, Travaglini] [Britto, Feng, Mastrolia] [Bern, Bjerrum-Bohr, Dunbar, Ita]. 1. Cut pieces previously derived, Cut completion, CR(z)

1 -loop example – recursive terms 9 recursive diagrams, the first 5 are zero, e. g. The remaining 4 are non-zero, e. g. + Similar results for other 3 diagrams

1 -loop example – The last two steps Auxiliary shift in legs 3 and 4 an on-shell recursion with one contributing term for [Inf An](z), 21 - Inf A 6 6+ 5+ 1 - 2 - 3 - 6+ Inf A 5 A 3 4+ Finally there are 3 overlap terms, e. g. 5+

1 -loop example – the final result Summing up all 4 different pieces gives us the complete rational part, Checks Has the correct factorisations and satisfies both the flip symmetries A 6(1 -, 2 -, 3 -, 4+, 5+, 6+)=A 6(3 -, 2 -, 1 -, 6+, 5+, 4+), A 6(1 -, 2 -, 3 -, 4+, 5+, 6+)=A 6(6+, 5+, 4+, 3 -, 2 -, 1 -). Matches a known numerical result. [Ellis, Giele, Zanderighi]

Solving the recurrence. Often have a recurrence with an unknown amplitude, with the same number of negative legs, on both sides. For example: An(1 -, 2 -, 3+…, n+) after shifting legs 1 and 2, Two types of contribution Known (gluonic) - Rsum + + A + + R A 3 - + = - R 3 -point Vertex - - R + + - + + All the pieces with fewer negative legs things we known. pieces with fewer positive legs same no. negative legs unknown. To solve this insert R into itself.

“Unwinding” and “Rewinding” Inserting A into itself until we reach Rsum + Rsum A 3 + Rsum …+ A 3 Both pieces known 2 +… A 3 Rsum A 3 j Recurrence solved Result is the sum of the product of known terms,

Solutions to all-multiplicity amplitudes The all-multiplicity 2 minus adjacent 1 -loop QCD amplitude [DF, Kosower] Technique applies generally, used to solve Non-adjacent 2 -minus amplitude, [Berger, Bern, Dixon, DF, Kosower] 3 -minus split helicity amplitude, [Berger, Bern, Dixon, DF, Kosower] Also tree level massive scalar amplitudes. [DF, Kosower] Small growth in complexity as number of legs increases.

Conclusions

On-shell recurrence for [ Inf An] Derive a recurrence relation for [Inf An] using a second shift. Pick a shift of legs l and m such that we have no [Inf An], this can have unknown factorisations, Perform the original shift in z on the legs i and j. If the unknown factorisations in Rn vanish when z→∞ then, From on-shell recursion. Extracted from known cut pieces. Have all the terms to construct the rational piece.

Can we avoid an [ Inf An](z) term? Spinors with complex momentum do not satisfy Spinor products are independent. Some 3 -point vertices no longer vanish, Momentum conservation p 1. p 2=p 2. p 3=p 1. p 3=0. For real momentum For complex momentum The 3 -point vertex can survive, e. g. for gluons p 1 p 3 p 2 =0

Unknown factorisations Complex spinors more complicated structure in the factorisation e. g. Complex factorisation ≠ real factorisation. Generally 1 -loop complex factorisation into 1 -loop A 3(++; +), A 3(++; -), A 3(--; -) and A 3(--; +) channels is unknown. + ± + Choose shifts to avoid these channels Generally leaves shifts with an [Inf An](z).

Choosing the shifts Can we always choose two such shifts? For gluonic 1 -loop amplitudes we have 1. a class of shifts with no unknown factorisations. 2. a class of shifts with no “boundary” (Inf) terms. For example shifts for gluonic amplitudes