Helicity amplitudes in quantum chromodynamics have a well-known behavior as momenta of external legs become collinear or soft [99, 20]. For the collinear case, at tree-level in quantum chromodynamics when two nearest neighboring legs in the color-stripped amplitudes become collinear, e.g., , , and , the amplitude behaves as [99]:

The function is a splitting amplitude, and is the helicity of the intermediate state . (The other helicity labels are implicit.) The contribution given in Eq. (21) is singular for parallel to ; other terms in the amplitude are suppressed by a power of , which vanishes in the collinear limit, compared to the ones in Eq. (21). For the pure glue case, one such splitting amplitude is where are spinor inner products, and is a momentum-dependent phase that may be found in, for example, Ref. [99]. In general, it is convenient to express splitting amplitudes in terms of these spinor inner products. The ‘’ and ‘’ labels refer to the helicity of the outgoing gluons. Since the spinor inner products behave as , the splitting amplitudes develop square-root singularities in the collinear limits. If the two collinear legs are not next to each other in the color ordering, then there is no singular contribution, e.g. no singularity develops in for collinear to .From the structure of the KLT relations it is clear that a universal collinear behavior similar to Eq. (21) must hold for gravity since gravity amplitudes can be obtained from gauge theory ones. The KLT relations give a simple way to determine the gravity splitting amplitudes, . The value of the splitting amplitude may be obtained by taking the collinear limit of two of the legs in, for example, the five-point amplitude. Taking parallel to in the five-point relation (11) and using Eq. (13) yields:

where More explicitly, using Eq. (22) then gives: Using the KLT relations at -points, it is not difficult to verify that the splitting behavior is universal for an arbitrary number of external legs, i.e.: (Since the KLT relations are not manifestly crossing-symmetric, it is simpler to check this formula for some legs being collinear rather than others; at the end all possible combinations of legs must give the same results, though.) The general structure holds for any particle content of the theory because of the general applicability of the KLT relations.In contrast to the gauge theory splitting amplitude (22), the gravity splitting amplitude (26) is not singular in the collinear limit. The factor in Eq. (25) has canceled the pole. However, a phase singularity remains from the form of the spinor inner products given in Eq. (23), which distinguishes terms with the splitting amplitude from any others. In Eq. (23), the phase factor rotates by as and rotate once around their sum as shown in Figure 8. The ratio of spinors in Eq. (26) then undergoes a rotation accounting for the angular-momentum mismatch of between the graviton and the pair of gravitons and . In the gauge theory case, the terms proportional to the splitting amplitudes (21) dominate the collinear limit. In the gravitational formula (27), there are other terms of the same magnitude as as . However, these non-universal terms do not acquire any additional phase as the collinear vectors and are rotated around each other. Thus, they can be separated from the universal terms. The collinear limit of any gravity tree amplitude must contain the universal terms given in Eq. (27) thereby putting a severe restriction on the analytic structure of the amplitudes.

Even for the well-studied case of momenta becoming soft one may again use the KLT relation to extract the behavior and to rewrite it in terms of the soft behavior of gauge theory amplitudes. Gravity tree amplitudes have the well known behavior [139]

as the momentum of graviton becomes soft. In Eq. (28) the soft graviton is taken to carry positive helicity; parity can be used to obtain the other helicity case.One can obtain the explicit form of the soft factors directly from the KLT relations, but a more symmetric looking soft factor can be obtained by first expressing the three-graviton vertex in terms of a Yang–Mills three-vertex [26] (see Eq. (40)). This three-vertex can then be used to directly construct the soft factor. The result is a simple formula expressing the universal function describing soft gravitons in terms of the universal functions describing soft gluons [26]:

where is the eikonal factor for a positive helicity soft gluon in QCD labeled by , and and are labels for legs neighboring the soft gluon. In Eq. (29) the momenta and are arbitrary null “reference” momenta. Although not manifest, the soft factor (29) is independent of the choices of these reference momenta. By choosing and the form of the soft graviton factor for used in, for example, Refs. [10, 22, 23] is recovered. The important point is that in the form (29), the graviton soft factor is expressed directly in terms of the QCD gluon soft factor. Since the soft amplitudes for gravity are expressed in terms of gauge theory ones, the probability of emitting a soft graviton can also be expressed in terms of the probability of emitting a soft gluon.One interesting feature of the gravitational soft and collinear functions is that, unlike the gauge theory case, they do not suffer any quantum corrections [23]. This is due to the dimensionful nature of the gravity coupling , which causes any quantum corrections to be suppressed by powers of a vanishing kinematic invariant. The dimensions of the coupling constant must be absorbed by additional powers of the kinematic invariants appearing in the problem, which all vanish in the collinear or soft limits. This observation is helpful because it can be used to put severe constraints on the analytic structure of gravity amplitudes at any loop order.

http://www.livingreviews.org/lrr-2002-5 |
© Max Planck Society and the author(s)
Problems/comments to |