We develop the derivation we proposed in [37] of the dressing phase of the S-matrix in the AdS/CFT correspondence in the framework of the underlying bare integrable model. We elaborate the configuration of the Bethe roots describing the physical vacuum, which consists of a long Bethe string stretched along the imaginary axis and stacks distributed along the real axis. We determine the distribution of all Bethe roots in the thermodynamic limit. We then directly compute the scattering phase of the fundamental excitations over the physical vacuum and reproduce the BHL/BES dressing phase.
JHEP12(2007)044so that it satisfies the crossing symmetry [13] and includes the previously known first two terms [14 -17] which reproduce the semi-classical string spectrum. Subsequently, a systematic way of its determination as the weak coupling expansion was presented [18]: The problem can be rephrased in terms of the cusp anomalous dimension [19] where the transcendentality principle [20], with some empirical rules, fully determines the dressing phase up to an overall multiplicative constant. The constant is readily singled out by comparison with either the perturbative computation [21] or the strong coupling result [12]. Ultimately the weak coupling result is identified with the strong coupling one by a sort of analytic continuation [18,22] and is nicely expressed in a closed integral formula. We call it the BHL/BES dressing phase after the authors of the articles [12,18]. Its properties, such as the pole structure [23] as well as the strong coupling limit [24 -27], have been further studied.Despite the success in the determination, the clear understanding of the scalar factor was still lacking. The above procedures do not explain why the scalar factor should exhibit its particular structure. It is also unsatisfactory that these procedures require some modelspecific computation of the string/gauge theory. Although there are some interesting results explaining part of its structure [23, 28 -30], one would desire a comprehensive explanation.Let us recall here that in the field of integrable models, there are two well-known approaches for the computation of the S-matrices: One is called the factorized bootstrap program or the phenomenological computation [31], the other is called the direct calculation, the microscopic derivation or the Bethe ansatz technique [32 -34].The former approach is to compute the S-matrices as an inverse problem. In twodimensional massive relativistic integrable models, two-body S-matrices of the fundamental particles satisfy the unitarity, the factorizability, and the crossing symmetry. These conditions constrain the form of the S-matrices up to the CDD ambiguity. The ambiguity can be removed by some additional requirements, such as the absence of the poles corresponding to unphysical particles.The latter approach is to compute the S-matrices as a direct problem. For example, in the anti-ferromagnetic Heisenberg spin-chain the physical vacuum is the anti-ferromagnetic state rather than the ferromagnetic ...