Loops in AdS: from the spectral representation to position space

Research output: Contribution to journalArticlepeer-review


We compute a family of scalar loop diagrams in AdS. We use the spectral representation to derive various bulk vertex/propagator identities, and these identities enable to reduce certain loop bubble diagrams to lower loop diagrams, and often to tree- level exchange or contact diagrams. An important example is the computation of the finite coupling 4-point function of the large-N conformal O(N) model on AdS3. Remarkably, the re-summation of bubble diagrams is equal to a certain contact diagram: the D¯1,1,32,32(zz¯) function. Another example is a scalar with ϕ4 or ϕ3 coupling in AdS3: we compute various 4-point (and higher point) loop bubble diagrams with alternating integer and half- integer scaling dimensions in terms of a finite sum of contact diagrams and tree-level exchange diagrams. The 4-point function with external scaling dimensions differences obeying ∆12 = 0 and ∆34 = 1 enjoys significant simplicity which enables us to compute in quite generality. For integer or half-integer scaling dimensions, we show that the M -loop bubble diagram can be written in terms of Lerch transcendent functions of the cross- ratios z and z¯. Finally, we compute 2-point bulk bubble diagrams with endpoints in the bulk, and the result can be written in terms of Lerch transcendent functions of the AdS chordal distance. We show that the similarity of the latter two computations is not a coincidence, but arises from a vertex identity between the bulk 2-point function and the double-discontinuity of the boundary 4-point function.

Original languageEnglish
Article number49
JournalJournal of High Energy Physics
Issue number6
StatePublished - 1 Jun 2020
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2020, The Author(s).


  • AdS-CFT Correspondence
  • Conformal Field Theory

ASJC Scopus subject areas

  • Nuclear and High Energy Physics


Dive into the research topics of 'Loops in AdS: from the spectral representation to position space'. Together they form a unique fingerprint.

Cite this