On multiple-brane solutions in OSFT

1. On multiple-brane solutions in OSFT Martin Schnabl Institute of Physics, Prague Academy of Sciences of the Czech Republic SFT 2010, Kyoto, October 18, 2010

2. String Field Theory 2011 • Prague, September 20-24, 2011 • organized by • Institute of Physicsof the Academy of Sciences of the Czech Republic

3. What is String Field Theory? • Field theoretic description of all excitations of a string (open or closed) at once. • Useful especially for physics of backgrounds: tachyon condensation or instanton physics • Single Lagrangian field theory which should around its various critical points describe physics of diverse D-brane backgrounds, possibly also gravitational backgrounds.

4. Open String Field Theory as the fundamental theory ? • Closed string field theory (Zwiebach 1990) ― presumably the fundamental theory of gravity ― is given by a technically ingenious construction, but it is hard to work with, and it also may be viewed as too perturbative and `effective’. • Open string field theory (see Kugo’s talk) might be more fundamental (Sen).Related to holography (Maldacena) and old 1960’s ideas of Sacharov.

5. First look at OSFT Open string field theory uses the following data Let all the string degrees of freedom be assembled in Witten (1986) proposed the following action

6. First look at OSFT This action has a huge gauge symmetry provided that the star product is associative, QB acts as a graded derivation and < . > has properties of integration. Note that there is a gauge symmetry for gauge symmetry so one expects infinite tower of ghosts – indeed they can be naturally incorporated by lifting the ghost number restriction on the string field.

7. Witten’s star product Defined by gluing three strings: It used to be a very complicated definition…

8. The elements of string field star algebra are states in the BCFT, they can be identified with a piece of a worldsheet. • By performing the path integral on the glued surface in two steps, one sees that in fact:

9. Witten’s star product as operator multiplication We have just seen that the star product obeys And therefore states obeyThe star product and operator multiplication are thus isomorphic!

10. In case you wonder what e-K is

11. Simple subsector of the star algebra • The star algebra is formed by vertex operators and the operator K. The simplest subalgebra relevant for tachyon condensation is therefore spanned by K and c. Let us be more generous and add an operator B such that QB=K. • The building elements thus obey • The derivative Q acts as

12. Classical solutions This new understanding lets us construct solutions to OSFT equations of motion easily. More general solutions are of the form Here F=F(K) is arbitraryOkawa 2006 (generalizing M.S. 2005)

13. Classes of tachyon vacuum solutions • The space of all such solutions has not been completely classified yet, although we are getting closer (Rastelli; Erler; M.S.). • Let us restrict our attention to different choices of F(K) only. • Let us call a state geometric if F(K) is of the formwhere f(α) is a tempered distribution. • Restricting to “absolutely integrable” distributions one gets the notion of L0–safe geometric states.

14. Classes of tachyon vacuum solutions • Therefore F(K) must be holomorphic for Re(K)>0 and bounded by a polynomial there. Had we further demanded boundedness by a constant we would get the so called Hardy space H∞ which is a Banach algebra. • Since formally and the state is trivial if is well defined

15. Classes of tachyon vacuum solutions • There is another useful criterion. One can look at the cohomology of the theory around a given solution. It is given by an operator • The cohomology is formally trivialized by an operator which obeys

16. Classes of tachyon vacuum solutions • Therefore in this class of solutions, the trivial ones are those for which F2(0) ≠ 1. • Tachyon vacuum solutions are those for which F2(0) = 1 but the zero of 1-F2 is first order • When the order of zero of 1-F2 at K=0 is of higher order the solution is not quite well defined, but it has been conjectured (Ellwood, M.S.) to correspond to multi-brane solutions.

17. Examples so far … trivial solution …. ‘tachyon vacuum’ only c and K turned on …. M.S. ‘05 … Erler, M.S. ’09 – the simplest solution so far

18. On multiple-brane solutions • Can one compute an energy of the general solution as a function of F? • The technology has been here for quite some time, but the task seemed daunting. Recently we addressed the problem with M. Murata.

19. On multiple-brane solutions We need to evaluate a general correlator of the form To achieve this we use and insert the identity in the form

20. On multiple-brane solutions • At the end all correlatorsare given be double integral over z and s. • To compute the energy is rather straightforwad (although requires some human-aided computer algebra) The final expression is simply • The second term is a total derivative and does not contribute provided G does not blow up at infinity.

21. On multiple-brane solutions • So the energy simplifies to • The integral can be reduced to a sum of residues, branch cut contributions and possibly a large z surface term by deforming the contour to the right. Note that G(z) has to be holomorphic for Re z>0. • For a function of the form we find

22. On multiple-brane solutions To get further support let us look at the Ellwood invariant(or Hashimoto-Itzhaki-Gaiotto-Rastelli-Sen-Zwiebach invariant) . Ellwood proposed that for general OSFT solutions it would obeylet us calculate the LHS:For we find !

23. Comments on level expansion • Is it really sensible to have inverse powers of K ?To test that we have computed several lowest level coefficients of • The coefficient of c1|0> diverges for negative n≤-4. This is good, as we do not like ‘ghost branes’ as solutions of OSFT. Cases n=-2 and n=-3 should be ruled out separately, n=-1 is allowed (tachyon vacuum).

24. Comments on level expansion • To study other coefficients we can easily compute the overlapwhere , and V is a matter field of scaling dimension hV. The ghosts contribute 1/t2, so the integral is perfectly convergent. For the integral however diverges for n≥3 asso that ghost coefficients starting with c-2 |0> are divergent.

25. Comments on level expansion • The most promising candidate for the n+1 brane solution is however • The reason is that the energy computation is free of surface terms, but perhaps more importantly 1/(1-F2) is holomorphic in Re(z)>0. (This is not true for F~z-n/2.)

26. What next ? • Find the ‘best’ choice of F • Understand in detail level expansion. Is there some F for which the level expansion is fully regular for all n>0? Is it possible to find regular twist symmetric solutions? • Work out the cohomology around these solutions. (We have seen some hints that we indeed get (n+1)2 copies of the original cohomology.)