betaFeedback

WorksPhysics and Astronomy

Remark on residual solutions in the KBcKBc algebra

Abstract

In this paper we study residual solutions of open bosonic string field theory in the KBcKBc algebra with the auxiliary identities (∂mc)2=0(\partial^mc)^2=0 of the ghost field. These solutions are not covered by the classification based on the defining relations alone. We show that the operator which controls the cohomology on the residual family depends only on s=γ1+γ2s=γ_1+γ_2. At positive level nn, cohomology can appear only on the lines s=±ns=\pm n. We compute it through level four and find one class at every point of each line s=−3s=-3 and s=−4s=-4. These classes give exact polynomial deformations of the residual family. At level six on s=−6s=-6, the cohomology instead vanishes generically and has dimension two at (−1,−5)(-1,-5) and (−5,−1)(-5,-1). We also compute the failure of the standard generator redefinition to preserve the auxiliary identities, and discuss the regularization of an explicit residual solution.

Introduction

In 2005 Schnabl found the first analytic solution for tachyon condensation in Witten’s open string field theory [1, 2]. In 2009 Erler and Schnabl found a simpler solution [3]; a short account is given in [4]. Since then the KBcKBc algebra has become a standard tool for constructing analytic solutions. Reviews can be found in [5, 6, 7], and a broader introduction to string field theory is given in [8]. Many solutions in this algebra contain the identity string field. Their interpretation requires care [9, 10], and identity-based solutions continue to appear in studies of open string renormalization group flow [11, 12].

In 2012 Erler classified solutions which admit an L−\mathcal{L}^{-} level expansion and satisfy the equations of motion by the defining relations of the KBcKBc algebra alone [13]. Under these assumptions there are six gauge orbits: the perturbative vacuum, the tachyon vacuum and four residual solutions. The residual solutions are singular from the point of view of the identity string field.

The ghost field satisfies additional relations, (∂mc)2=0(\partial^{m}c)^{2}=0 for m≥1m \ge1, which do not follow from the defining relations. These are the auxiliary identities of [13]. With them, the four residual solutions become points of a two-parameter family. The classification does not cover solutions which need these identities. It was also observed in [13] that the cohomology of the residual family may be enhanced at special points, with a possible connection to multibrane solutions [14].

Here we study the cohomology of this family. We first give a criterion for imposing the auxiliary identities: only the part antisymmetric in the ghost derivatives survives. We then show that the operator which controls the cohomology depends only on s=γ1+γ2s=\gamma_{1}+\gamma_{2}. At positive level nn, a class can occur only when s=±ns=\pm n. Through level four the positive lines have no cohomology. On the negative lines we recover the known level-two class at the ghost brane point and find one class at every point of each line s=−3s=-3 and s=−4s=-4. These classes give exact polynomial solutions without higher-order corrections.

We also examine level six on s=−6s=-6. Here the cohomology vanishes at generic points and has dimension two at (−1,−5)(-1,-5) and (−5,−1)(-5,-1). Thus the classes found on the whole lines at levels three and four do not imply the same behavior at higher levels. Finally, we compute explicitly how the generator redefinition used in [13] fails to preserve the auxiliary identities.

Our interest in this question started from the family ψb=c(1+K)Bc((1+K)−1+b)\psi_b=c(1+K)Bc((1+K)^{-1}+b), discussed in the lecture [15]. This is a special case of the two-parameter solution in § 2.3 of [16]. It also appears in equation (86) of [17], with λ=1\lambda=1 and α=1+b\alpha=1+b. We return to this family and its regularization in § 7.

The cohomology calculations and the new polynomial solutions are statements in the algebra with auxiliary identities. Their interpretation as regular string backgrounds, and the definition of their energy, require further analysis.1

The rest of the paper is organized as follows. In § 2 we review the KBcKBc algebra and the auxiliary identities. § 3 describes the residual family, and § 4 studies its cohomology through level four. In § 5 we use the cohomology classes to construct new solutions. § 6 discusses the generator redefinition, and § 7 examines the explicit family ψb\psi_b and its regularization. We conclude in § 8. Appendix § 10 gives the closed string overlap, and appendix § 11 contains the level-six calculation.

The KBc algebra and auxiliary identities

The KBc algebra

The string fields KK, BB and cc have Grassmann parity even, odd and odd, respectively. Their defining relations are

{B,c}=1,B2=c2=0,[K,B]=0,QB=K,QK=0,Qc=cKc.(1)\{B,c\}=1,\qquad B^2=c^2=0,\qquad[K,B]=0,\qquad QB=K,\qquad QK=0,\qquad Qc=cKc. \tag*{(1)}

We suppress the star product symbol and write Ωt=e−tK\Omega^t=e^{-tK}. The BRST operator around ψ\psi is QψX=QX+ψX−(−1)∣X∣XψQ_\psi X=QX+\psi X-(-1)^{|X|}X\psi. For any function X(K)X(K), the defining relations give

cBc=c,BcXBc=BXc,Q(Bc)=cKBc,(Bc)2=Bc.(2)cBc=c,\qquad BcXBc=BXc,\qquad Q(Bc)=cKBc,\qquad(Bc)^2=Bc. \tag*{(2)}

Thus P=BcP=Bc and Pˉ=cB=1−P\bar{P}=cB=1-P are complementary projectors. For a function G(K)G(K), set M=K/(1−G)M=K/(1-G). Then

ψG=cMBcG(3)\psi_G=cMBcG \tag*{(3)}

is the left form of Okawa’s solution [18]. The case G=(1+K)−1G=(1+K)^{-1} gives the Erler–Schnabl solution ψES\psi_{\mathrm{ES}}.2 We normalize the energy so that E[ψES]=−1/(2π2)E[\psi_{\mathrm{ES}}]=-1/(2\pi^2).

Auxiliary identities

Write ∂X=[K,X]\partial X=[K,X] and cj=∂jcc_j=\partial^j c. In addition to the defining relations, the ghost field obeys

cj2=0(j≥1),and hence{ci,cj}=0(i,j≥0).(4)c_j^2=0\quad(j\ge1),\qquad\text{and hence}\qquad\{c_i,c_j\}=0\quad(i,j\ge0). \tag*{(4)}

The anticommutators follow by differentiating the squared identities, including c2=0c^2=0. More explicitly, induction on i+ji+j relates adjacent pairs of indices, while the middle pair is fixed by a squared identity or its derivative. These relations are independent of (1) [13]. They allow us to write every polynomial string field as

∑ar;i1…ip;ηKrci1⋯cipBη,r≥0,i1<⋯<ip,η∈{0,1},(5)\sum a_{r;i_1\ldots i_p;\eta}K^r c_{i_1}\cdots c_{i_p}B^\eta,\qquad r\ge0,\quad i_1<\cdots<i_p,\quad\eta\in\{0,1\}, \tag*{(5)}

using the rules

[K,cj]=cj+1,{B,cj}=δj0,Qcj=∑r=0j(jr)crcj−r+1.(6)[K,c_j]=c_{j+1},\qquad\{B,c_j\}=\delta_{j0},\qquad Qc_j=\sum_{r=0}^{j}\binom{j}{r}c_r c_{j-r+1}. \tag*{(6)}

A monomial in (5) has ghost number p−ηp-\eta and level

ℓ=r+∑a=1p(ia−1)+η,(7)\ell=r+\sum_{a=1}^{p}(i_a-1)+\eta, \tag*{(7)}

At fixed ghost number and level the state space is therefore finite-dimensional. We compute cohomology in this polynomial algebra.3

A simple criterion

The auxiliary identities can also be imposed in a polynomial representation. Associate the word

f0(K)cf1(K)c⋯cfm(K)[B],(8)f_0(K)c f_1(K)c\cdots c f_m(K)[B], \tag*{(8)}

with a polynomial F(x0,…,xm)F(x_0,\ldots,x_m), where xix_i is the argument of fif_i, and define gi=xi−xi−1g_i=x_i-x_{i-1}. Differentiating the ii-th ghost multiplies FF by −gi-g_i. The relations {cp,cq}=0\{c_p,c_q\}=0 therefore remove the part symmetric under the exchange of two neighbouring gaps. Since neighbouring transpositions generate SmS_m, we obtain

F=0 in the auxiliary quotient⟺∑π∈Smsgn⁡(π)F(x0;gπ(1),…,gπ(m))=0.(9)F=0\ \text{in the auxiliary quotient}\quad\Longleftrightarrow\quad\sum_{\pi\in S_m}\operatorname{sgn}(\pi)F(x_0;g_{\pi(1)},\ldots,g_{\pi(m)})=0. \tag*{(9)}

Thus only the part antisymmetric in the ghost derivatives survives. This criterion reproduces the cohomology around the four residual solutions given in table 2 of [13] at levels −1-1 to 2.

The residual family

With the auxiliary identities, the level-zero residual solutions belong to the family [13]

Ψ0=γ1cK+γ2Kc−(1+γ1+γ2)cKBc,s=γ1+γ2.(10)\Psi_0=\gamma_1cK+\gamma_2Kc-(1+\gamma_1+\gamma_2)cKBc,\qquad s=\gamma_1+\gamma_2. \tag*{(10)}

The points (0,0)(0,0), (−1,0)(-1,0), (0,−1)(0,-1) and (−1,−1)(-1,-1) are the residual perturbative vacuum, the residual tachyon vacuum, the residual conjugate tachyon vacuum and the ghost brane solution [19]. Only these four points satisfy the equations of motion by (1) alone. The remaining points require the auxiliary identities. We write D=QΨ0D=Q\Psi_0.

Following [13], introduce the odd derivation β=12B−\beta=\frac{1}{2}\mathcal{B}^{-}, with βK=B\beta K=B and βB=βc=0\beta B=\beta c=0. It is well defined in the auxiliary quotient, since βc1=1\beta c_1=1 and βcj=0\beta c_j=0 for j≠1j\ne1, and hence it preserves the anticommutation relations. Using β(cK)=−cB\beta(cK)=-cB, β(Kc)=Bc\beta(Kc)=Bc and β(cKBc)=0\beta(cKBc)=0, we obtain βΨ0=γ2−scB\beta\Psi_0=\gamma_2-scB. Consequently,

L≡{D,β}=12L−−s ad⁡cB.(11)\mathcal{L}\equiv\{D,\beta\}=\frac{1}{2}\mathcal{L}^{-}-s\,\operatorname{ad}_{cB}. \tag*{(11)}

The operator therefore depends only on ss. If Dφ=0D\varphi=0 and Lφ=hφ\mathcal{L}\varphi=h\varphi with h≠0h\ne0, then φ=D(βφ/h)\varphi=D(\beta\varphi/h), so the cohomology lies in the kernel of L\mathcal{L}. Since cBcB is a projector, ad⁡cB\operatorname{ad}_{cB} has the eigenvalues e=−1,0,+1e=-1,0,+1 on PXPˉPX\bar{P}, PXP+PˉXPˉPXP+\bar{P}X\bar{P} and PˉXP\bar{P}XP, respectively. Moreover D(cB)=0D(cB)=0, so DD preserves these sectors and the level. A cohomology class at level mm must therefore satisfy

m=se.(12)m=se. \tag*{(12)}

At ghost number one the possible classes are the marginal directions at m=0m=0; the state cc at m=−1m=-1, e=+1e=+1, which appears only for s=−1s=-1; and states at positive level m=nm=n, with e=−1e=-1 on the lines s=−ns=-n or with e=+1e=+1 on the lines s=ns=n.

At level zero, the tangent vectors v1=cK−cKBcv_1=cK-cKBc and v2=Kc−cKBcv_2=Kc-cKBc are closed and independent. The only other state with ghost number one is cKBccKBc, and using (6) we find

D(cKBc)=sKc0c1−γ1c0c2.(13)D(cKBc)=sKc_0c_1-\gamma_1c_0c_2. \tag*{(13)}

Both states with ghost number zero, 11 and BcBc, are annihilated by DD. Hence the level-zero cohomology is two-dimensional everywhere except at (0,0)(0,0), where the residual perturbative vacuum has the additional class cKBccKBc [13]. At level −1-1 the state cc is closed only for s=−1s=-1, and then it is not exact.

Cohomology at positive level

For 1≤n≤41 \le n \le4, the sector e=+1e=+1 on s=ns=n has no closed states of ghost number one. We therefore consider e=−1e=-1 on s=−ns=-n, and set γ1=a\gamma_1=a, γ2=−n−a\gamma_2=-n-a. In this sector the states with ghost number one at level nn are KrcicjBK^{r}c_i c_j B with 1≤i<j1 \le i < j and r+i+j=n+1r+i+j=n+1, and the states with ghost number zero are Kn−jcjBK^{n-j}c_jB with 1≤j≤n1 \le j \le n. Since DD preserves the level and the sectors, exactness in this sector is the same as exactness in the whole algebra.

This sector has no ghost number one states at level one. At level two its only state is u=c1c2Bu=c_1c_2B. We find

Du=0,D(Kc1B)=−(a+1)u,D(c2B)=0.(14)Du=0,\qquad D(Kc_1B)=-(a+1)u,\qquad D(c_2B)=0. \tag*{(14)}

A class therefore survives only at a=−1a=-1, the ghost brane point [13, 19].

At level three a basis is u=Kc1c2Bu=Kc_1c_2B, v=c1c3Bv=c_1c_3B. The differential is

Du=Dv=0,D(K2c1B)=−(2a+3)u+(a+2)v,D(Kc2B)=D(c3B)=0.(15)Du=Dv=0,\qquad D(K^2c_1B)=-(2a+3)u+(a+2)v,\qquad D(Kc_2B)=D(c_3B)=0. \tag*{(15)}

Since 2a+32a+3 and a+2a+2 never vanish together, the image is one-dimensional for every aa. The state

φ3=2u−v=(2Kc1c2−c1c3)B(16)\varphi_3=2u-v=(2Kc_1c_2-c_1c_3)B \tag*{(16)}

is closed and is not exact for any aa, since det⁡(−(2a+3)a+22−1)=−1\det\begin{pmatrix} -(2a+3) & a+2 \\ 2 & -1 \end{pmatrix}=-1.

At level four, write

u1=K2c1c2B,u2=Kc1c3B,u3=c1c4B,u4=c2c3B,w=c1c2c3B.(17)u_1=K^2c_1c_2B,\qquad u_2=Kc_1c_3B,\qquad u_3=c_1c_4B,\qquad u_4=c_2c_3B,\qquad w=c_1c_2c_3B. \tag*{(17)}

On the ghost number one states,

Du1=Du2=−(a+3)w,Du3=−2w,Du4=0.(18)Du_1=Du_2=-(a+3)w,\qquad Du_3=-2w,\qquad Du_4=0. \tag*{(18)}

while on the ghost number zero states,

D(K3c1B)=−3(a+2)u1+(3a+8)u2−(a+3)u3,D(K2c2B)=D(Kc3B)=(a+3)u4,D(c4B)=2u4.(19)\begin{aligned} D(K^3c_1B)&=-3(a+2)u_1+(3a+8)u_2-(a+3)u_3,\\ D(K^2c_2B)&=D(Kc_3B)=(a+3)u_4,\qquad D(c_4B)=2u_4. \tag*{(19)} \end{aligned}

The kernel is therefore three-dimensional and the image two-dimensional for every aa. The state

φ4=u1−u2=(K2c1c2−Kc1c3)B(20)\varphi_4=u_1-u_2=(K^2c_1c_2-Kc_1c_3)B \tag*{(20)}

is closed. For a≠−3a \ne-3 it is not exact, since it has no u3u_3 component, and at a=−3a=-3 the first image vector becomes 3u1−u23u_1-u_2, which is independent of u1−u2u_1-u_2. Thus φ4\varphi_4 is not exact anywhere on the line. We collect these results in Table 1.

s=−ns=-n, level nn, e=−1e=-1dim⁡ker⁡D\dim\ker Ddim⁡im⁡D\dim\operatorname{im}Ddim⁡H1\dim H^{1}
n=1n=1000
n=2n=2, a≠−1a\ne-1110
n=2n=2, a=−1a=-1101
n=3n=3, all aa211
n=4n=4, all aa321

Table 1. Ghost number one cohomology in the sector e=−1e=-1 at level nn on the line γ1+γ2=−n\gamma_{1}+\gamma_{2}=-n, with γ1=a\gamma_{1}=a.

At level six on s=−6s=-6, the cohomology is instead zero at generic points and two-dimensional at (−1,−5)(-1,-5) and (−5,−1)(-5,-1). The calculation is given in appendix § 11.

New solutions

The classes at levels two, three and four give exact solutions. At a point Ψ0\Psi_{0} on s=−ns=-n, take

Ψ=Ψ0+ϵφn,(21)\Psi=\Psi_{0}+\epsilon\varphi_{n}, \tag*{(21)}

where ϵ\epsilon is a number and φn\varphi_{n} is one of the states found in the previous section, with φ2=c1c2B\varphi_{2}=c_{1}c_{2}B at the ghost brane point. Since Ψ0\Psi_{0} is a solution, the equation of motion gives

QΨ+Ψ2=ϵDφn+ϵ2φn2.(22)Q\Psi+\Psi^{2}=\epsilon D\varphi_{n}+\epsilon^{2}\varphi_{n}^{2}. \tag*{(22)}

The first term is zero because Dφn=0D\varphi_{n}=0. Also φn2=0\varphi_{n}^{2}=0: all ghosts in φn\varphi_{n} carry derivatives, so the final BB in the first factor can be moved through the second factor without a contraction, giving B2=0B^{2}=0. Equivalently, φn\varphi_{n} lies in PXPˉPX\bar{P}, and PˉP=0\bar{P}P=0. Therefore (21) is a solution for every value of ϵ\epsilon.

On the line s=−3s=-3 the solution is

Ψ=a cK−(3+a) Kc+2cKBc+ϵ(2K∂c ∂2c−∂c ∂3c)B,(23)\Psi=a\,cK-(3+a)\,Kc+2cKBc+\epsilon\left(2K\partial c\,\partial^{2}c-\partial c\,\partial^{3}c\right)B, \tag*{(23)}

and on the line s=−4s=-4

Ψ=a cK−(4+a) Kc+3cKBc+ϵ(K2∂c ∂2c−K∂c ∂3c)B,(24)\Psi=a\,cK-(4+a)\,Kc+3cKBc+\epsilon\left(K^{2}\partial c\,\partial^{2}c-K\partial c\,\partial^{3}c\right)B, \tag*{(24)}

where aa can be any number. At the ghost brane point, the known level-two class [13] similarly gives

Ψ=−cK−Kc+cKBc+ϵ ∂c ∂2c B.(25)\Psi=-cK-Kc+cKBc+\epsilon\,\partial c\,\partial^{2}c\,B. \tag*{(25)}

Since φn\varphi_{n} is not exact, these are nontrivial deformations of Ψ0\Psi_{0}.4 These deformations require the auxiliary identities; for a nonzero deformation the defining relations (1) alone do not give the equations of motion.

More general formal power-series deformations along these classes are also unobstructed. At order k≥2k\ge2 the obstruction lies at level knkn, where L\mathcal{L} has the eigenvalue n(k+e)>0n(k+e)>0. Therefore a closed obstruction SkS_{k} is exact, and it can be removed by

χk=−∑eβ(PeSk)n(k+e),(26)\chi_{k}=-\sum_{e}\frac{\beta(P_{e}S_{k})}{n(k+e)}, \tag*{(26)}

Here PeP_e projects onto the sector with eigenvalue ee. For the representatives above, φn2=0\varphi_n^2=0 and all higher-order corrections vanish.

These solutions are polynomial in KK and are therefore identity-based. The positive-level terms increase their singularity. Computing an action or a boundary state requires a suitable regularization.

The generator redefinition

The energy prescription in [13] uses the generator redefinition

K^=K1+δK,B^=B1+δK,c^=c(1+δK)Bc.(27)\widehat{K}=\frac{K}{1+\delta K},\qquad\widehat{B}=\frac{B}{1+\delta K},\qquad\widehat{c}=c(1+\delta K)Bc. \tag*{(27)}

This preserves (1) and commutes with QQ. The energy argument also assumes that a BRST-exact identity-like term has zero overlap. The redefined generators need not obey the auxiliary identities [13]. Expanding to first order, K^=K−δK2+O(δ2)\widehat{K}=K-\delta K^2+O(\delta^2) and c^=c+δcKBc\widehat{c}=c+\delta cKBc, and using (6) we find

(∂^c)2=−δc1c2+O(δ2)≠0,(28)(\widehat{\partial}c)^2=-\delta c_1c_2+O(\delta^2)\ne0, \tag*{(28)}

where ∂^=[K^,⋅]\widehat{\partial}=[\widehat{K},\mathord{\cdot}]. The redefinition (27) therefore preserves the KBcKBc algebra but does not descend to the auxiliary quotient. For a solution which needs these identities, the redefined field solves the equations of motion only if the redefinition maps the defect QΨ+Ψ2Q\Psi+\Psi^2 back into the auxiliary ideal, and this has to be checked case by case.

This explains an obstruction behind the energy puzzle in [13]. The energy rule found there is proportional to s=γ1+γ2s=\gamma_1+\gamma_2, while the auxiliary identities connect the residual solutions by marginal directions, which suggests that their energies should be equal. Since the regularization behind the energy rule does not respect the auxiliary identities, the redefinition is guaranteed to preserve the equations of motion only at the four points which satisfy the defining relations. Even there, the energy assignment still depends on the assumed trace prescription. It cannot simply be extended along the family. An energy calculation on this family requires a regularization compatible with the auxiliary identities.

An explicit example

For the family discussed in the introduction, set G=(1+K)−1G=(1+K)^{-1} and M=1+KM=1+K. Then

ψb=c(1+K)Bc(11+K+b)=ψES+bΦ,Φ=c(1+K)Bc.(29)\psi_b=c(1+K)Bc\left(\frac{1}{1+K}+b\right)=\psi_{\mathrm{ES}}+b\Phi,\qquad\Phi=c(1+K)Bc. \tag*{(29)}

Using (2), we find QΦ=cKcQ\Phi=cKc, Φ2=ΦψES=0\Phi^2=\Phi\psi_{\mathrm{ES}}=0 and ψESΦ=−cKc\psi_{\mathrm{ES}}\Phi=-cKc, so ψb\psi_b solves the equations of motion for every bb.5 Expanding around K=0K=0, ψb=(1+b)c−cK+(1+b)cKBc+⋯\psi_b=(1+b)c-cK+(1+b)cKBc+\cdots, so for b≠−1b\ne-1 it is in the tachyon vacuum orbit and for b=−1b=-1 it is in the residual tachyon vacuum orbit: its level-zero term is −cK-cK [13]. The gauge transformation U=1+λBcU=1+\lambda Bc, with λ≠−1\lambda\ne-1, acts as 1+b↦(1+λ)(1+b)1+b\mapsto(1+\lambda)(1+b), so b=−1b=-1 is a fixed point.

At b=−1b=-1 there is no homotopy field in this KBcKBc algebra. A field of ghost number minus one has the form BF(K)BF(K), and

Qψ−1(BF)=−[cKF+FMc(G−1)]B,(30)Q_{\psi_{-1}}(BF)=-\left[cKF+FMc(G-1)\right]B, \tag*{(30)}

and the right-hand side vanishes when multiplied by BB from the right, while 1⋅B≠01\cdot B\ne0. For functions with suitable Laplace representations the closed string overlap [20] is (see appendix § 10)

W[f0cf1]=−(f1f0)′(0),W[f0cf1cf2B]=−f1′(0)(f2f0)(0),(31)W[f_0cf_1]=-(f_1f_0)'(0),\qquad W[f_0cf_1cf_2B]=-f_1'(0)(f_2f_0)(0), \tag*{(31)}

which gives W[ψG]=−M(0)G′(0)W[\psi_G]=-M(0)G'(0) for regular Okawa solutions. Formally, the energy and the overlap of ψb\psi_b are those of the tachyon vacuum for every bb [13, 15, 16], but for b≠0b\ne0 the traces contain identity-like terms, and one needs a prescription to define them. The interpretation of singular string fields in terms of generalized states is discussed in [21]. For related discussions of identity-based solutions and the distinction between regular and singular gauge transformations, see [17].

To regularize the gauge parameter, take Rμ=(1+μK)−1R_\mu=(1+\mu K)^{-1}, μ>0\mu>0, and Ubμ=1+bBcRμU_b^\mu=1+bBcR_\mu. Then

ψbRμ=(Ubμ)−1(Q+ψES)Ubμ=f−1ψG~f,f=1+bRμ,G~=G+bRμf,(32)\psi_b^{R_\mu}=(U_b^\mu)^{-1}(Q+\psi_{\mathrm{ES}})U_b^\mu=f^{-1}\psi_{\widetilde{G}}f,\qquad f=1+bR_\mu,\qquad\widetilde{G}=\frac{G+bR_\mu}{f}, \tag*{(32)}

Thus the regularized field is an Okawa solution conjugated by a function of KK. For b>−1b>-1 the pole of f−1f^{-1} lies on the negative real axis, so the functions in the gauge parameter and its inverse admit Laplace representations supported at nonnegative wedge widths. With the usual wedge-state prescription this gives a regular gauge transformation of the tachyon vacuum. Its coefficientwise limit as μ→0\mu\to0 is ψb\psi_b; this does not establish continuity of singular traces.6 For b=−1b=-1, however, fμ=μK/(1+μK)f_\mu=\mu K/(1+\mu K) vanishes at K=0K=0 and G~μ=−(1−μ)G/μ\widetilde{G}_\mu=-(1-\mu)G/\mu, so (32) is a singular conjugation. Its coefficientwise limit in the localized algebra can be computed explicitly:

lim⁡μ→0ψ−1Rμ=−K−1c(1+K)KBcK1+K≡ψ∗,(ψ∗−ψ−1)B=K−1[K,c]KB≠0.(33)\lim_{\mu\to0}\psi_{-1}^{R_\mu}=-K^{-1}c(1+K)KBc\frac{K}{1+K}\equiv\psi_*, \qquad(\psi_*-\psi_{-1})B=K^{-1}[K,c]KB\ne0. \tag*{(33)}

The regularized family therefore does not return to the residual field. For traces which are defined at finite cutoff, one can see where the difficulty comes from: for SN=∑j=0NRjS_N=\sum_{j=0}^{N}R^j we have Tr⁡[SNX(1−R)]=Tr⁡X−Tr⁡[RN+1X]\operatorname{Tr}[S_NX(1-R)]=\operatorname{Tr}X-\operatorname{Tr}[R^{N+1}X], and the boundary term has to be controlled before an energy can be assigned to the singular conjugation.

For b<−1b<-1, a real continuous regulator with R(0)=1R(0)=1 and R(∞)=0R(\infty)=0 makes 1+bR1+bR vanish at some K∗>0K_*>0, where the numerator G(K∗)−1G(K_*)-1 of G~\widetilde{G} does not vanish. Thus this regulated real projector ansatz cannot connect the two signs of 1+b1+b by a regular transformation, although more general gauge parameters are not excluded. For the role of K=0K=0 and K=∞K=\infty in singular solutions see [22, 23].

Conclusions

In this paper we studied residual solutions of the KBcKBc algebra with the auxiliary identities of the ghost field. The operator which controls the cohomology depends only on s=γ1+γ2s=\gamma_1+\gamma_2 and restricts possible classes at positive level nn to s=±ns=\pm n. Through level four we find no cohomology on the positive lines. On the negative lines there is one class at every point of s=−3s=-3 and s=−4s=-4, while at level two the class survives only at the ghost brane point. The explicit representatives give exact polynomial deformations of the residual solutions.

The calculation in appendix § 11 shows a different behavior at level six. On s=−6s=-6 the cohomology vanishes generically and has dimension two at (−1,−5)(-1,-5) and (−5,−1)(-5,-1). A general description must therefore account for both classes which persist along a whole line and enhancement at isolated points.

We also showed that the standard generator redefinition breaks the auxiliary identities already at first order. The energy prescription based on this redefinition cannot be extended directly to the residual family. For the explicit family ψb\psi_b, regularizing the gauge parameter gives a regular gauge transformation for b>−1b>-1 with the stated wedge-state prescription. At b=−1b=-1 the coefficientwise limit differs from the original residual field.

The main open question is the cohomology at arbitrary level. It would also be useful to find a regularization compatible with the auxiliary identities and to determine whether any of these formal solutions admit a boundary-state or multibrane interpretation [14]. Other realizations of the KBcKBc algebra, such as those in [24], should be checked against the same identities.

Acknowledgments

I am deeply grateful to Loriano Bonora, with whom this project was started and from whom I learned this subject. I am also grateful to T. Erler for correspondence on the family (29). I acknowledge support from the Istanbul Integrability and Stringy Topics Initiative (IStringy). Algebraic checks and the preparation of the draft were assisted by Claude (Anthropic), and ChatGPT (OpenAI).

The closed string overlap

Let VV be an on-shell closed string vertex operator of ghost number two and weight (0,0)(0,0), inserted at the midpoint. Normalize its correlator by ⟨Vc(x)⟩CL=(L/π)CV\langle Vc(x)\rangle_{\mathrm{CL}}=(L/\pi)C_V. If FF has a Laplace representation with an integrable first moment, then

F(K)=∫0∞dt F~(t)e−tK,Tr⁡V[cF]=−CVπF′(0),(34)F(K)=\int_{0}^{\infty}dt\,\widetilde{F}(t)e^{-tK},\qquad\operatorname{Tr}_{V}[cF]=-\frac{C_V}{\pi}F'(0), \tag*{(34)}

Cyclicity gives the first formula in (31). For the second, choose the orientation of the BB line so that, for widths ti≥0t_i\geq0,

Tr⁡V[Ωt0cΩt1cΩt2B]=CVπt1.(35)\operatorname{Tr}_{V}[\Omega^{t_0}c\Omega^{t_1}c\Omega^{t_2}B]=\frac{C_V}{\pi}t_1. \tag*{(35)}

Integrating over the widths we obtain

Tr⁡V[f0cf1cf2B]=−CVπf1′(0)(f2f0)(0).(36)\operatorname{Tr}_{V}[f_0cf_1cf_2B]=-\frac{C_V}{\pi}f_1'(0)(f_2f_0)(0). \tag*{(36)}

This sign agrees with Q(BcF)=cKF−cKcFBQ(BcF)=cKF-cKcFB and the vanishing overlap of BRST exact fields. We define WW as the coefficient of CV/πC_V/\pi. For regular Okawa solutions, writing Bc=1−cBBc=1-cB gives W[Ψ]=−(MG)′(0)+M′(0)G(0)=−M(0)G′(0)W[\Psi]=-(MG)'(0)+M'(0)G(0)=-M(0)G'(0). These formulas assume that the traces converge; they cannot be used to cancel singular factors inside otherwise undefined traces.

Cohomology at level six

Here we give the level-six calculation on s=−6s=-6. Set γ1=a\gamma_1=a and γ2=−6−a\gamma_2=-6-a. In the sector e=−1e=-1, bases at ghost numbers zero, one and two are

hj=K6−jcjB,j=1,…,6,(37)h_j=K^{6-j}c_jB,\qquad j=1,\ldots,6, \tag*{(37)}
(u1,u2,u3)=(K4c1c2B, K3c1c3B, K2c1c4B),(u_1,u_2,u_3)=\left(K^4c_1c_2B,\ K^3c_1c_3B,\ K^2c_1c_4B\right),
(u4,u5,u6)=(Kc1c5B, c1c6B, K2c2c3B),(u_4,u_5,u_6)=\left(Kc_1c_5B,\ c_1c_6B,\ K^2c_2c_3B\right),
(u7,u8,u9)=(Kc2c4B, c2c5B, c3c4B),(38)(u_7,u_8,u_9)=\left(Kc_2c_4B,\ c_2c_5B,\ c_3c_4B\right), \tag*{(38)}

and

(w1,w2,w3,w4)=(K2c1c2c3B, Kc1c2c4B, c1c2c5B, c1c3c4B),(39)(w_1,w_2,w_3,w_4)=\left(K^2c_1c_2c_3B,\ Kc_1c_2c_4B,\ c_1c_2c_5B,\ c_1c_3c_4B\right), \tag*{(39)}

respectively. The matrices of DD in these ordered bases are

D0=(v05×504×1−A),D1=(A04×4),(40)D_0=\begin{pmatrix}v&0_{5\times5}\\0_{4\times1}&-A\end{pmatrix},\qquad D_1=\begin{pmatrix}A&0_{4\times4}\end{pmatrix}, \tag*{(40)}

where

v=(−5a−1510a+40−10a−455a+24−a−5),A=(−6a−26−3a−12−2004a+190−2a−9−50−a−500−a−5−90a+5a+50−5).(41)v=\begin{pmatrix}-5a-15\\10a+40\\-10a-45\\5a+24\\-a-5\end{pmatrix},\qquad A=\begin{pmatrix} -6a-26&-3a-12&-2&0&0\\ 4a+19&0&-2a-9&-5&0\\ -a-5&0&0&-a-5&-9\\ 0&a+5&a+5&0&-5 \end{pmatrix}. \tag*{(41)}

The relation Av=0Av=0 gives D1D0=0D_1D_0=0, and vv is nonzero for every aa. The greatest common divisor of the nonzero 4×44\times4 minors of AA is proportional to (a+1)(a+5)(a+1)(a+5). At a=−1a=-1 and a=−5a=-5 the rank is three, while it is four everywhere else. The cohomology is therefore

dim⁡H1∣ℓ=6,s=−6={2,a=−1 or a=−5,0,a∉{−1,−5}.(42)\dim H^1\big|_{\ell=6,s=-6}= \begin{cases} 2,&a=-1\text{ or }a=-5,\\ 0,&a\notin\{-1,-5\}. \end{cases} \tag*{(42)}

Indeed, dim⁡ker⁡D1=9−rank⁡A\dim\ker D_1=9-\operatorname{rank}A and dim⁡im⁡D0=1+rank⁡A\dim\operatorname{im}D_0=1+\operatorname{rank}A. Thus the cohomology is enhanced at the two points (γ1,γ2)=(−1,−5)(\gamma_1,\gamma_2)=(-1,-5) and (−5,−1)(-5,-1), and vanishes at generic points of the line. At this level the other two projector sectors are contractible by (11), so (B.6) gives the full ghost number one cohomology at level six on s=−6s=-6.

References

References

  1. [1]E. Witten, “Non-commutative geometry and string field theory,” Nucl. Phys. B 268 no. 2, (1986) 253–294.DOI
  2. [2]M. Schnabl, “Analytic solution for tachyon condensation in open string field theory,” Adv. Theor. Math. Phys. 10 no. 4, (2006) 433–501, arXiv:hep-th/0511286.arxiv.org/abs/hep-th/0511286
  3. [3]T. Erler and M. Schnabl, “A Simple Analytic Solution for Tachyon Condensation,” JHEP 10 (2009) 066, arXiv:0906.0979 [hep-th].DOI
  4. [4]T. Erler, “A simple analytic solution for tachyon condensation,” Theor. Math. Phys. 163 no. 3, (2010) 705–709.
  5. [5]L. Bonora, “A review on open SFT,” PoS ICMP2013 (2014) 001.
  6. [6]L. Bonora, “Prospects in string field theory,” Facta Univ. Ser. Phys. Chem. Technol. 12 no. 2, (2014) 85–99.DOI
  7. [7]T. Erler, “Four lectures on analytic solutions in open string field theory,” Phys. Rept. 980 (2022) 1–95, arXiv:1912.00521 [hep-th].DOI
  8. [8]I. Y. Aref’eva, “String field theory: From high energy to cosmology,” Theor. Math. Phys. 163 no. 3, (2010) 697–704.DOI
  9. [9]I. Kishimoto and T. Takahashi, “Vacuum structure around identity-based solutions,” Prog. Theor. Phys. 122 no. 2, (2009) 385–399, arXiv:0904.1095 [hep-th].DOI
  10. [10]S. Zeze, “Tachyon potential in KBc subalgebra,” Prog. Theor. Phys. 124 no. 4, (2010) 567–580, arXiv:1004.4351 [hep-th].DOI
  11. [11]J. Hristov, “Open String Renormalization Group Flow as a Field Theory,” Fortsch. Phys. 72 no. 11, (2024) 2300266.DOI
  12. [12]J. Hristov, “Comments on the RG-Flow in Open String Field Theory,” Fortsch. Phys. 73 no. 12, (2025) e70045.DOI
  13. [13]T. Erler, “The Identity String Field and the Sliver Frame Level Expansion,” JHEP 11 (2012) 150, arXiv:1208.6287 [hep-th].DOI
  14. [14]M. Murata and M. Schnabl, “On Multibrane Solutions in Open String Field Theory,” Prog. Theor. Phys. Suppl. 188 (2011) 50–55, arXiv:1103.1382 [hep-th].DOI
  15. [15]I. Gahramanov, “On Analytical Solutions in Witten’s Cubic Open String Field Theory,” arXiv:2412.20981 [hep-th]. in Turkish.
  16. [16]E. A. Arroyo, “Generating Erler–Schnabl-type solution for the tachyon vacuum in cubic superstring field theory,” J. Phys. A 43 no. 44, (2010) 445403, arXiv:1004.3030 [hep-th].DOI
  17. [17]L. Bonora and D. D. Tolla, “Comments on lump solutions in SFT,” Eur. Phys. J. C 76 no. 4, (2016) 203, arXiv:1412.0936 [hep-th].arxiv.org/abs/1412.0936
  18. [18]Y. Okawa, “Comments on Schnabl’s analytic solution for tachyon condensation in Witten’s open string field theory,” JHEP 04 (2006) 055, arXiv:hep-th/0603159.DOI
  19. [19]T. Masuda, T. Noumi, and D. Takahashi, “Constraints on a class of classical solutions in open string field theory,” JHEP 10 (2012) 113, arXiv:1207.6220 [hep-th].DOI
  20. [20]I. Ellwood, “The closed string tadpole in open string field theory,” JHEP 08 (2008) 063, arXiv:0804.1131 [hep-th].DOI
  21. [21]L. Bonora and S. Giaccari, “Generalized states in SFT,” Eur. Phys. J. C 73 (2013) 2644, arXiv:1304.2159 [hep-th].DOI
  22. [22]H. Hata and T. Kojita, “Winding Number in String Field Theory,” JHEP 01 (2012) 088, arXiv:1111.2389 [hep-th].DOI
  23. [23]H. Hata and T. Kojita, “Singularities in K-space and Multi-brane Solutions in Cubic String Field Theory,” JHEP 02 (2013) 065, arXiv:1209.4406 [hep-th].DOI
  24. [24]H. Hata, D. Takeda, and J. Yoshinaka, “Generating string field theory solutions with matter operators from KBc algebra,” PTEP 2022 no. 9, (2022) 093B09, arXiv:2205.14599 [hep-th].arxiv.org/abs/2205.14599

Paper details

vlnc:W_qqp8gkpq
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Claude, ChatGPT

View statement
Looking for related papers…
Contents