-
Because we are going to find both
$ \alpha' $ and loop corrected solutions, we write the full perturbative structure of the closed string effective action,$ \begin{aligned}[b] I = & \int {\rm d}^{d+1}x\sqrt{-g}\bigg\{ {\rm e}^{-2\phi}\Big[\Big(R+4(\partial\phi)^{2}- \frac{1}{12}{\cal H}^{2} \Big)+\frac{\text{{\alpha'}}}{4}(R_{\mu\nu\sigma\rho}R^{\mu\nu\sigma\rho}+\cdots)+{\cal O}(\alpha'^{2})\big]\\ & + \Big[(c_{R}^{1}R+c_{\phi}^{1}(\partial\phi)^{2}+c_{{\cal H}}^{1} {\cal H}^{2})+\alpha'(c_{\alpha'R}^{1}R_{\mu\nu\sigma\rho}R^{\mu\nu\sigma\rho}+\cdots)+{\cal O}(\alpha'^{2})\Big]\\ & + {\rm e}^{2\phi}\Big[(c_{R}^{2}R+c_{\phi}^{2}(\partial\phi)^{2}+c_{{\cal H}}^{2} {\cal H}^{2})+\alpha'(c_{\alpha'R}^{2}R_{\mu\nu\sigma\rho}R^{\mu\nu\sigma\rho}+\cdots)+{\cal O}(\alpha'^{2})\Big] + \cdots\bigg\}. \end{aligned} $ (1) This action contains three massless fields, the metric
$ g_{\mu\nu} $ , dilaton ϕ, and antisymmetric field$ b_{\mu\nu} $ , whose field strength is$ {\cal H}_{\mu\nu\rho}=3\partial_{[\mu}b_{\nu\rho]} $ . We set$ b_{\mu\nu}=0 $ in this study. All$ c_{[\cdots]}^{i} $ are unknown up to now. We first consider the$ \alpha' $ corrected solutions. To this end, we focus on the loop tree level, that is, the first line of the above action. In the FLRW background,$ \begin{equation} {\rm d} s^{2}=-{\rm d}t^{2}+a^{2}\left(t\right)\delta_{ij}{\rm d}x^{i}{\rm d}x^{j}. \end{equation} $
(2) The loop tree level action with all orders in the
$ \alpha^{\prime} $ corrections is given by Hohm and Zwiebach in Refs. [17, 18]:$ \begin{aligned}[b] I_{\alpha'} = & \int {\rm d}^{D}x\sqrt{-g} {\rm e}^{-2\phi}\Big(R+4\left(\partial\phi\right)^{2}\\&+\frac{1}{4}\alpha^{\prime}\left(R^{\mu\nu\rho\sigma}R_{\mu\nu\rho\sigma}+\ldots\right)+\alpha^{\prime2}(\ldots)+\ldots\Big), \end{aligned} $
(3) $ = \int {\rm d}t{\rm e}^{-\Phi}\left(-\dot{\Phi}^{2}+\sum\limits_{k=1}^{\infty}\left(\alpha^{\prime}\right)^{k-1}c_{k}\mathrm{tr}\left(\dot{\mathcal{S}}^{2k}\right)\right), $
(4) where the notation
$ \mathcal{S} $ is defined as$ \begin{equation} \mathcal{S}=\left(\begin{array}{cc} 0 & a^{2}\left(t\right)\\ a^{-2}\left(t\right) & 0 \end{array}\right). \end{equation} $
(5) In the action (4), we can only determine the coefficients
$ c_{1}=-\dfrac{1}{8} $ and$ c_{2}=\dfrac{1}{64} $ for bosonic string theory through the one-loop and two-loop beta functions of the non-linear sigma model, and$ c_{k\geq3} $ are undetermined constants. We define$ \begin{aligned}[b] H\left(t\right) = & \frac{\dot{a}\left(t\right)}{a\left(t\right)}, \\ f\left(H\right) = & d\sum\limits_{k=1}^{\infty}\left(-\alpha^{\prime}\right)^{k-1}2^{2\left(k+1\right)}kc_{k}H^{2k-1} \\ =&-2dH-2d\alpha^{\prime}H^{3}+\mathcal{O}\left(\alpha^{\prime3}\right), \\ g\left(H\right) = & d\sum\limits_{k=1}^{\infty}\left(-\alpha^{\prime}\right)^{k-1}2^{2k+1}\left(2k-1\right)c_{k}H^{2k} \\ =& -dH^{2}-\frac{3}{2}d\alpha^{\prime}H^{4}+\mathcal{O}\left(\alpha^{\prime}\right), \end{aligned} $
(6) where
$ H\left(t\right) $ is the Hubble parameter. Note that$ g'(H)=Hf'(H),\quad{\rm and}\quad g(H)=Hf(H)-\int_{0}^{H}f(x){\rm d}x, $
where
$f'(H)\equiv\dfrac{{\rm d}}{{\rm d}H}f(H)$ . The action (4) is simplified to the Hohm-Zwiebach action$ I_{HZ} = \int {\rm d}t{\rm e}^{-\Phi}\left(-\dot{\Phi}^{2}+g(H)-Hf(H)\right), $
(7) and after variation, the EOM of Hohm-Zwiebach action (7) is
$ \begin{aligned}[b] \ddot{\Phi}+\frac{1}{2}Hf\left(H\right) = 0, \quad \frac{{\rm d}}{{\rm d}t}\left({\rm e}^{-\Phi}f\left(H\right)\right) = 0, \quad \dot{\Phi}^{2}+g\left(H\right) = 0. \end{aligned} $
(8) In the perturbative region
$ t\to \infty $ , the perturbative solution can be obtained iteratively using (6),$ \begin{aligned}[b] H\left(t\right) = & \frac{\sqrt{2}}{\sqrt{\alpha^{\prime}}}\left[\frac{t_{0}}{t}-160c_{2}\frac{t_{0}^{3}}{t^{3}}+\frac{256\left(770c_{2}^{2}+19c_{3}\right)}{3}\frac{t_{0}^{5}}{t^{5}}\right.\\ & \left.-\frac{2048\left(88232c_{2}^{3}+4644c_{3}c_{2}+41c_{4}\right)}{5}\frac{t_{0}^{7}}{t^{7}}+\mathcal{O}\left(\frac{t_{0}^{9}}{t^{9}}\right)\right], \\ \Phi\left(t\right) = & -\frac{1}{2}\log\left(\beta^2\frac{t^{2}}{t_{0}^{2}}\right)-32c_{2}\frac{t_{0}^{2}}{t^{2}}+\frac{256\left(44c_{2}^{2}+c_{3}\right)}{3}\frac{t_{0}^{4}}{t^{4}} \\ & -\frac{2048\left(6976c_{2}^{3}+352c_{3}c_{2}+3c_{4}\right)}{15}\frac{t_{0}^{6}}{t^{6}}+\mathcal{O}\left(\frac{t_{0}^{8}}{t^{8}}\right), \end{aligned} $
(9) with
$ \begin{aligned}[b] f\left(H\left(t\right)\right) = & -2dH-128c_{2}d\alpha^{\prime}H^{3}+768c_{3}d\alpha^{\prime2}H^{5}-4096c_{4}d \alpha^{\prime3}H^{7}+\mathcal{O}\left(\alpha^{\prime4}H^{9}\right), \\ = & \frac{\sqrt{d}}{t_{0}}\left[-\frac{2t_{0}}{t}+64c_{2}\frac{t_{0}^{3}}{t^{3}}-\frac{512\left(50c_{2}^{2}+c_{3}\right)}{3}\frac{t_{0}^{5}}{t^{5}}\right. \left.+\frac{4096\left(2632c_{2}^{3}+124c_{3}c_{2}+c_{4}\right)}{5} \frac{t_{0}^{7}}{t^{7}}+\mathcal{O}\left(\frac{t_{0}^{9}}{t^{9}}\right)\right],\\ g\left(H\left(t\right)\right) = & -dH^{2}-96c_{2}d\alpha^{\prime}H^{4} +640c_{3}d\alpha^{\prime2}H^{6}-3584c_{4}d\alpha^{\prime3} H^{8}+\mathcal{O}\left(\alpha^{\prime4}H^{10}\right),\\ = & \frac{1}{t_{0}^{2}}\left[-\frac{t_{0}^{2}}{t^{2}}+128c_{2}\frac{t_{0}^{4}}{t^{4}}-\frac{2048\left(50c_{2}^{2}+c_{3}\right)}{3}\frac{t_{0}^{6}}{t^{6}}\right. \left.+\frac{8192\left(24448c_{2}^{3}+1136c_{3}c_{2}+9c_{4}\right)}{15}\frac{t_{0}^{8}}{t^{8}}+\mathcal{O}\left(\frac{t_{0}^{10}}{t^{10}}\right)\right], \end{aligned} $ where β is an integration constant,
$ t_{0}\equiv\dfrac{\sqrt{\alpha'}}{\sqrt{2d}} $ , and the universal$ c_{1}=-\dfrac{1}{8} $ is used. This solution is singular at the non-perturbative region$ t=0 $ . From scale factor duality,$ H(t)\to-H(t) $ ,$ \Phi(t)\to\Phi(t) $ ,$f(t)\to -f(t)$ , and$ g(t)\to g(t) $ is also a solution.A non-perturbative non-singular solution of the EOM (8) should meet two conditions: (1) It must match the perturbative solution (9) in the perturbative region
$ \dfrac{\sqrt{\alpha'}}{t}\to 0 $ , and (2) it must be non-singular for any$ \dfrac{\sqrt{\alpha'}}{t} $ . Because$ c_{k\ge3} $ are still unknown up to now, a good solution only needs to match the first two orders in$ \alpha' $ . Such a solution was constructed in Ref. [23]. However, an important question is can we construct a general solution that can easily match all given$ c_{k\le n} $ for some$ n>2 $ ? In other words, a general solution expressed in terms of$ c_{k} $ . In ref. [24], two such solutions were given. In this study, we obtain another simpler solution. Every term in our solution is non-singular. Let us first define a dimensionless parameter:$ \tau \equiv \frac{t}{t_0} = \sqrt{\frac{2d}{\alpha'}}t. $
(10) After extensive calculation, we find a non-perturbative solution of the EOM (8),
$ H(t)= \frac{\sqrt{\dfrac{2}{\alpha' \beta ^2 \lambda _0}} \left({\rm e}^{-\sum _{k=1}^{\infty } \textstyle\frac{\lambda _k}{\tau ^{2 k}+1}} \left(\left(\tau ^2+1\right)^2 \sum _{k=1}^{\infty } \left(\dfrac{8 k^2 \lambda _k \tau ^{4 k-2}}{\left(\tau ^{2 k}+1\right)^3}-\dfrac{2 k (2 k-1) \lambda _k \tau ^{2 k-2}}{\left(\tau ^{2 k}+1\right)^2}\right)+\tau ^2-1\right)\right)}{\left(\tau ^2+1\right)^{3/2}} , $ (11) $ \Phi(t) = \frac{1}{2}\log\frac{\lambda_0}{1+\tau^2} +\sum\limits_{n=1}^\infty \frac{\lambda_n}{1+\tau^{2n}}, $
(12) $ f(H(t))= -2 d \sqrt{\frac{2 \beta ^2 \lambda _0}{\left(\tau ^2+1\right) \alpha '}} {\rm e}^{\sum _{k=1}^{\infty }\textstyle\frac{\lambda _k}{\tau ^{2 k}+1}} = -2d H(t) -2d\alpha' H(t)^3 + {\cal O}(\alpha^{\prime 2}) , $
(13) $ \begin{aligned}[b] g(H(t))=& -\frac{2 d}{\alpha' } \left(\sum\limits _{k=1}^{\infty } \frac{2 k \lambda _k \tau ^{2 k-1}}{\left(\tau ^{2 k}+1\right)^2}+\frac{\tau }{\tau ^2+1}\right)^2\\ =& -d H(t)^2 -\frac{3}{2} d\alpha' H(t)^4 + {\cal O}(\alpha^{\prime 2}). \end{aligned} $
(14) Applying
$ c_1 =-1/8 $ and$ c_2=1/64 $ and expanding this solution in the perturbative region$ t/\sqrt{\alpha'}\to \infty $ to match the perturbative solution (9), we identify$ \begin{aligned}[b] \lambda_0 =& 1/\beta^2,\quad \lambda_1 = 0,\quad \lambda_2= \frac{11+1024c_3}{12},\\ \lambda_3 =& -\frac{4}{15}(13+2816c_3+1536c_4)... \end{aligned} $
(15) It is clear that every single term in the sums of the solution, such as
$ \sum _{n=1}^\infty \dfrac{\lambda_n}{1+\tau^{2n}} $ , is non-singular everywhere. However, note that there could be a small possibility that the summation is not non-singular for fine-tuned parameters,$ \lambda_n, n>0 $ 1 . If choosing$ \lambda_{n\ge 1}=0 $ , the solution reduces to$ \begin{aligned}[b] H(t) = & -\frac{\sqrt{2}}{\sqrt{\alpha'}}\frac{\left(1-\tau^{2}\right)}{\left(1+\tau^{2}\right)^{3/2}}, \\ \Phi(t) = & -\frac{1}{2}\log\beta^2-\frac{1}{2}\log\left(1+\tau^{2}\right),\\ f(t) = & -\frac{2\sqrt{2}d}{\sqrt{\alpha'}}\frac{1}{\sqrt{1+\tau^{2}}}, \\ g(t) = & -\frac{2d}{\alpha'}\frac{\tau^{2}}{\left(1+\tau^{2}\right)^{2}}. \end{aligned} $
(16) This is exactly the solution given in Ref. [23].
-
After constructing the non-perturbative non-singular
$ \alpha' $ corrected solution, we want to find the corresponding loop corrected non-perturbative non-singular solution. We do not know the form and coefficients of the higher loop terms in the full perturbative action (1). For the FLRW background (2), an effective loop corrected action is expressed as$ \begin{aligned}[b] I_{\rm Loop}=&\int {\rm d}^{d+1}x\sqrt{-g}{\rm e}^{-2\phi}\left[R+4\left(\partial_{\mu}\phi\right)^{2} - V\left({\rm e}^{-\Phi\left(x\right)}\right)\right],\\ =& \int {\rm d} t {\rm e}^{-\Phi} \big[ -\dot\Phi + d H^2 -V({\rm e}^{-\Phi})\big], \end{aligned} $
(17) where the
$ O(d,d) $ non-local dilaton is [9, 10]$ \begin{aligned}[b] {\rm e}^{-\Phi\left(t\right)}=&V_{d}\int {\rm d} t'\left|\frac{{\rm d}\left(2\phi\right)}{{\rm d}t'} \right|\sqrt{-g\left(t'\right)}{\rm e}^{-2\phi\left(t'\right)}\delta\left(2\phi\left(t\right)-2\phi\left(t^{\prime}\right)\right)\\=&V_{d}\sqrt{-g\left(t\right)}{\rm e}^{-2\phi\left(t\right)}. \end{aligned} $
(18) The EOM is
$ \begin{aligned}[b] 2\ddot{\Phi}_L -2 dH_L^2 - \frac{\partial V}{\partial\Phi_L} =&0, \\ \dot{\Phi}_L^{2}-dH_L^{2}-V = & 0, \\ \dot{H}_L-H_L\dot{\Phi}_L = & 0, \end{aligned} $
(19) where the subscript L indicates that the quantities belong to the loop corrected theory. For a positive integer n and arbitrary parameters
$ m_n $ and$ \sigma_n $ , a class of solutions of the above EOM was constructed in Refs. [10, 11],$ \begin{aligned}[b] \Phi^{(n)}_L(t) = & \frac{1}{2n}\log\left(\frac{\sigma_n^{2n}}{1+\left(m_n t\right)^{2n}}\right),\\ H^{(n)}_{L} (t) = & \frac{1}{\sqrt d} \frac{m_n}{\sigma_n} {\rm e}^{\Phi^{(n)}_L \left(t\right)}\\=& \frac{m_n}{\sqrt d} \left[\frac{1}{1+\left(m_n t\right)^{2n}}\right]^{1/2n}. \end{aligned} $
(20) This had the potential
$ \begin{equation} V^{(n)}_{L} = \left(\frac{m_n}{\sigma_n}\right)^{2}{\rm e}^{2\Phi^{(n)}_L \left(t\right)}\left[\left(1 - \sigma_n^{-2n} {\rm e}^{2n\Phi^{(n)}_L \left(t\right)}\right)^{\frac{2n-1}{n}}-1\right]. \end{equation} $
(21) As argued in Refs. [10, 11], because the factor
${\rm e}^\Phi$ roughly represents the coupling constant, the integer n effectively represents the loop number. The potential indicates the non-perturbative contributions by the nth loop.In Ref. [24], a map was constructed between the
$ \alpha' $ and loop corrected theories,$ \begin{array}{*{20}{l}} H_{L} & \leftrightarrow & f\left(H_{\alpha^{\prime}}\right), \\ -V_{L} &\leftrightarrow & g\left(H_{\alpha^{\prime}}\right)+df\left(H_{\alpha^{\prime}}\right)^{2} \\ \Phi_L &\leftrightarrow & \Phi_{\alpha'} + \Phi_0. \end{array} $
(22) Here,
$ \Phi_0 $ is a constant. Through this map, we can find a loop corrected non-perturbative non-singular solution from our$ \alpha' $ corrected solution (14). Because${\rm e}^{n\Phi_L^{(n)}}$ indicates the contribution from the n-th loop, we use Eq. (20) to express the solution in terms of${\rm e}^{\Phi_L^{(n)}}$ . Then, after some tedious calculation, we obtain the loop corrected solution,$ \begin{aligned}[b] \Phi_L(t) =& \Phi_L^{(1)} + \sum\limits_{n=1}^\infty {\rm e}^{2n\Phi_L^{(n)}} \\ H_L(t)=& \frac{m_1}{\sigma_1} \exp\left[\Phi_L^{(1)} + \sum\limits_{n=1}^\infty {\rm e}^{2n\Phi_L^{(n)}} \right] \\ V_L(\Phi_L^{(n)}(t))=& \left( \dot \Phi_L^{(1)} + \sum\limits_{n=1}^\infty 2n \dot \Phi_L^{(n)} {\rm e}^{2n\Phi_L^{(n)}} \right)^2 \\&- \frac{d m_1^2}{\sigma_1^2} \exp\left[2\Phi_L^{(1)} + 2 \sum\limits_{n=1}^\infty {\rm e}^{2n\Phi_L^{(n)}} \right], \end{aligned} $
(23) where we set
$ m_n = \sqrt{2d/\alpha'} $ , and$ \sigma_i $ are free parameters to be determined using coefficients calculated from the effective low energy action. Obviously, this solution is non-singular around the non-perturbative region$ t\sim 0 $ .
Two non-perturbative α' or loop corrected string cosmological solutions
- Received Date: 2023-05-30
- Available Online: 2023-09-15
Abstract: In this paper, we present two non-perturbative string cosmological solutions without curvature singularities for the bosonic gravi-dilaton system. These solutions are general in that they can straightforwardly match the perturbative solution to arbitrarily high orders in the perturbative region. The first solution includes non-perturbative