# Off-shell strings
The off-shell string formalism is an approach to string theory that allows strings to propagate on backgrounds that do not satisfy the Einstein equation, i.e., off-shell backgrounds. Some advantages of this formulation are:
- one can derive target space effective action directly from string worldsheet, without having to obtain EOM first;
- one can compute $n$-point correlators without having to take the insertions to infinity;
- allows e.g. a conical singularity which is an off-shell configuration so that the entropy derivation can be performed.
## Notations
- $I_g$: string action at genus $g$ correction ($I_g^{\text{eff}}=-Z_g$)
- $I_0$: classical string (EFT) action
- $I_{g\ge1}$: quantum effective actions
- $K_{g,n}$: CFT correlator for inserting $n$ vertex operators onto the genus $g$ worldsheet
- $K_0=\sum\limits_{n}K_{0,n}$
- string amplitude
- $g\ge1$: $A_{\mathrm{g}, n}=\int[\mathrm{d} \tau] \frac{K_{\mathrm{g}, n}}{\operatorname{Vol}(\mathrm{KG})}$
- $g=0$: $A_{0, n}=\frac{\partial}{\partial \log \epsilon} K_{0, n}$ (T1)
- sum over $n$: $Z_0=\sum_n A_{0, n}=\frac{\partial}{\partial \log \epsilon} K_0$
- $\epsilon$: UV cutoff on the worldsheet
- $n$: number of operator insertions
- $\mathcal{M}_{0,n}$: moduli space
- $\mathcal{P M}_{0, n}$: pre-moduli space
- $\mathcal{M}_{0,n}=\mathcal{P M}_{0, n} / \mathrm{SL}(2, \mathbb{C})$
- $\mathcal{P M}_{0, n}^{(\epsilon)}$: regulated pre-moduli space
- $\mathcal{M}_{0, n}^{(\epsilon)}:=\mathcal{P M}_{0, n}^{(\epsilon)} / \mathrm{SL}(2, \mathbb{C})$: regulated moduli space
## Two serious problems and resolutions
- theory depends on an arbitrary choice of the Weyl frame, $\omega$
- not a problem: changing $\omega$ can be absorbed into field redefinition of target space fields
- corresponds to renormalisation of worldsheet QFT
- non-compact conformal Killing group for the sphere diagram
- Tseytlin does not deal with it by fixing three points, which is not generalisable to off-shell case
- the effect of Tseytlin's sphere prescription is to mod out by the gauge orbits (*not* dividing by the volume of the gauge group)
## III. Defining the partition function off-shell
- choice of Weyl frame
- $g_{a b}=e^{2 \omega} \gamma_{a b}$
- off-shell gauge fixing
- define a function $\tilde{\mathcal{L}}[X, g]$ that is the unique Weyl-invariant function which agrees with $\mathcal{L}_{\mathrm{QFT}}[X, \gamma, \omega]$ when $g_{a b}=e^{2 \omega} \gamma_{a b}$
- then can repeat Faddeev-Popov gauge fixing for this $\tilde {\mathcal{L}}$
- field redefinition
- under a small Weyl rescaling
- $\frac{\delta}{\delta \omega(z)} Z[\omega]=\langle\langle T(z)\rangle\rangle_\omega=\sum_a \beta^a\left\langle\left\langle\mathcal{O}_a\right\rangle\right\rangle_\omega$ $\propto$ EOM
- perturbatively two nearby Weyl frames give equivalent results up to a field redefinition
- conformal perturbation theory
- any QFT can be regarded as a CFT with a gas of vertex operator insertions
- $\mathcal{L}=\mathcal{L}_{\mathrm{CFT}}+\sum_a \epsilon^{2\left(h_a-1\right)} \phi^a \mathcal{O}_a$
- QFT amplitude:
- $K_{\mathrm{g}}(\tau)=\sum_{n=0}^{\infty} K_{\mathrm{g}, n}(\tau)$
- at order $\phi^n$: $K_{\mathrm{g}, n}=\sum_{a_1 \ldots a_n}\left(\prod_n \epsilon^{2\left(h_a-1\right)} \phi^a\right) \frac{1}{n !}\left\langle\left\langle V_{a_1} \ldots V_{a_n}\right\rangle\right\rangle_{\mathrm{CFT}}$
- momentum basis: $\mathcal{O}_a \sim \exp \left(i P_\mu X^\mu\right) \mathcal{O}_{\aleph}$
- if $\mathcal{O}_a$ is an $(1,1)$ primary $\to$ physical state with $P^2+M^2=0$
- string amplitude
- $g>0$:
- $A_{\mathrm{g}, n}=\int[\mathrm{d} \tau] \frac{K_{\mathrm{g}, n}}{\operatorname{Vol}(\mathrm{KG})}$
- n.b. we need to *choose* $\hat g$ s.t. CKG acts as a normal KG (doable only when $g>0$)
- $g=0$:
- T1: $A_{0, n}=\frac{\partial}{\partial \log \epsilon} K_{0, n}$
- so $Z_0=\frac{\partial}{\partial \log \epsilon} K_0$
- wrongly implies that there is a tree-level tadpole associated with the tachyon
- $I_0^{\mathbf{T} \mathbf{1}}=-\frac{\partial K_0}{\partial \log \epsilon}=\sum_a \beta^a \frac{\partial K_0}{\partial \phi^a}$
- T2: $Z_0=\left(\frac{\partial}{\partial \log \epsilon}+\frac{1}{2} \frac{\partial^2}{(\partial \log \epsilon)^2}\right) K_0$
- $I_0^{\mathbf{T} 2}=I_0^{\mathbf{T} 1}+\frac{1}{2}\left(\sum_{a, b} \beta^a \beta^b \frac{\partial^2 K_0}{\partial \phi^a \partial \phi^b}+\beta^b \frac{\partial \beta^a}{\partial \phi^b} \frac{\partial K_0}{\partial \phi^a}\right)$
- n.b. taking derivative in $\log \epsilon$ is related to the $\beta$ functions
- n.b. the action vanishes on-shell: $\beta=0$
- on-shell amplitude
- $S_{\mathrm{g}, n}=A_{\mathrm{g}, n} \prod_n \delta\left(P_a^2+M_a^2\right)$, i.e., forcing external legs on-shell
- two-point amplitude
- ...
- string tadpoles
- $A_{0,1}\ne0$
## IV. Renormalisation and propagating strings
- divergences (separating degenerations): a single string propagator becomes long
- momentum-dependent ones: when the propagator satisfies mass-shell
- momentum-independent ones: when vertex operators approach each other
- regulated propagator
- ...
- locality and cutoff
- $\epsilon\to0$: WS theory exactly local; target space gives S-matrix
- $\epsilon\sim1$: WS non-local; target space confusing (discretised worldsheet?)
- perturbatively in $\epsilon$ (WS perturbatively local)
- $\log \epsilon^{-1} \sim 1$: $I^\text{eff}_0$ local perturbatively in $l_s$ (local over scales $\Delta X\gg l_s$)
- $\log \epsilon^{-1} \gg 1$: $I^\text{eff}_0$ non-local even at scales $\Delta X\gg l_s$
## V. Obtaining the tree-level S-matrix
- takeaway: T1 or T2 is equivalent to quotienting by the gauge orbits of $\mathrm{SL}(2,\mathbb{C})$
- the gauge orbits of $\mathrm{SL}(2,\mathbb{C})$
- refer to each element of $\mathcal{M}_{0, n}^{(\epsilon)}:=\mathcal{P M}_{0, n}^{(\epsilon)} / \mathrm{SL}(2, \mathbb{C})$ as $\Omega$
- i.e., each $\Omega$ is a gauge orbit of $\mathrm{SL}(2,\mathbb{C})$
- the nontrivial contribution to $\Omega$ comes from $\frac{\mathrm{SL}(2, \mathbb{C})}{\mathrm{SU}(2)}=H_3$, where $\mathrm{SU}(2)$ is the compact rotation group
- regulated volume in $H_3$ per solid angle:
- $\int_0^{\log (a / \epsilon)+O\left(\epsilon^2\right)} \mathrm{d} \lambda \sinh ^2(\lambda)=\frac{a^2}{8} \epsilon^{-2}+\frac{1}{2} \log (\epsilon)+b+O\left(\epsilon^2\right)$
- n.b. for any ball of radius $r$ in $H^n$:
- $\operatorname{Vol}(B(r))=\operatorname{Vol}\left(S^{n-1}\right) \int_0^r \sinh ^{n-1}(t) d t$
- integrate up to $\log \epsilon$: these are orbits for which each pair of points are at least $\epsilon$ apart on the sphere
- n.b. for $n=2$, $\operatorname{Vol}(\Omega)=\infty$: will not consider in this section; for $n\ge3$, the (regulated) volume is finite:
- ![[AhmadainWall2022a_fig5.png]]
- then T2 gives $\lim _{\epsilon \rightarrow 0}\left(\frac{\partial}{\partial \log \epsilon}+\frac{1}{2} \frac{\partial^2}{(\partial \log \epsilon)^2}\right) \operatorname{Vol}(\Omega) \propto 1$
- generic momenta: fixing 3 points
- when $n=3$ (6 real parameters), only one orbit $\mathrm{SL}(2,\mathbb{C})$ orbit
- $K_{0, n}=\operatorname{Vol}\left(\Omega_3\right) F_n$
- for all $n\ge3$
- $F_n$ is the tree level amplitude defined by fixing 3 points
- to show this, start with $n$ insertions and act with $\mathrm{SL}(2,\mathbb{C})$ on *all* $n$ insertions
- $A_{0,n}=F_n$
- by applying T2 to both sides above
- this shows that Tseytlin's prescription is the same as fixing 3 points
## VI. Obtaining the classical EOM
- goal: show that the tree-level string action $I_0$ obtains the correct EOM to all orders in perturbation theory in $n$
- plan: show at orders $n=1,2$; then show it for marginal primaries when $n\ge0$, which is sufficient, as $n=1,2$ dominate in non-marginal cases
## Refs
- originals
- [[1985#Fradkin, Tseytlin (PRB158)]]
- [[1985#Fradkin, Tseytlin (NPB261)]]
- [[1985#Fradkin, Tseytlin (PRB160)]]
- Tseytlin's proposal
- [[1988#Tseytlin]]
- [[1989#Tseytlin]]
- careful study [[2022#Ahmadain, Wall (a)]]
- entropy calculation [[2022#Ahmadain, Wall (b)]]
- [[1988#Tseytlin]]: some useful results needed for [[1994#Susskind, Uglum]]