Past exam of the mathematics course of the University of Cambridge / 2003 / iii / Paper 50 / 5

Use units $k_B=\hslash =c=1$. In critical light-cone gauge in string theory, each chiral sector of the closed bosonic string has 24 transverse string oscillator species. Their generating function isEach factor sums the independent occupation numbers of the string oscillators at level $m$. For $q=e^{-t}$ with $t\downarrow 0$, the modular transformation of the Dedekind eta function givesExtract the coefficient by a contour integral and apply the saddle-point approximation to the exponent $(N-1)t+4{\pi }^2/t$. The saddle is $t_{\ast }=2\pi /\sqrt{N-1}$, with exponent $4\pi \sqrt{N-1}$ and second derivative $(N-1{)}^{3/2}/\pi$. The factor $t^{12}$ gives $(N-1{)}^{-6}$, and the Gaussian width gives $(N-1{)}^{-3/4}/\sqrt{2}$. HenceFor a noncompact closed string, closed-string level matching imposes equal left and right levels, but does not correlate their transverse polarizations. Therefore the large-level degeneracy of a closed bosonic string isThe mass relation is ${\alpha }^{\prime }{M}^2=4(N-1)$. Thus $8\pi \sqrt{N-1}={\beta }_{H}M$ withDistinguish degeneracy per discrete level from a smoothed density per unit rest mass. Since $dN/dM={\alpha }^{\prime }M/2$, the latter isThe exponential growth comes from the many ways of distributing string oscillator energy along a string; it is much faster than a fixed finite collection of particle species. The corresponding high-energy entropy is $S(E)={\beta }_{H}E+O(\log E)$, so its microcanonical temperature tends to $T_H$. Additional energy can go into very long strings rather than raising the temperature substantially.

The Boltzmann suppression competes with this exponential density. For the string oscillator rest-mass sum alone the large-$M$ tail is $\int dM{M}^{-26}{e}^{-(\beta -{\beta }_H)M}$. It converges for $\beta >{\beta }_H$ and fails for $\beta <{\beta }_H$. To include translational states in 25 noncompact spatial dimensions, take the one-string partition function per unit spatial volume. At large $M$,This follows by expanding the energy as $M+p^2/(2M)+\cdots$ and doing a Gaussian momentum integral. Its high-mass contribution is consequently proportional toThis integral is finite at $\beta ={\beta }_H$, although nonanalytic there: its first twelve derivatives in $\beta$ are finite, while the thirteenth diverges. The endpoint expansion contains a term proportional to $(\beta -{\beta }_H{)}^{25/2}$. Thus one must not assert that the canonical partition function necessarily diverges exactly at $T_H$; the polynomial prefactor and the spatial spectrum matter. This is the finite canonical partition function at the bosonic Hagedorn threshold distinction. The multi-string ideal gas has the same leading Hagedorn threshold, since the higher occupation-cycle terms probe $\ell \beta$, $\ell \ge 2$, and are nonsingular at ${\beta }_H$.

There is also a spacetime diagnosis. Euclidean temperature makes time a circle of radius $R_{\beta }=\beta /(2\pi )$. A string oscillator ground state winding once around that circle, with zero momentum number, hasIt becomes massless at ${\beta }_H$ and tachyonic for $T>T_H$. This thermal winding instability of a closed bosonic string signals failure of the perturbative low-temperature gas and motivates a transition to a different high-energy phase, often described in terms of long-string or winding-mode condensation. Free state counting fixes the threshold and its singular behavior, not the interacting phase's endpoint or the order of a transition.

Finally, the ordinary bosonic tachyon is already an infrared instability at zero temperature. The above canonical discussion is a formal analysis of the high-level spectrum with that separate infrared problem regulated. It does not establish a stable thermal vacuum of the unmodified bosonic theory; an actual equilibrium transition requires an appropriate stable setting and control of interactions.

The first massless level of a closed string contains a spin-two field, an antisymmetric two-form and a scalar: the graviton, Kalb–Ramond field and dilaton. The polarization redundancies found in Question 2 become, in position space, the linearized transformations $\delta G_{\mu \nu }={\partial }_{\mu }{\xi }_{\nu }+{\partial }_{\nu }{\xi }_{\mu }$ and $\delta B_{\mu \nu }={\partial }_{\mu }{\Lambda }_{\nu }-{\partial }_{\nu }{\Lambda }_{\mu }$. Thus the massless spin-two field already carries the gauge symmetry of linearized general relativity.

To describe slowly varying backgrounds, promote these polarizations to fields of the embedding coordinates. The Euclidean string nonlinear sigma model isHere ${ϵ}^{ab}$ is the antisymmetric coordinate density. Quantum consistency requires worldsheet Weyl anomaly cancellation, not merely the classical background equations. At first order in ${\alpha }^{\prime }$ the bosonic sigma-model Weyl anomaly coefficients givewhere $H=dB$ is the Kalb-Ramond field strength. The scalar anomaly condition, combined with the trace of the metric condition in the critical dimension, givesThese are the Euler–Lagrange equations of the leading string-frame massless effective actionFor example, variation of $B$ gives the divergence equation above. Varying $\Phi$ and integrating its kinetic variation by parts gives $R+4{\nabla }^2\Phi -4(\nabla \Phi {)}^2-H^2/12=0$. Combining this with the metric variation yields the Ricci equation. Thus the string's quantum conformal consistency reproduces target-space gravity coupled to matter. Equivalently, expanding string scattering amplitudes at small momenta gives the same action: massless exchange poles are reproduced by these fields, while analytic terms from massive exchange become local higher-derivative interactions.

The apparent dilaton-dependent coefficient of $R$ is a frame choice. Put $\varphi =\Phi -{\Phi }_0$ and, in dimension $D>2$, define the Einstein-frame metricTo see the cancellation explicitly, write $G=e^{2\omega }{g}^{(E)}$, $\omega =2\varphi /(D-2)$. ThenThe Laplacian term is a boundary term. Combining the last term with the original positive dilaton kinetic term produces $-4({\nabla }_E\varphi {)}^2/(D-2)$. The three inverse metrics in $H^2$ give its remaining exponential factor. HenceFor the uncompactified bosonic theory $D=26$, these coefficients are $-1/6$ and $e^{-\varphi /3}$. The Ricci term is now the Einstein-Hilbert action; its metric variation gives the Einstein field equations with stress-energy tensor from the scalar and two-form. The leading low-energy theory is general relativity coupled to the massless string fields.

The approximation requires characteristic momenta $E\sqrt{{\alpha }^{\prime }}\ll 1$ and small curvature in string units. Massive string modes have masses of order $1/\sqrt{{\alpha }^{\prime }}$ and can be integrated out, generating corrections with extra powers of ${\alpha }^{\prime }{E}^2$ or ${\alpha }^{\prime }R$. The string genus expansion supplies independent loop corrections controlled by $g_s=e^{{\Phi }_0}$; at weak coupling tree-level gravity dominates. Dimensional analysis gives ${\kappa }_{26}^2\propto g_s^2({\alpha }^{\prime }{)}^{12}$.

Compactification supplies lower-dimensional gravity and additional matter: metric and two-form components give gauge fields, internal geometry gives scalar moduli, and light modes of Kaluza-Klein theory must be kept if their masses lie below the chosen cutoff. Open strings add gauge fields, and supersymmetric string theories add spacetime fermions. Their particular matter spectrum depends on the compactification and projection; universal massless spin-two interactions retain the gravitational form. Finally, the bosonic massless-sector action is a formal truncation around a tachyonically unstable vacuum. It does not remove that instability. Tachyon-free superstring backgrounds provide stable applications of this low-energy gravitational interpretation.