String Theory

From the original dual resonance model to M-theory, D-branes, holography, and the landscape. What string theory actually predicts, what it doesn't, and why physicists still work on it.

1. Why string theory?

Every successful quantum field theory relies on a crucial assumption: the fundamental objects are point particles — zero-dimensional objects occupying a single spacetime point at any instant. For the strong, weak, and electromagnetic forces this works beautifully. The Standard Model built from this assumption has passed every experimental test for fifty years. The trouble starts when you try to add quantum gravity.

Gravity is described by general relativity, and its force carrier is the graviton — a massless spin-2 boson. If you write down a quantum field theory of gravitons as point particles and compute loop diagrams (corrections due to virtual graviton exchanges), you find that each loop integral diverges worse than the previous one. A theory that requires infinitely many independent counter-terms to absorb infinities is called non-renormalizable: perturbation theory breaks down completely at the Planck scale $E_{Pl} = \sqrt{\hbar c^5 / G} \approx 1.2 \times 10^{19}$ GeV.

The ultraviolet divergence problem in quantum gravity is not an accident of bad bookkeeping — it is a structural feature of point particles. When two point particles collide, their interaction happens at a single point of zero extent. The contribution to the amplitude from that collision grows without bound as you integrate over arbitrarily short distances. There is simply no geometric mechanism to cut off the integral.

The string-theoretic fix is conceptually simple: replace point particles with one-dimensional objects — strings — of a characteristic length $\ell_s$ (the string length). When two strings interact, the interaction is spread out over the string's extent. Instead of a pointlike vertex, you have a smooth worldsheet merging and splitting. Loop integrals receive a natural damping at momenta above $1/\ell_s$, making the UV behavior far better. In fact, it was proved in the 1980s that superstring theory is UV finite at every order in perturbation theory — the infinities that plague point-particle quantum gravity simply do not arise.

Why is the spin-2 mode compelling? In any consistent relativistic quantum theory, a massless spin-2 particle mediating long-range interactions is forced by the spin-2 uniqueness theorem to obey the same equations as a graviton. So string theory does not merely accommodate gravity — it requires it. Every consistent string theory contains gravity. This is remarkable: you start with a theory of 1D oscillating objects, and gravity falls out automatically.

2. Bosonic string theory

The simplest string theory contains only bosons — no fermions, no supersymmetry. It is not physically realistic (it has a tachyon in its spectrum), but it cleanly illustrates the mathematical structure that all string theories share.

The worldsheet

A point particle sweeps out a worldline in spacetime — a one-dimensional curve parameterized by proper time $\tau$. A string sweeps out a worldsheet — a two-dimensional surface parameterized by $(\tau, \sigma)$, where $\tau$ is a timelike coordinate and $\sigma \in [0, \pi]$ is a spacelike coordinate running along the string. The embedding of the worldsheet in $D$-dimensional spacetime is given by $D$ functions $X^\mu(\tau, \sigma)$, $\mu = 0, 1, \ldots, D-1$.

The classical action that minimizes the area of the worldsheet is the Nambu-Goto action:

The Nambu-Goto action has a square root, which makes quantization awkward. The equivalent but more tractable form is the Polyakov action, which introduces an auxiliary worldsheet metric $h_{ab}$:

Mode expansion and the Virasoro algebra

In conformal gauge ($h_{ab} \propto \eta_{ab}$), the Polyakov equations of motion become free 2D wave equations $(\partial_\tau^2 - \partial_\sigma^2) X^\mu = 0$, solved by a Fourier mode expansion. A closed string (periodic in $\sigma$) has both left-movers and right-movers:

Each sector contains an infinite tower of oscillator modes $\alpha^\mu_n$ (right) and $\tilde\alpha^\mu_n$ (left), $n \in \mathbb{Z}$. Canonical quantization promotes these to operators satisfying $[\alpha^\mu_m, \alpha^\nu_n] = m \delta_{m+n,0} \eta^{\mu\nu}$. The physical state conditions — that the stress-energy tensor of the worldsheet vanishes — generate the Virasoro algebra $[L_m, L_n] = (m-n)L_{m+n} + \frac{c}{12}(m^3-m)\delta_{m+n,0}$.

Critical dimension

When you normal-order the Virasoro constraints, you find a vacuum energy contribution from each spacetime dimension. The Virasoro algebra has a central charge $c$ equal to the number of bosonic worldsheet fields, which equals the number of spacetime dimensions $D$. Requiring that the quantum constraints be consistent — specifically that ghost states decouple from all physical observables — forces $c = 26$ for the bosonic string, i.e., $D = 26$.

The mass spectrum includes a ground state (lowest-mode oscillation) with $m^2 = -1/\alpha'$ — a tachyon, signaling the instability of the bosonic string vacuum. The next level contains a massless spin-2 state — the graviton — plus massless scalars. Higher levels give an infinite tower of massive particles with masses $m^2 \sim n/\alpha'$. For open strings (with Neumann or Dirichlet boundary conditions at the endpoints), the spectrum is halved: only right-movers or only left-movers, and the gauge group resides on the endpoints.

3. Superstrings and the critical dimension D=10

The bosonic string has two fatal problems: the tachyon and the absence of fermions. Both are cured by adding worldsheet supersymmetry: for each bosonic coordinate $X^\mu$ you add a fermionic partner $\psi^\mu$ — a two-component Majorana spinor on the worldsheet.

NS and R sectors

Fermions on a worldsheet can be periodic or antiperiodic. This gives two sectors:

• Neveu-Schwarz (NS) sector: $\psi^\mu$ antiperiodic around the string. The ground state is a spacetime boson. The tachyon lives here but is projected out.
• Ramond (R) sector: $\psi^\mu$ periodic. The ground state is a spacetime spinor — the first appearance of fermions.

The GSO projection (Gliozzi, Scherk, Olive, 1977) keeps only states of definite worldsheet fermion parity. The result is: the tachyon is projected out; the spectrum is tachyon-free; and the NS and R sectors pair up into a spacetime supersymmetric spectrum. The critical dimension drops from 26 to $D = 10$.

The five superstring theories

Depending on how left-movers and right-movers are handled and what boundary conditions are applied, there are five consistent superstring theories in 10 dimensions:

The heterotic theories are hybrid constructions where left-movers are bosonic (living in 26 dimensions) and right-movers are supersymmetric (living in 10 dimensions). The extra 16 dimensions of the left-movers are compactified on one of two special lattices, producing either $SO(32)$ or $E_8 \times E_8$ gauge symmetry. The $E_8 \times E_8$ heterotic string was for a long time considered the most promising for phenomenology because $E_8 \supset SU(5) \supset SU(3) \times SU(2) \times U(1)$ — the Standard Model gauge group can be embedded inside it.

4. M-theory and dualities

By the early 1990s, physicists had five superstring theories and one 11-dimensional supergravity. That felt like five too many. In 1995, Edward Witten gave a celebrated talk proposing that all five superstring theories — plus 11D supergravity — are different limiting descriptions of a single 11-dimensional theory he called M-theory. The web of dualities connecting them is one of the most striking structures in modern theoretical physics.

T-duality

T-duality is a perturbative duality visible even at weak string coupling. Compactify one spatial dimension on a circle of radius $R$. A string can have momentum modes (quantized as $n/R$) and winding modes (wrapping the circle $w$ times, with energy $wR/\alpha'$). The spectrum is invariant under the exchange $R \leftrightarrow \alpha'/R$ combined with swapping $n \leftrightarrow w$. This maps Type IIA on a small circle to Type IIB on a large circle, and vice versa. At the self-dual radius $R = \sqrt{\alpha'}$, the spectrum develops an enhanced gauge symmetry.

S-duality

S-duality maps a theory at string coupling $g_s$ to another theory at coupling $1/g_s$. This exchanges the weakly coupled description with the strongly coupled one. Type I with coupling $g_s$ is dual to the $SO(32)$ heterotic string at coupling $1/g_s$. Type IIB is self-dual under S-duality. S-duality implies that heavy non-perturbative objects (D-branes) in one theory become light perturbative strings in another.

The 11D limit

Type IIA string theory at strong coupling $g_s \to \infty$ grows an extra circular dimension of radius $R_{11} = g_s^{2/3} \ell_s$. The tower of D0-branes becomes the Kaluza-Klein momentum modes of an 11-dimensional theory — 11D supergravity. The fundamental string becomes the M2-brane wrapped on the emerging circle. This is why the web of dualities converges on an 11-dimensional theory: the string coupling is not a free parameter but a geometric modulus — the radius of an extra dimension.

M-theory also contains M2-branes (membranes, 2 spatial dimensions) and M5-branes (5 spatial dimensions). These are the fundamental extended objects of M-theory, and all the branes of the five string theories can be understood as M-branes wrapped or reduced on the extra dimensions.

5. D-branes

In 1995, Joseph Polchinski showed that certain extended objects called D-branes are required for the consistency of string dualities — they are not optional add-ons but necessary components of the theory. A Dp-brane is an object with $p$ spatial dimensions: a D0-brane is a point, a D1-brane is a string, a D2-brane is a membrane, and so on.

The tension (energy per unit volume) of a Dp-brane scales as $T_{Dp} \propto 1/(g_s \ell_s^{p+1})$. At weak coupling $g_s \ll 1$, D-branes are very heavy — much heavier than fundamental strings (which have tension $\sim 1/\ell_s^2$, independent of $g_s$). This is why they do not appear in perturbation theory. At strong coupling, D-branes become light and dominate.

Gauge theory from D-branes

The open string spectrum on a single Dp-brane contains a massless $U(1)$ gauge field $A_\mu$ (the photon) living on the brane's $(p+1)$-dimensional worldvolume, plus $9-p$ real scalar fields describing the brane's transverse position in spacetime. Now stack $N$ coincident Dp-branes. An open string can begin on any of the $N$ branes and end on any of the $N$ branes, giving $N^2$ possible string types labelled by a pair of indices $i, j \in \{1, \ldots, N\}$ — the Chan-Paton factors. The massless vector fields then form an $N \times N$ matrix — a non-abelian $U(N)$ gauge field. At low energies, the D-brane worldvolume theory is $U(N)$ Yang-Mills gauge theory.

This is a profound result: non-abelian gauge theories live on D-branes. It is the starting point for the AdS/CFT correspondence (Section 7) and for string phenomenology attempts to derive the Standard Model from intersecting D-branes wrapping cycles of a compact Calabi-Yau manifold.

6. Compactification and the landscape

String theory requires 10 spacetime dimensions. We observe 4. The standard resolution is compactification: the 6 extra spatial dimensions are curled up on a compact manifold too small to see at accessible energies. The shape and topology of this manifold determine the effective 4-dimensional physics — the particle content, gauge group, and coupling constants.

Calabi-Yau manifolds

Preserving $\mathcal{N}=1$ supersymmetry in 4D (needed for the lightest superpartners to remain light and protect the Higgs mass) requires the internal 6-manifold to be a Calabi-Yau manifold — a complex three-fold with $SU(3)$ holonomy and vanishing first Chern class $c_1 = 0$. These are characterized by two Hodge numbers $h^{1,1}$ and $h^{2,1}$. The total number of massless scalar fields (moduli) in 4D is $h^{1,1} + h^{2,1}$: the Kähler moduli describe the sizes of the internal two-cycles, and the complex structure moduli describe their shapes. The Euler characteristic of the Calabi-Yau determines the number of chiral generations in the 4D theory.

The moduli problem and flux compactification

A compactification with unfixed moduli is phenomenologically disastrous: massless scalars couple to matter with gravitational strength, mediating unobserved fifth forces. The moduli must be stabilized — given masses by some mechanism that generates a potential.

Flux compactification turns on quantized background fluxes — higher-dimensional generalizations of magnetic flux threading the compact cycles. Each choice of flux quantum numbers (integers) gives a different vacuum, with different values of the moduli, different effective cosmological constant, and different low-energy physics. The KKLT mechanism (Kachru, Kallosh, Linde, Trivedi, 2003) stabilizes all moduli in a supersymmetric anti-de Sitter vacuum and then lifts it to a de Sitter (positive cosmological constant) vacuum using anti-D3-branes.

The landscape

This enormous number has a profound implication: if the landscape is real, our universe is just one particular vacuum, selected not by any theoretical principle but perhaps by anthropic reasoning: we live in a vacuum compatible with the existence of observers. This was anticipated by Steven Weinberg's 1987 prediction that the cosmological constant cannot be much larger than the observed value $\Lambda \sim 10^{-122} M_{Pl}^4$, because larger values would have prevented galaxies from forming. The prediction was confirmed in 1998 when supernovae measurements revealed the accelerating expansion of the universe.

The swampland

The swampland program asks: which effective field theories can be consistently embedded in quantum gravity? Swampland conjectures (Vafa and collaborators, 2005–present) propose constraints on any consistent theory of quantum gravity. Two prominent examples: the de Sitter conjecture suggests that stable de Sitter vacua (with positive $\Lambda$) cannot exist in string theory, in tension with KKLT; the distance conjecture says that as you move far in field space, an infinite tower of light states appears, preventing you from taking a simple effective field theory limit. These conjectures offer a new approach to falsifiability: they make predictions about what kinds of physics (dark energy models, inflationary models) are or are not consistent with quantum gravity.

7. AdS/CFT and holography

In 1997, Juan Maldacena published a paper proposing an exact equivalence between a theory of quantum gravity and a quantum field theory with no gravity at all — living in one fewer dimension. This is the AdS/CFT correspondence, also called the Maldacena duality or gauge/gravity duality. It is the most important discovery in string theory since the second superstring revolution.

The dictionary

The precise identification of parameters between the two sides is:

Bulk–boundary dictionary

Every local operator $\mathcal{O}(x)$ in the boundary CFT corresponds to a bulk field $\phi$ with the appropriate quantum numbers. The AdS radial direction $z$ encodes the renormalization group scale of the boundary theory — physics near the boundary ($z \to 0$) is UV physics; physics deep in the bulk is IR physics. A black hole in AdS at temperature $T_H$ is dual to the boundary CFT in a thermal state at the same temperature. The area of the bulk black hole horizon gives the entropy of the thermal boundary state via the Bekenstein-Hawking formula.

Applications: holographic QCD and condensed matter

AdS/CFT allows computation of strongly coupled gauge theory observables using classical gravity — turning an intractable field theory calculation into a tractable general relativity problem. Key results:

• Quark-gluon plasma viscosity. The ratio of shear viscosity to entropy density for any theory with an Einstein gravity dual is $\eta/s = 1/(4\pi)$ (in units of $\hbar/k_B$). This is a universal lower bound conjectured to hold for all fluids. The RHIC heavy-ion experiments at Brookhaven measure $\eta/s$ of the quark-gluon plasma and find values close to $1/(4\pi)$, remarkably consistent with the holographic prediction for a strongly coupled fluid.
• Holographic superconductors. A charged black hole in AdS develops a scalar hair below a critical temperature, dual to a superconducting phase transition in the boundary theory. Holographic models reproduce the qualitative (and sometimes quantitative) features of strongly coupled superconductors.
• Entanglement entropy. The Ryu-Takayanagi formula gives the entanglement entropy of a region $A$ in the CFT in terms of a minimal surface in the bulk.

8. Strings and black holes

String theory has had its most concrete quantitative success in explaining the entropy of black holes — not as a macroscopic thermodynamic approximation but as an exact count of microstates.

Bekenstein-Hawking entropy from D-brane counting

In 1996, Andrew Strominger and Cumrun Vafa computed the entropy of a specific class of extremal (zero-temperature) charged black holes in 5 dimensions by counting the number of D-brane configurations with the same mass and charges. They found:

where $N_{\text{micro}}$ is the number of D-brane bound states — an exact match, not just an order-of-magnitude agreement. This was the first derivation of black hole entropy from a microscopic theory and one of the most celebrated results in string theory. It works because the extremal black hole and the corresponding D-brane system are related by supersymmetry and protected from quantum corrections.

The information paradox

Stephen Hawking showed in 1974 that black holes emit thermal radiation with a temperature $T_H = \hbar c^3 / (8\pi G_N M k_B)$. As the black hole radiates it loses mass and eventually evaporates. The radiation appears perfectly thermal — with no dependence on what fell in. If true, information is destroyed: you cannot reconstruct the initial quantum state from the final Hawking radiation. This violates unitarity, a cornerstone of quantum mechanics.

AdS/CFT resolves the paradox in principle: the black hole in AdS is dual to a thermal state of the boundary CFT, and the boundary CFT evolves unitarily. Therefore the evaporation must be unitary — information is preserved. But how exactly information escapes remained unclear for twenty years.

The Page curve and the island formula

The Page curve describes how, in a unitary theory, the entanglement entropy of Hawking radiation should first increase, peak at the Page time (when about half the black hole entropy has been radiated), and then decrease to zero as the black hole fully evaporates. Classical Hawking calculations give entropy that increases monotonically — violating unitarity.

In 2019-2020, a breakthrough emerged from the Ryu-Takayanagi formula and its generalizations. The entropy of the radiation, computed using the quantum extremal surface formula, receives contributions from island regions — patches of spacetime inside or near the horizon that contribute to the entropy budget of the radiation. When islands are included, the entropy follows the Page curve exactly. The island formula:

where $I$ is an island region and $R$ is the radiation region, reproduces the Page curve and suggests that black hole evaporation is indeed unitary.

ER = EPR

In 2013, Maldacena and Susskind proposed the ER=EPR conjecture: two black holes connected by a wormhole (Einstein-Rosen bridge) are equivalent to two maximally entangled black holes (an EPR pair). More broadly: any quantum entanglement between two systems corresponds to a geometric connection in spacetime. Entanglement builds geometry. Spacetime connectivity is a consequence of quantum correlations. This is the deepest insight from the AdS/CFT era: geometry and entanglement may be two languages for the same underlying reality.

9. Experimental signatures and the status question

What string theory predicts experimentally, and what it does not, is one of the most important questions in theoretical physics — and also one of the most contested.

What string theory predicts (and doesn't)

String theory as currently formulated does not make unique, falsifiable predictions about the Standard Model parameters, particle masses, or coupling constants. The landscape of $\sim 10^{500}$ vacua means that almost any observed values can be "accommodated" somewhere in the landscape — but this is not the same as predicting them. The main experimental handles that string theory's low-energy limits suggest are:

• Supersymmetric partners. All superstring theories require spacetime supersymmetry. If broken at low enough scales, the LHC would produce superpartners — squarks, sleptons, gauginos. Despite extensive searches, no superpartners have been found at the LHC up to masses of $\sim 2$ TeV (as of 2026). Naturalness arguments suggested they should appear below $\sim 1$ TeV. This is the primary experimental problem for low-energy string phenomenology.
• Extra dimensions. If large extra dimensions exist (ADD or Randall-Sundrum scenarios), they could be detected through missing-energy signatures at colliders, modifications to gravity at sub-millimeter scales, or Kaluza-Klein graviton production. None found.
• Cosmic strings. Fundamental strings stretched to cosmological scales could leave imprints on the CMB or produce gravitational wave backgrounds. Constraints are becoming tight.

What string theory HAS given us

String theory's practical contributions to physics and mathematics are substantial, even setting aside the question of whether it describes nature:

• Holographic quark-gluon plasma. The $\eta/s = 1/(4\pi)$ lower bound from AdS/CFT is consistent with RHIC and LHC heavy-ion measurements of the quark-gluon plasma — a real, experimentally tested number that came from string theory.
• Scattering amplitude methods. Twistor string theory (Witten, 2003) inspired the BCFW recursion relations and the discovery of the amplituhedron (Arkani-Hamed and Trnka, 2013) — a geometric structure encoding scattering amplitudes in $\mathcal{N}=4$ SYM. These methods dramatically speed up collider calculations that are now used at the LHC.
• Black hole entropy. The Strominger-Vafa calculation remains the only microscopically precise derivation of black hole entropy from first principles.
• Mirror symmetry. The mathematical duality discovered in string theory between pairs of Calabi-Yau manifolds solved open conjectures in algebraic geometry about counting rational curves on Calabi-Yau threefolds.
• Information paradox resolution. The island formula and the Page curve calculation are string-theory-derived insights about quantum gravity that go beyond what we knew from classical GR or naive QFT.

Honest assessment

Interactive: Calabi-Yau cross-section

A Calabi-Yau manifold is a 6-real-dimensional (3-complex-dimensional) space. It cannot be visualized directly, but projections of its structure can be. The animated figure below shows a 2D cross-section of a quintic Calabi-Yau — a complex hypersurface defined by a degree-5 polynomial in $\mathbb{CP}^4$. The rippling surface detail represents the non-trivial curvature structure (Ricci-flat Kähler metric) that makes Calabi-Yau manifolds special. At every point of ordinary 4D spacetime, a manifold of this shape is attached — too small to see, but determining the physics.

Cheat sheet

See also