string theory


String theory


physics, mathematical physics, philosophy of physics

Surveys, textbooks and lecture notes

theory (physics), model (physics)

experiment, measurement, computable physics


Quantum field theory

Higher spin geometry

Elliptic cohomology



String theory is a theory in fundamental physics. Certainly it is, mathematically, a structure that contains in various limits a plethora of quantum field theories. Its interest for experimental high energy physics lies in the hypothesis that it provides a theory of everything in the sense of fundamental physics, but the jury on that is still out.

Below we indicate the basic idea and provide pointers to further details. See also the string theory FAQ.

Conceptually: String perturbation theory from QFT worldline formalism

Perturbative string theory is something at least close to a categorification of the following description of perturbative quantum field theory in terms of sums over Feynman diagrams.

Recall that in quantum field theory one approach to make sense of the path integral is the perturbation series expansion, which interprets the path integral for the scattering amplitudes (the S-matrix) as a certain sum over graphs of certain numbers assigned to each graph.

The graphs are called Feynman diagrams, the numbers assigned to them are called (renormalized) scattering amplitudes and the sum over graphs of (renormalized) amplitudes is the perturbation series or S-matrix.

The amplitude assigned to a single graph with nn external edges is interpreted as the amplitude for nn “quanta” or “particles” of the fields in question to interact in the way indicated by the graph.

Crucial for the motivation of the idea of string theory is the observation that this (renormalized) amplitude assigned to a graph is itself the correlator of a 1-dimensional quantum field theory on that graph: the “worldline quantum field theory” describing the (relativistic) quantum mechanics of these particles. This is usually a sigma-model with parameter space the given graph and target space the spacetime on which the fields live for which the perturbation series computes the path integral.

When made explicit this is called the worldline formalism for computing the quantum field perturbation series. (See there for more details.)

The premise of perturbative string theory is to replace the perturbation series over correlators of a 1-d QFT over graphs by a sum of correlators of a 2-dQFT over 2-dimensional surfaces – called worldsheets and hence produce an S-matrix this way. Again in simple cases this 2d QFT is a sigma-model whose target is the spacetime in which one computes interactions.

In analogy to the previous case, one thinks of the amplitude assigned this way to a surface as the amplitude for the boundary arcs – the strings – to interact in the way given by the surface.

Some of the motivations for considering this replacement of graphs by surfaces have been the following:

  • the 2-d correlators are better behaved in that they don’t have to be renormalized. The “counterterms” appearing in renormalization of ordinary QFT can be identified with contributions to the correlators that come from the linear extension of the strings (see the above reference for more on this);

  • there are fewer choices involved: a Feynman graph is really a decorated graph with the decoration from some more or less arbitrary index set, describing the nature of the particles associated with a given edge and the nature of the interaction associated with a given vertex. In the sum over surfaces there is no extra decoration (except on the boundary of the surfaces) and one finds that instead a single string diagram (a 2d QFT correlator for a given surface) encodes already a sum over (infinitely) many particle species decorations and all possible interaction decorations for them;

  • while there are fewer choices to be made by hand, it turns out that the effective particle content that does arise automatically from this prescription happens to be structurally of the kind one would hope for: the massless effective particles described by the string perturbation series happen to be gauge bosons, fermions charged under them, and, notably, gravitons. This is structurally exactly the Yang-Mills theory input of the standard model of particle physics combined with perturbative quantum gravity that one would hope to see.

These aspects have motivated the impression that the string perturbation series might be considerably closer to the true formalism of fundamental physics than ordinary perturbative quantum field theory. This impression is however offset by the following problems:

  • while the worldsheet 2d QFT whose correlators are summed over surfaces are themselves much easier to handle than the full target space quantum physics they are used to encode, a fully complete and rigorous theory of 2d QFT is available only in simple special cases.

  • In particular, even though there are fewer arbitrary choices involved in the string perturbation series as compared to the ordinary Feynman perturbation series, one crucial choice still present is that of this worldsheet 2d QFT. By the above, every choice of worldsheet QFT (called a choice of “vacuum”) corresponds to a choice of effective target space geometry (to be thought of as the one that the perturbation series computes the quantum perturbations about) and particle content (see 2-spectral triple for more on that). One would therefore like to understand the space of all worldsheet QFTs whose effective target space geometry and particle content is close to the one experimentally observed. After many years of rather naïve approaches to handle or not to handle this, it has more recently at least come to the general attention that there is something to be better understood here.

  • More fundamentally, already the role of the original perturbation series in quantum field theory is actually not fully understood. Its main success is the observation that truncating or resumming the perturbation series in a more or less ad hoc way, it does yield values that very well describe a plethora of real world measurements. One imagines that there is a non-perturbative definition of quantum field theory such that in certain well-defined circumstances the perturbation series does yield an approximation to it and is a posteriori justified. If so, there should be an analogous nonperturbative definition of string theory. There is a large ratio of speculations as to what that might be over solid results about it.


While therefore the premise of perturbative string theory is conceptually suggestive for various reasons, there is to date no connection to experimental phenomenology (apart from the fact that conceptual insights into string theory have helped analyze quantum field theoretic data, see at string theory results applied elsewhere). As a result much of the substantial outcome of string theory research is more in mathematical physics (if done well, at least), exploring the general theory space of quantum field theories and their UV-completions, than in realistic model building (though there is no lack of trying), where it remains very speculative. This has led to public or semi-public debates about the value of string theory for actual physics. See at criticism of string theory for pointers.

Scales and (no) parameters

  • string tension? T=1/(2πα)T = 1/(2\pi \alpha')

  • string scale l s=αl_s = \sqrt{\alpha'}

  • string coupling constant g s=e λg_s = e^\lambda

    (= radius of M-theory compactification circle (Witten 95))

  • coupling constant of gravity: κg s(α) 6=g sl s 12\kappa \propto g_s (\alpha')^6 = g_s l_s^{12}

  • Newton's constant (before compactification) G N=κ 2/8πG_N = \kappa^2/8\pi

  • Planck length l P=(8πG N) 1/24=κ 1/12l sg s 1/12l_P = (8 \pi G_N)^{1/24} = \kappa^{1/12} \propto l_s g_s^{1/12}

  • Planck mass M p=l p 1M_p = l_p^{-1}

(see e.g. arXiv:0908.0333)

Critical string theories and quantum anomalies

The action functional for the string-sigma model in general has a quantum anomaly of both kinds:

  1. For both the bosonic string and the superstring the corresponding Polyakov action has a gauge anomaly for the conformal symmetry, depending on the dimension dd of target space, and on the strength of the dilaton background field. For vanishing dilaton field this anomaly vanishes exactly for d=26d = 26 for the bosonic model, and in d=10d = 10 for the superstring.

    For target spaces of these dimensions one speaks of critical string theory. In as far as string theory is expected to have relevance for physics at all, it is usually expected to be in this critical dimension. But also noncritical string models can and have been considered.

  2. Apart from the gauge anomaly, the action functional of the string-sigma-model also in general has an anomalous action functional , for two reasons:

    1. The higher holonomy of the higher background gauge fields is in general not a function, but a section of a line bundle;

    2. The fermionic path integral over the worldsheet-spinors of the superstring produces as section of a Pfaffian line bundle.

    In order for the action functional to be well-defined, the tensor product of these different anomaly line bundles over the bosonic configuration space must have trivial class (as bundles with connection, even). This gives rise to various further anomaly cancellation conditions:

    1. For the heterotic string (necessarily closed) the anomaly cancellation condition is known as the Green-Schwarz mechanism : it says that the background fields of gravity and B-field must organize to a twisted differential string structure whose twist is given by the background Yang-Mills field.

    2. For the open type II string the condition is known as the Freed-Witten anomaly cancellation condition: it says that the restriction of the B-field to any D-brane must consistute the twist of a twisted spin^c structure on the brane.

      A more detailed analysis of these type II anomalies is in (DFMI) and (DFMII).

      See also Diaconescu-Moore-Witten anomaly.



Critical string models

Extended objects

Table of branes appearing in supergravity/string theory (for classification see at brane scan).

branein supergravitycharged under gauge fieldhas worldvolume theory
black branesupergravityhigher gauge fieldSCFT
D-branetype IIRR-fieldsuper Yang-Mills theory
(D=2n)(D = 2n)type IIA\,\,
D0-brane\,\,BFSS matrix model
D4-brane\,\,D=5 super Yang-Mills theory with Khovanov homology observables
D6-brane\,\,D=7 super Yang-Mills theory
(D=2n+1)(D = 2n+1)type IIB\,\,
D1-brane\,\,2d CFT with BH entropy
D3-brane\,\,N=4 D=4 super Yang-Mills theory
(D25-brane)(bosonic string theory)
NS-branetype I, II, heteroticcircle n-connection\,
string\,B2-field2d SCFT
NS5-brane\,B6-fieldlittle string theory
D-brane for topological string\,
M-brane11D SuGra/M-theorycircle n-connection\,
M2-brane\,C3-fieldABJM theory, BLG model
M5-brane\,C6-field6d (2,0)-superconformal QFT
M9-brane/O9-planeheterotic string theory
topological M2-branetopological M-theoryC3-field on G2-manifold
topological M5-brane\,C6-field on G2-manifold
solitons on M5-brane6d (2,0)-superconformal QFT
self-dual stringself-dual B-field
3-brane in 6d

Scattering amplitudes

Elliptic genera, elliptic cohomology and tmftmf

A properly developed theory of elliptic cohomology is likely to shed some light on what string theory really means. (Witten 87, very last sentence)

The large volume limit of the partition function of the superstring on a given target spacetime is an elliptic genus of that manifold (Witten 87), the Witten genus (see there for more).

Since the Witten genus in turn is the decategorification of the string orientation of tmf, this suggests that tmf-generalized (Eilenberg-Steenrod) cohomology classifies full string theories, in refinement of how the classification of D-brane charge (just the boundary conditions for open strings) is given by K-theory.

A non-trivial conistency check of this idea is announced in (Nikolaus 14).

Topological strings

String phenomenology

String theory results applied elsewhere

Beyond the speculative hypothetized role of string theory as a theory behind observed particle physics, the theory has shed light on many aspects of quantum field theory, both on the conceptual structure of quantum field theory as such as well as on concrete theories and their concrete properties. Some of these string theory results enter crucially in computations that are used to interpret particle physics experiments such as the LHC.

For more see


Survey and introduction

A large body of references is organized at the

For another list of literature see the entry

A quick survey of the big picture as of 2016 is in

  • Ashoke Sen, What is string theory?, talk at YITP50, Stony Brook 2016 (pdf)

A useful survey of the status of string theory as a theory of quantum gravity is in

Some reflections on the mathematical physics involved in string theory are in

An article summarizing information about cohomological models for aspects of string theory and listing plenty of useful further references is

A book trying to summarize the state of the art of capturing mathematical structures fundamental to the definition of perturbative string theory is

Technical details


A key article for the idea that string theory provides a framework for realistic (chiral) grand unified theories combined with quantum gravity was

  • Edward Witten, Some properties of O(32)O(32) superstrings, Phys. Lett. B, volume 149, December 1984 (pdf)

based on the insights of Green-Schwarz anomaly cancellation.

Elliptic genera, elliptic homology and tmf

The partition function of the superstring was understood to be an elliptic genus (the Witten genus) in

  • Edward Witten, Elliptic Genera And Quantum Field Theory , Commun.Math.Phys. 109 525 (1987) (Euclid)

A nontrivial consistency check on the suggestion that this means that string backgrounds are classified in tmf is given in

Quantum anomalies

Discussion of type II quantum anomalies is in

Discussion of superstring perturbation theory is in

Fields medal (and other) work related to string theory

Pure mathematics work which is closely related string theory and was awared with a Fields medal includes the following.

Richard Borcherds, 1998

Maxim Kontsevich, 1998

Edward Witten, 1990

Grigori Perelman, 2006

In (Madsen 07) it says with respect to the proof (Madsen-Weiss 02) of the Mumford conjecture:

These tools [[ used in the proof ]] are all rather old, known for at least twenty years, and one may wonder why they have not before been put to use in connection with the Riemann moduli space. Maybe we lacked the inspiration that comes from the renewed interaction with physics, exemplified in conformal field theories.

See also


Revised on December 1, 2016 12:24:04 by Urs Schreiber (