Computational High Energy Physics

Cengiz Pehlevan, Daniel Doro Ferrante and Gerald S. Guralnik

1. Complex Langevin Equation 2. Three-Dimensional Gravity and Symmetry Breaking 3. Topology Change, Ramificaitons and Langlands Duality

1.1 Motivation 2.1 Chern-Simons Theory 3.1 Symmetry Breaking and Topology Change
The solution space of Schwinger-Dyson equations led us to complexified path integrals. The con- Chern-Simons (CS) theory has a curious history: It was discovered in the context of anomalies in Using a conformal transformation of the metric, we can recast the Lagrangian of any theory solely
ventionel numerical methods to study path integrals, i.e. Monte Carlo methods, work on Euclidean the 70’s. It was only by the mid-80’s that it was realized that ordinary Einstein gravity in (2 + 1)- in terms of a kinetic term:
space path integrals of real actions, i.e. the sign problem. We need new numerical methods for dimensions is a natural example of a CS system. There is an intrinsic connection between CS and
1
complexified path integrals. The new numerical method will be of use in many other applications, the [mathematical] theory of knots and link invariants. The table below shows some of the analogies: L= g(π, π) − V (φ) ;
e.g. simulating QFT in Minkowski space, nuclear physics problems with complex actions, QCD with 2
Knot Theory Chern-Simons Theory ≡ g̃ E (π, π) ;
chemical potential, etc. knots and links Wilson loops
polynomial invariants vev’s of products of Wilson loops where,
We consider complex Langevin equations as a numerical method for this purpose. The crucial insight singular knots operators of singular knots
is the correspondence between the stationary distributions of a complex Langevin equation and the


Vassiliev invariants coefficients of the perturbative series g̃ E = 2 E − V (φ) g ;
solutions of the Schwinger-Dyson equation of the associated QFT. configuration space integral Landau gauge
where E is the total energy of the system. The different phases of this theory are, then, given by
1.2 Langevin Equations and Stochastic Quantization The key to construct the CS form in 3-dim is as follows: the Pontryagin form, P = tr(F ∧ F), the geodesics (γ) of this metric, which can all be found via,
Stochastic quantization using Langevin equations in Euclidean space was first proposed by Parisi and is closed, dP = 0. (F = dA + A ∧ A is the curvature of the Lie algebra-valued connection
1-form A, taken in the adjoint representation. Upon a gauge transformation, F 7→ F0 = g −1 F g,
 
Wu. Consider an n dimensional Euclidean QFT with action S[φ(x)]. Add a new coordinate to dγ dγ
g̃ E , =1;
the fields φ, the fictitious time τ , φ(x) 7→ φ(x, τ ). The fictitious time evolution of φ(x, τ ) where g ∈ g, the Lie algebra of the gauge group G. So using the cyclic property of the trace we
 ds ds 
is governed by a stochastic equation that allows for relaxation to equilibrium, i.e. the Langevin see that the Pontryagin form, P, remains invariant under gauge transformations.) By Poincaré’s
dγ dγ
Lemma, P is locally exact, i.e., P = dQ. Thus, Q = LSC is the CS Lagrangian, found to be ⇒ 2 E − V (γ) g , =1;
equation, ds ds
∂φ(x, τ ) δS Q = LSC = tr(A ∧ dA + 32 A ∧ A ∧ A). s
=− + η(x, τ ) ; , dγ 1
∂τ δφ(x, τ ) ∴ =
;
More than sixty years of frustrated efforts to quantize this theory can explain the immediate attention ds 2 E − V (γ)
where η(x, τ ) is a Gaussian white noise.
drawn by Witten’s observation that gravity in (2 + 1)-dim is an exactly solvable model! This means with the initial condition γ(0) = 0 .
The main assertion of stochastic quantization is that the probability distribution P (φ, τ ) associated that the quantum theory can be completely and explicitly spelled out. This is due to the fact that
(2 + 1)-dim gravity has no propagating degrees of freedom and, therefore, its quantum description Note that the potential V (φ) will depend on some parameters, and it is by analyzing these param-
with the Langevin process relaxes to the Euclidean path integral independent of the initial condition,
R is like that of a system of point particles. It is a particular case of a Topological QFT! eters (i.e., analyzing the geodesic equation in parameter space) that we will find all of the distinct
n
e− d xS[φ(x)] solutions and phases of the theory. For concreteness sake, let us assume [without loss of generality]
lim P (φ, τ ) = Z . 2.2 Witten: Monster Group and Number Theory that V (φ) = φn + an−1 φn−1
τ −→∞ Q+ · · · + a1 φ + a0, i.e., V is a polynomial in its field variables.
n
R
Dφ e− d xS[φ(x)] Witten just introduced the concept that pure 3-dim gravity with negative cosmological constant Therefore, we know that ∆ = i<j (ri − rj )2, the discriminant of this polynomial, will com-
Stochastic quantization is useful in quantization of gauge theories, it enables quantization without may be holomorphic factorized only for certain (discrete) values of the coupling constant. According pletely determine the solutions of the theory: whether ∆ vanishes, or is positive, or is negative, we
gauge-fixing. It provides also a new numerical method for the study of QFT’s. to this theory, the Monster group may be the first in a discrete series of CFTs that are dual to have distinct relations among its solutions, which will affect the solution of the geodesic equation
3-dim gravity; where the partition functions are determined on a hyperelliptic Riemann surface of above for V (γ) = (γ − r1) · · · (γ − rn). In particular, the ramifications at the roots of this
1.3 Complex Langevin Equation any genus. Therefore, a theory that was thought to have no physical degrees of freedom, now seem polynomial are related to the singularities of the Higgs Bundle that can be constructed to this
Can one use the Langevin equation in Minkowski space? Or more generally for path integrals with to have revealed its kinematical variables in another “phase” of the theory (related to BTZ black theory, making its connection to the Geometric Langlands Duality quite apparent.
complex weights? Parisi and Klauder suggested the use of complex Langevin equation for this holes). R
purpose. One still uses the Langevin equation but now the drift term is complex. Because the drift However, this hardly comes as a surprise to us, once all we need to do is realize that the Chern- Furthermore, we note that S = g̃ E (γ, γ) dγ is known as the “energy functional” in Morse
is complex, no precise mathematical analysis can be given (yet). Numerical studies show noncon- Simons Lagrangian is just a non-abelian extension of the scalar φ3 model: that is, all we have to do Theory, a fact that renders our Path Integral a topological object, where the classical solution has
vergence, convergence to the desired complex distributions and convergence to other distributions. is extend φ from being a scalar-valued 0-form to a Lie algebra-valued 1-form, A. Here is how we one given topology and its quantum corrections are “handles” attached to it. This fact realizes the
Why are we interested in this then? If the complex Langevin equation converges, the resulting do it: quantum corrections as a sequence of surgeries between the initial cobordism (given by the clas-
φ3
Z
complex distribution satisfies Schwinger-Dyson equations! sical configuration space) and the final cobordism (given by the quantum configuration space) —
ei 3
+i J φ
dφ
Ai(J ) remember that cobordisms are equivalence relations between manifolds based on their boundaries.
Path Integral: Z [J ] = C Z φ3
≡ ; (1)
Let’s do this in 0 dimensions for simplicity. Applying the rules of Ito calculus to the complex Langevin Ai(0) 3.2 Higgs Bundles, Langlands Duality
ei 3 dφ
equation will give the identity, C
* + φ7→−i ∂J
One of the central features of Higgs bundles is that they have continuous moduli, i.e., they come in
∂ 2F Schwinger-Dyson: φ2 + J = 0 = ⇒ (∂J2 − J ) Z = 0 .

d ∂F ∂S ===== (2) families parameterized by the points of a moduli space. This moduli space is described by the follow-
hF (φ(τ )) i = − ,
dτ ∂φ2 ∂φ ∂φ This shows that a model usually considered non-physical, does indeed have physical solutions. These ing gauge theoretic equations for φ and an SU (n)-connection A compatible with the holomorphic
formulas above, together with the very same contour C, can be used to extend φ from scalar-valued structure of the bundle:
For stationary distributions the LHS is zero. Then substituting F (φ) = φk , the Schwinger-Dyson
identities are produced order by order. This stationary distribution must be a linear combination of to matrix-valued or Lie Algebra-valued — this is a nice result by Varadarajan: the two solutions
to the Schwinger-Dyson equation (2) are Airy functions, Ai(J ) and Bi(J ).
the complex path integral solutions to the Schwinger-Dyson equation, FA + [φ, φ∗] = 0 ;
Z
2.3 Symmetry Breaking, Topology Change and Phase Transitions d00
Z dφ φne−S(φ) Aφ = 0 ;
X Γ In this setting, the connections with Braid and Knot Theory and the Configuration Space [of this
P̃eq (φR) = aΓI P̃ΓI (φR), dφR φn(φR)P̃ΓI (φR) ≡ ZI ; where FA is the curvature of A and d00
theory] seem to jump before our eyes: does having two distinct Generating Functions implies that Aφ is the anti-holomorphic part of the covariant derivative
ΓI < dφ e−S(φ) we shall need two different polynomial invariants, or does it imply that the representation of the of φ. These equations express both the flatness of an SL(n, C)-connection A + φ + φ∗ and the
ΓI
Lie group in question can be dynamically determined (by the two possible contours). Given that harmonicity condition for a metric in the resulting flat bundle.
where ΓI are independent contours that connect the zeros of e−S(φ) on the complex φ plane! the Chern-Simons form is a topological invariant, one can use it in order to measure whether a Thus, the ramifications of V (φ) (Number Theory) are related to this connection (Geometry) and
If one can control the flow to the stationary state, this suggests a new numerical method. topological phase transition has occurred. via the construction above they are also related to the distinct topology of each phase of the theory.