57 Page 4 of 20 Eur. Phys. J. C (2019) 79 :57
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0
0.2
0.4
0.6
0.8
1.0
x
1
Fig. 2 Convex hull of the projected set of vector exponents
In the following sub-sections we will demonstrate a new
way of solving the problem of finding regions based on the
singularities of Feynman integral in Alpha representation.
2.3 Determination of the regions using the analytic
structure of the propagator
The different regions, where a particular mass or kinemati-
cal scale becomes important can be linked to the underlying
singularities of the Feynman integral. In the following, we
will introduce the main concepts and motivate the ideas that
will lead to the development of the final algorithm. We first
give an overview of the singularities that are of interest for
our problem, followed by a review of the basic understand-
ing of particle thresholds as pinched singularities
in momen-
tum space. This interpretation is well understood and can be
mathematically expressed using a set of equations called the
Landau equations. Since expansions in the neighborhood of
the singular surfaces give us the leading order behavior of
Feynman amplitudes, we perform similar expansions in the
Alpha-parameter space in carefully chosen neighborhoods of
the singular points. This requires us to use techniques from
the field of power geometry. We then motivate the use of
Gröbner basis for the identification of all neighborhoods of
the singular points.
2.3.1 Singularities and threshold processes
Understanding the analytic structure of the amplitude is cru-
cial to identifying the different regions. The poles in the inte-
grand of the amplitude for a given process are functions of
kinematical invariants, loop momenta etc. Therefore, when
these parameters vary, the poles in the integration domain
move. In the case of isolated singularities, it is always pos-
sible to deform the contour of integration to avoid these sin-
gular points. However, sometimes, the poles migrate so as to
pinch the contour of integration (pinch singularities) or move
to one of the end point of the integration (end point singular-
Fig. 3 Types of singularities: P1 is a simple pole, P2 and P3 are
Pinched Singularities
and P4 is an End Point Singularity. While the
contour between the points A and B can be deformed so as to avoid
the simple pole P1, the same is not true for the pinch and end-point
singularities
ities) as illustrated in Fig. 3. In such cases, these singularities
cannot be avoided by contour deformations.
The condition for a point to be one of these unavoidable
singular points is the usual condition for establishing a singu-
larity for a polynomial. For an arbitrary polynomial g({x
i
})
that appears in the denominator of the Feynman integral, the
point x
i
is singular point iff
g({x
i
}) = 0,
∂g
∂x
i
= 0. (9)
Therefore, at these unavoidable singular points, hereby
referred to as just singular points, the integrand diverges. We
will now adopt the approach of Coleman and Norton [25],
also discussed in [26], to explain the connection between the
singularities and physical events.
Consider a general Feynman amplitude in the Alpha rep-
resentation
I =
i
d
d
k
i
j
dα
j
δ
⎛
⎝
j
α
j
− 1
⎞
⎠
f ({q
j
}) D
−n
(10)
123