2
which can be seen in various physical situations. Historically such a detachment has been
first discussed in detail in the context of spin glasses [3]. In this case the control parameter
was identified with the eigenvalue of the matrix which centers the Gaussian ensemble with
the shifted mean. More recently, this phenomenon was considered in the probability theory
for the behavior of eigenvalues of some covariance matrix [2]. The control parameter is the
deviation of a single eigenvalue of the covariance matrix from the unity. Later more general
situation with several control parameters was discussed as well. Exactly at the transition
point the new universality class has been identified which distribution which now is known
as the “Baik-Ben Arous-P´ech´e” (BBP) distribution. Similar to the Tracy-Widom (TW)
distribution which is given by the solution to the Painleve II equation with the specific
asymptotics, the BBP distribution is described by the pair of differential equations which
involve the solution to Painleve II equation with the peculiar monodromy [4].
The shifted mean Gaussian ensemble has been generalized to the chiral case in [5] and the
position of the separated eigenvalue as a function of the order parameter has been derived
analytically. Later on more eigenvalue detachment phase transitions were found in such
physical problems as: last passage percolation [6]; Asymmetric Simple Exclusion Process
(ASEP) with particular spiked initial conditions [7, 8]; q-version of Totally Asymmetric
Simple Exclusion Process (q-TASEP) with slow particles [9]; percolation in 2D [10]; spin
glass-paramagnetic transition in the mean field approximation [11]. In all these cases the
nature of the phenomena is one and the same: first one zooms the spectral edge where the
TW distribution emerges, and then the perturbation is introduced.
In our study we focus our attention not at the distribution of spectral fluctuations, rather
at the distribution of paths’ fluctuations in space-time. The fermionic nature of eigenvalues
in the matrix model can be mapped onto the vicious walkers problem. So, the detachment of
one eigenvalue from the bulk of the spectrum gets mapped onto the emergence of the outliers
in the Brownian ensemble. The dynamics of the Brownian motion of large number of walkers
with a few outliers has been also discussed in the mathematical literature in [12] where the
transition from the Gaussian to TW fluctuations has been derived analytically. Instead
of looking at detached eigenvalues, the same pattern can be modelled in the ensemble of
individual random walkers nearby the extended defect. In the polymer language, the BBP
transition occurs when a (1+1)D random walk is located exactly at the threshold of the
formation of a bound state with some type of defect, or attractive boundary. As shown in
examples below, the interaction of a random path with a defect can be introduced either
explicitly of effectively.
The example of an explicit interaction is provided in the work [13], where the BBP-
like transition emerges when a part of a polymer trajectory gets localized on an extended
defect at some critical value of coupling between the polymer and the defect. Mapping
of the polymer problem onto the matrix model in the context of the BBP transition has
been discussed in [13]. Conceptually the model under consideration is as following. Take