
The resemblance between the methods used in quantummany body physics and in
machine learning has drawn considerable attention. In particular, tensor
networks (TNs) and deep learning architectures bear striking similarities to
the extent that TNs can be used for machine learning. Previous results used
onedimensional TNs in image recognition, showing limited scalability and
flexibilities. In this work, we train twodimensional hierarchical TNs to solve
image recognition problems, using a training algorithm derived from the
multiscale entanglement renormalization ansatz. This approach introduces
mathematical connections among quantum manybody physics, quantum information
theory, and machine learning. While keeping the TN unitary in the training
phase, TN states are defined, which encode classes of images into quantum
manybody states. We study the quantum features of the TN states, including
quantum entanglement and fidelity. We find these quantities could be properties
that characterize the image classes, as well as the machine learning tasks.

(Please refer to arXiv:1810.08050, which has completely different aims but
contains all the main contents of this paper) In this work, we propose to
access the information of criticality and excitations of onedimensional
quantum systems by a matrix product state (MPS) defined in the (imaginary) time
direction. This state, dubbed as time MPS (tMPS), is a boundary state of tensor
network (TN) that represents the groundstate simulation after TrotterSuzuki
decomposition. We show that the tMPS exhibits the structure of the continuous
MPS originally proposed for the field theories. The information of excitations,
e.g., dynamic correlation length and energy gap, can be accurately calculated
from the tMPS. The nonuniversal renormalization of velocity of the excitations
is given by a ratio between the correlations of the ground state and the tMPS.
When the system is at the quantum critical point, the tMPS is found to show the
logarithmic scaling law, where the scaling coefficient gives the central charge
that characterizes the criticality of the theory. Our work also implies that
the spectra of the transfer matrices defined by the ground state and tMPS
provide information of the lowlying masses of the theory in the continuous
limit. This could help to understand the role of the finiteness of the bond
dimension as an infra red regulator. Furthermore, we show from the perspective
of methodology that, the tMPS emerges from a generalized TN abinitio
optimization principle scheme, which unifies the infinite density matrix
renormalization group and the infinite timeevolving block decimation
algorithms.

Searching for simple models that possess nontrivial controlling properties
is one of the central tasks in the field of quantum technologies. In this work,
we construct a quantum spin$1/2$ chain of finite size, termed as controllable
spin wire (CSW), in which we have $\hat{S}^{z} \hat{S}^{z}$ (Ising)
interactions with a transverse field in the bulk, and $\hat{S}^{x} \hat{S}^{z}$
and $\hat{S}^{z} \hat{S}^{z}$ couplings with a canted field on the boundaries.
The Hamiltonians on the boundaries, dubbed as tuning Hamiltonians (TH's), bear
the same form as the effective Hamiltonians emerging in the socalled `quantum
entanglement simulator' that is originally proposed for mimicking infinite
models. We show that tuning the TH's (parametrized by $\alpha$) can trigger
nontrivial controlling of the bulk properties, including the degeneracy of
energy/entanglement spectra, and the response to the magnetic field $h_{bulk}$
in the bulk. A universal point dubbed as $\alpha^s$ emerges. For $\alpha >
\alpha^s$, the groundstate diagram versus $h_{bulk}$ consists of three
`phases', which are Ne\'eL and polarized phases, and an emergent pseudomagnet
phase, distinguished by entanglement and magnetization. For $\alpha <
\alpha^s$, the phase diagram changes completely, with no steplike behaviors to
distinguish phases. Due to its controlling properties and simplicity, the CSW
could potentially serve in future the experiments for developing quantum
devices.

Tensor network (TN), a young mathematical tool of high vitality and great
potential, has been undergoing extremely rapid developments in the last two
decades, gaining tremendous success in condensed matter physics, atomic
physics, quantum information science, statistical physics, and so on. In this
lecture notes, we focus on the contraction algorithms of TN as well as some of
the applications to the simulations of quantum manybody systems. Starting from
basic concepts and definitions, we first explain the relations between TN and
physical problems, including the TN representations of classical partition
functions, quantum manybody states (by matrix product state, tree TN, and
projected entangled pair state), time evolution simulations, etc. These
problems, which are challenging to solve, can be transformed to TN contraction
problems. We present then several paradigm algorithms based on the ideas of the
numerical renormalization group and/or boundary states, including density
matrix renormalization group, timeevolving block decimation,
coarsegraining/corner tensor renormalization group, and several distinguished
variational algorithms. Finally, we revisit the TN approaches from the
perspective of multilinear algebra (also known as tensor algebra or tensor
decompositions) and quantum simulation. Despite the apparent differences in the
ideas and strategies of different TN algorithms, we aim at revealing the
underlying relations and resemblances in order to present a systematic picture
to understand the TN contraction approaches.

It is a fundamental, but still elusive question whether methods based on
quantum mechanics, in particular on quantum entanglement, can be used for
classical information processing and machine learning. Even partial answer to
this question would bring important insights to both fields of both machine
learning and quantum mechanics. In this work, we implement simple numerical
experiments, related to pattern/images classification, in which we represent
the classifiers by quantum matrix product states (MPS). Classical machine
learning algorithm is then applied to these quantum states. We explicitly show
how quantum features (i.e., singlesite and bipartite entanglement) can emerge
in such represented images; entanglement characterizes here the importance of
data, and this information can be practically used to improve the learning
procedures. Thanks to the low demands on the dimensions and number of the
unitary matrices, necessary to construct the MPS, we expect such numerical
experiments could open new paths in classical machine learning, and shed at
same time lights on generic quantum simulations/computations.

We explore the frustrated spin$1/2$ Heisenberg model on the star lattice
with antiferromagnetic (AF) couplings inside each triangle and ferromagnetic
(FM) intertriangle couplings ($J_e<0$), and calculate its magnetic and
thermodynamic properties. We show that the FM couplings do not sabotage the
magnetic disordering of the ground state due to the frustration from the AF
interactions inside each triangle, but trigger a fully gapped
inversionsymmetrybreaking trimerized valence bond crystal (TVBC) with
emergent spin1 degrees of freedom. We discover that with strengthening $J_e$,
the system scales exponentially, either with or without a magnetic field $h$:
the order parameter, the five critical fields that separate the $J_e$$h$
groundstate phase diagram into six phases, and the excitation gap obtained by
lowtemperature specific heat, all depend exponentially on $J_e$. We calculate
the temperature dependence of the specific heat, which can be directly compared
with future experiments.

Due to the presence of strong correlations, theoretical or experimental
investigations of quantum manybody systems belong to the most challenging
tasks in modern physics. Stimulated by tensor networks, we propose a scheme of
constructing the fewbody models that can be easily accessed by theoretical or
experimental means, to accurately capture the groundstate properties of
infinite manybody systems in higher dimensions. The general idea is to embed a
small bulk of the infinite model in an "entanglement bath" so that the
manybody effects can be faithfully mimicked. The approach we propose is
efficient, simple, flexible, signproblemfree, and it directly accesses the
thermodynamic limit. The numerical results of the spin models on honeycomb and
simple cubic lattices show that the groundstate properties including quantum
phase transitions and the critical behaviors are accurately captured by only
$\mathcal{O}(10)$ physical and bath sites. Moreover, since the fewbody
Hamiltonian only contains local interactions among a handful of sites, our work
provides new ways of studying the manybody phenomena in the infinite
stronglycorrelated systems by mimicking them in the fewbody experiments using
cold atoms/ions, or developing novel quantum devices by utilizing the manybody
features.

Characterizing criticality in quantum manybody systems of dimension $\ge 2$
is one of the most important challenges of the contemporary physics. In
principle, there is no generally valid theoretical method that could solve this
problem. In this work, we propose an efficient approach to identify the
criticality of quantum systems in higher dimensions. Departing from the
analysis of the numerical renormalization group flows, we build a general
equivalence between the higherdimensional ground state and a onedimensional
(1D) quantum state defined in the imaginary time direction in terms of the
socalled time matrix product state (tMPS). We show that the criticality of the
targeted model can be faithfully identified by the tMPS, using the mature
scaling schemes of correlation length and entanglement entropy in 1D quantum
theories. We benchmark our proposal with the results obtained for the
Heisenberg antiferromagnet on honeycomb lattice. We demonstrate critical
scaling relation of the tMPS for the gapless case, and a trivial scaling for
the gapped case with spatial anisotropy. The critical scaling behaviors are
insensitive to the system size, suggesting the criticality can be identified in
small systems. Our tMPS scheme for critical scaling shows clearly that the
spin1/2 kagom\'e Heisenberg antiferromagnet has a gapless ground state. More
generally, the present study indicates that the 1D conformal field theories in
imaginary time provide a very useful tool to characterize the criticality of
higher dimensional quantum systems.

We investigate the ground state and finitetemperature properties of the
spin1/2 Heisenberg antiferromagnet on an infinite octakagome lattice by
utilizing stateoftheart tensor networkbased numerical methods. It is shown
that the ground state has a vanishing local magnetization and possesses a
$1/2$magnetization plateau with updownupup spin configuration. A quantum
phase transition at the critical coupling ratio $J_{d}/J_{t}=0.6$ is found.
When $0<J_{d}/J_{t}<0.6$, the system is in a valence bond state, where an
obvious zeromagnetization plateau is observed, implying a gapful spin
excitation; when $J_{d}/J_{t}>0.6$, the system exhibits a gapless excitation,
in which the dimerdimer correlation is found decaying in a power law, while
the spinspin and chiralchiral correlation functions decay exponentially. At
the isotropic point ($J_{d}/J_{t}=1$), we unveil that at low temperature ($T$)
the specific heat depends linearly on $T$, and the susceptibility tends to a
constant for $T\rightarrow 0$, giving rise to a Wilson ratio around unity,
implying that the system under interest is a fermionic algebraic quantum spin
liquid.

In this work, a simple and fundamental numeric scheme dubbed as abinitio
optimization principle (AOP) is proposed for the ground states of translational
invariant stronglycorrelated quantum lattice models. The idea is to transform
a nondeterministicpolynomialhard ground state simulation with infinite
degrees of freedom into a single optimization problem of a local function with
finite number of physical and ancillary degrees of freedom. This work
contributes mainly in the following aspects: 1) AOP provides a simple and
efficient scheme to simulate the ground state by solving a local optimization
problem. Its solution contains two kinds of boundary states, one of which play
the role of the entanglement bath that mimic the interactions between a
supercell and the infinite environment, and the other give the ground state in
a tensor network (TN) form. 2) In the sense of TN, a novel decomposition named
as tensor ring decomposition (TRD) is proposed to implement AOP. Instead of
following the contractiontruncation scheme used by many existing TNbased
algorithms, TRD solves the contraction of a uniform TN in an opposite clue by
encoding the contraction in a set of selfconsistent equations that
automatically reconstruct the whole TN, making the simulation simple and
unified; 3) AOP inherits and develops the ideas of different wellestablished
methods, including the density matrix renormalization group (DMRG), infinite
timeevolving block decimation (iTEBD), network contractor dynamics, density
matrix embedding theory, and etc., providing a unified perspective that is
previously missing in this fields; 4) AOP as well as TRD gives novel
implications to existing TNbased algorithms: a modified iTEBD is suggested and
the 2D AOP is argued to be an intrinsic 2D extension of DMRG that is based on
infinite projected entangled pair state.

Determination and characterization of criticality in twodimensional (2D)
quantum manybody systems belong to the most important challenges and problems
of quantum physics. In this paper we propose an efficient scheme to solve this
problem by utilizing the infinite projected entangled pair state (iPEPS), and
tensor network (TN) representations. We show that the criticality of a 2D state
is faithfully reproduced by the ground state (dubbed as boundary state) of a
onedimensional effective Hamiltonian constructed from its iPEPS
representation. We demonstrate that for a critical state the correlation length
and the entanglement spectrum of the boundary state are essentially different
from those of a gapped iPEPS. This provides a solid indicator that allows to
identify the criticality of the 2D state. Our scheme is verified on the
resonating valence bond (RVB) states on kagom\'e and square lattices, where the
boundary state of the honeycomb RVB is found to be described by a $c=1$
conformal field theory. We apply our scheme also to the ground state of the
spin1/2 XXZ model on honeycomb lattice, illustrating the difficulties of
standard variational TN approaches to study the critical ground states. Our
scheme is of high versatility and flexibility, and can be further applied to
investigate the quantum criticality in many other phenomena, such as
finitetemperature and topological phase transitions.

Density matrix renormalization group (DMRG) is one of the most powerful
numerical methods available for manybody systems. In this work, we develop a
perturbation theory of DMRG (PTDMRG) to largely increase its accuracy in an
extremely simple and efficient way. By using the canonical matrix product state
(MPS) representation for the ground state of the considered system, a set of
orthogonal basis functions $\left\lbrace  \psi_i \rangle \right\rbrace$ is
introduced to describe the perturbations to the ground state obtained by the
conventional DMRG. The Schmidt numbers of the MPS that are beyond the bond
dimension cutoff are used to define such perturbation terms. The perturbed
Hamiltonian is then defined as $\tilde{H}_{ij}= \langle \psi_i  \hat{H} 
\psi_j \rangle$; its ground state permits to calculate physical observables
with a considerably improved accuracy as compared to the original DMRG results.
We benchmark the secondorder perturbation theory with the help of
onedimensional Ising chain in a transverse field and the Heisenberg chain,
where the precision of DMRG is shown to be improved $\rm O(10)$ times.
Furthermore, for moderate length $L$ the errors of DMRG and PTDMRG both scale
linearly with $L^{1}$. The linear relation between the dimension cutoff of
DMRG and that of PTDMRG with the same precision shows a considerable
improvement of efficiency, especially for large dimension cutoff's. In
thermodynamic limit we show that the errors of PTDMRG scale with
$\sqrt{L^{1}}$. Our work suggests an effective way to define the tangent space
of the ground state MPS, which may shed lights on the properties beyond the
ground state. Such secondorder PTDMRG can be readily generalized to higher
orders, as well as applied to the models in higher dimensions.

The anisotropic antiferromagnetic Ising model on the fractal Sierpi\'{n}ski
gasket is intensively studied, and a number of exotic properties are disclosed.
The ground state phase diagram in the plane of magnetic fieldinteraction of
the system is obtained. The thermodynamic properties of the three plateau
phases are probed by exploring the temperaturedependence of magnetization,
specific heat, susceptibility and spinspin correlations. No phase transitions
are observed in this model. In the absence of a magnetic field, the unusual
temperature dependence of the spin correlation length is obtained with $0
\leq$J$_b/$J$_a<1$, and an interesting crossover behavior between different
phases at J$_b/$J$_a=1$ is unveiled, whose dynamics can be described by the
J$_b/$J$_a$dependence of the specific heat, susceptibility and spin
correlation functions. The exotic spinspin correlation patterns that share the
same special rotational symmetry as that of the Sierpi\'{n}ski gasket are
obtained in both the $1/3$ plateau disordered phase and the $5/9$ plateau
partially ordered ferrimagnetic phase. Moreover, a quantum scheme is formulated
to study the thermodynamics of the fractal Sierpi\'{n}ski gasket with
Heisenberg interactions. We find that the unusual temperature dependence of the
correlation length remains intact in a small quantum fluctuation.

We propose a systematic scheme to reach the properties of twodimensional
(2D) statistical and quantum systems by studying the effective
(1+1)dimensional theory that is constructed from the tensor network
representation. On on hand, we discover that the degeneracy of the 2D system
can be determined by the purity of the boundary thermal state, which is the
density operator of the effective theory at zero (effective) temperature. On
the other hand, we find that the gapped (or critical) 2D system leads to a
gapped (or critical) effective (1+1)dimensional theory whose criticality can
be accessed by the entanglement entropy $S$ of its ground state dubbed as
boundary pure state. We also uncover that for the critical systems, $S$ obeys
the same logarithmic law as that found in the critical 1D quantum chains, which
reads $S = (\kappa c/6)\log_2 D + const.$, with $c$ the central charge and
$\kappa$ a constant related to the scaling property of the correlation length
$\xi$ as $\xi \sim D^{\kappa}$. Such a scaling law presents an efficient way to
characterize the critical universality class of the original 2D systems. An
important implication of our work is that many wellestablished theories for 1D
quantum chains become available for studying 2D systems with the help of the
proposed lower dimensional correspondence.

By utilizing tensornetworkbased methods, we investigate the zero and
finitetemperature properties of the spin1/2 Heisenberg antiferromagnetic
(HAF) model on an infinite Husimi lattice that contains 3/2 sites per triangle.
The ground state of this model is found to possess vanishing local
magnetization and is featureless; the spinspin and dimerdimer correlation
functions are verified to decay exponentially; and its groundstate energy per
site is determined to be $e_0=0.4343(1)$, which is very close to that
[$e_0=0.4386(5)$] of the intriguing kagome HAF model. The magnetization curve
shows the absence of a zeromagnetization plateau, implying a gapless
excitation. A $1/3$magnetization plateau with spin upupdown state is
observed, which is selected and stabilized by quantum fluctuations. A ground
state phase diagram under magnetic fields is presented. Moreover, both magnetic
susceptibility and the specific heat are studied, whose lowtemperature
behaviors reinforce the conclusion that the HAF model on the infinite Husimi
lattice owns a gapless and featureless spin liquid ground state.

Based on the tensor network state representation, we develop a nonlinear
dynamic theory coined as network contractor dynamics (NCD) to explore the
thermodynamic properties of twodimensional quantum lattice models. By invoking
the rank$1$ decomposition in the multilinear algebra, the NCD scheme makes
the contraction of the tensor network of the partition function be realized
through a contraction of a local tensor cluster with vectors on its boundary.
An imaginarytimesweep algorithm for implementation of the NCD method is
proposed for practical numerical simulations. We benchmark the NCD scheme on
the square Ising model, which shows a great accuracy. Besides, the results on
the spin1/2 Heisenberg antiferromagnet on honeycomb lattice are disclosed in
good agreement with the quantum Monte Carlo calculations. The
quasientanglement entropy $S$, Lyapunov exponent $I^{lya}$ and loop character
$I^{loop}$ are introduced within the dynamic scheme, which are found to display
the ``nonlocality" near the critical point, and can be applied to determine the
thermodynamic phase transitions of both classical and quantum systems.

The unusual ground state and thermodynamic properties of spin1/2 twoleg
honeycomb (HC) spin ladder are systematically studied by jointly utilizing
various analytical and numerical methods. The HC spin ladder is found to
exhibit behaviors dramatically different from those of the conventional square
spin ladder. A strong relevant term nJ and a half saturation magnetization
plateau that can be attributed to the formation of diluted dimer states are
observed in the HC ladder, both of which are absent in the square ladder. The
ground state phase diagram of the HC spin ladder is identified, and the
thermodynamic properties of the specific heat and susceptibility for different
couplings are thoroughly explored, where two kinds of excitations are unveiled.
The distinct Wilson ratios for both spin ladders at the lower critical fields
are also obtained. Our calculated result is well fitted to the experimental
data of the twoleg HC spin ladder compound [Cu2L1(N3)4]n.

The unusual reentrant phenomenon is observed in the anisotropic 3state Potts
model on a gen eralized Kagome lattice. By employing the linearized tensor
renormalization group method, we find that the reentrance can appear in the
region not only under a partial ordered phase as commonly known but also a
phase without a local order parameter, which is uncovered to fall into the uni
versality of the KosterlitzThouless (KT) type. The region of the reentrance
depends strongly on the ratios of the next nearest couplings {\alpha} = J2 /J1
 and {\beta} = J3 /J1 . The phase diagrams in the plane of temperature
versus {\beta} for different {\alpha} are obtained. Through massive
calculations, it is also revealed that the quasientanglement entropy can be
used to accurately detect the KT transition temperature.

A novel algorithm based on the optimized decimation of tensor networks with
superorthogonalization (ODTNS) that can be applied to simulate efficiently and
accurately not only the thermodynamic but also the ground state properties of
twodimensional (2D) quantum lattice models is proposed. By transforming the 2D
quantum model into a threedimensional (3D) closed tensor network (TN)
comprised of the tensor product density operator and a 3D brickwall TN, the
free energy of the system can be calculated with the imaginary time evolution,
in which the network Tucker decomposition is suggested for the first time to
obtain the optimal lowerdimensional approximation on the bond space by
transforming the TN into a superorthogonal form. The efficiency and accuracy
of this algorithm are testified, which are fairly comparable with the quantum
Monte Carlo calculations. Besides, the present ODTNS scheme can also be
applicable to the 2D frustrated quantum spin models with nice efficiency.

The ground state phase diagrams and thermodynamic properties of a spin1
bondalternating Heisenberg antiferromagnetic chain with a singleion
anisotropy in longitudinal and transverse magnetic fields are investigated
jointly by means of the infinite time evolving block decimation, the linearized
tensor renormalization group and the density matrix renormalization group
methods. It is found that in the magnetic fieldbond alternating ratio plane
six phases such as singlet dimer, Haldane, two TomonagaLuttinger liquid, 1/2
magnetization plateau and spin polarized phases are identified in a
longitudinal field, while in a transverse field there are fives phases
including singlet dimer, Haldane, Z$_2$ symmetry breaking, quasi1/2
magnetization plateau and quasi spin polarized phases. A reentrant behavior of
the staggered magnetization in a transverse field is observed. The quantum
critical behaviors in longitudinal and transverse magnetic fields are disclosed
to fall into different universality classes with corresponding conformal field
central charge $c=1$ and 1/2, respectively. The experimental data of the
compound NTENP under both longitudinal and transverse magnetic fields are
nicely fitted, and the Luttinger liquid behavior of low temperature specific
heat experimentally observed is also confirmed.

We propose a novel algorithm with a modified Tucker decomposition for tensor
network that allows for efficiently and precisely calculating the ground state
and thermodynamic properties of twodimensional (2D) quantum spin lattice
systems, and is coined as the fast linearized tensor renormalization group
(fLTRG). Its amazing efficiency and precision are examined by studying the
spin1/2 anisotropic Heisenberg antiferromagnet on a honeycomb lattice, and the
results are found to be fairly in agreement with the quantum Monte Carlo
calculations. It is also successfully applied to tackle a quasi2D spin1/2
frustrated bilayer honeycomb Heisenberg model, where a quantum phase transition
from an ordered antiferromagnetic state to a gapless quantum spin liquid phase
is found. The thermodynamic behaviors of this frustrated spin system are also
explored. The present fLTRG algorithm could be readily extended to other
quantum lattice systems.

A linearized tensor renormalization group (LTRG) algorithm is proposed to
calculate the thermodynamic properties of onedimensional quantum lattice
models, that is incorporated with the infinite timeevolving block decimation
technique, and allows for treating directly the twodimensional transfermatrix
tensor network. To illustrate its feasibility, the thermodynamic quantities of
the quantum XY spin chain are calculated accurately by the LTRG, and the
precision is shown to be comparable with (even better than) the transfer matrix
renormalization group (TMRG) method. Unlike the TMRG scheme that can only deal
with the infinite chains, the present LTRG algorithm could treat both finite
and infinite systems, and may be readily extended to boson and fermion quantum
lattice models.

The twodimensional Ising model on a distorted Kagom\'{e} lattice is studied
by means of exact solutions and the tensor renormalisation group (TRG) method.
The zerofield phase diagrams are obtained, where three phases such as
ferromagnetic, ferrimagnetic and paramagnetic phases, along with the
secondorder phase transitions, have been identified. The TRG results are quite
accurate and reliable in comparison to the exact solutions. In a magnetic
field, the magnetization ($m$), susceptibility and specific heat are studied by
the TRG algorithm, where the $m=1/3$ plateaux are observed in the magnetization
curves for some couplings. The experimental data of susceptibility for the
complex Co(N$_3$)$_2$(bpg)$\cdot$ DMF$_{4/3}$ are fitted with the TRG results,
giving the couplings of the complex $J=22K$ and $J'=33K$.