Optimization of intelligent compaction based on finite element simulation and nonlinear multiple regression

Chengyong Chen; Fagang Chang; Li Li; Wenqiang Dou; Changjing Xu; Chengyong Chen; Fagang Chang; Li Li; Wenqiang Dou; Changjing Xu

doi:10.3934/era.2023140

Electronic Research Archive

2023, Volume 31, Issue 5: 2775-2792. doi: 10.3934/era.2023140

Previous Article Next Article

Research article Special Issues

Optimization of intelligent compaction based on finite element simulation and nonlinear multiple regression

1.
Shandong Hi-Speed infrastructure construction co., LTD, Jinan 250000, China
2.
Shandong Hi-Speed Mansion, Jinan 250000, China

Academic Editor: Hui Yao

Received: 04 December 2022 Revised: 15 January 2023 Accepted: 18 January 2023 Published: 13 March 2023

In intelligent compaction, a critical issue is determining the combination of construction parameters (e.g., the rolling speed and the number of passes) for achieving optimal compaction results. In this paper, a finite element model was developed based on the Mohr-Coulomb elasto-plastic model to simulate the field compaction process of subgrade, which was validated by field compaction tests. Nonlinear multiple regression was used to match the impacts of construction factors on compaction quality based on the model simulation. Then, the linear search approach was used to find the ideal combination of construction parameters that optimizes the compaction quality. The findings indicated that the ideal combination of construction parameters for reaching the ideal compaction degree is a rolling speed of 1.3 m/s with 4 roller passes.

Keywords:

Citation: Chengyong Chen, Fagang Chang, Li Li, Wenqiang Dou, Changjing Xu. Optimization of intelligent compaction based on finite element simulation and nonlinear multiple regression[J]. Electronic Research Archive, 2023, 31(5): 2775-2792. doi: 10.3934/era.2023140

Related Papers:

[1]	Liping Yang, Hu Li . A hybrid algorithm based on parareal and Schwarz waveform relaxation. Electronic Research Archive, 2022, 30(11): 4086-4107. doi: 10.3934/era.2022207
[2]	Chunhua Chu, Yijun Chen, Qun Zhang, Ying Luo . Joint linear array structure and waveform design for MIMO radar under practical constraints. Electronic Research Archive, 2022, 30(9): 3249-3265. doi: 10.3934/era.2022165
[3]	Xuerong Hu, Yuxiang Han, Junyan Lu, Linxiang Wang . Modeling and experimental investigation of quasi-zero stiffness vibration isolator using shape memory alloy springs. Electronic Research Archive, 2025, 33(2): 768-790. doi: 10.3934/era.2025035
[4]	Qian He, Wenxin Du, Feng Shi, Jiaping Yu . A fast method for solving time-dependent nonlinear convection diffusion problems. Electronic Research Archive, 2022, 30(6): 2165-2182. doi: 10.3934/era.2022109
[5]	Li Tian, Ziqiang Wang, Junying Cao . A high-order numerical scheme for right Caputo fractional differential equations with uniform accuracy. Electronic Research Archive, 2022, 30(10): 3825-3854. doi: 10.3934/era.2022195
[6]	Yi Wei . The Riccati-Bernoulli subsidiary ordinary differential equation method to the coupled Higgs field equation. Electronic Research Archive, 2023, 31(11): 6790-6802. doi: 10.3934/era.2023342
[7]	Adrian Nicolae Branga . Some conditions for the existence and uniqueness of monotonic and positive solutions for nonlinear systems of ordinary differential equations. Electronic Research Archive, 2022, 30(6): 1999-2017. doi: 10.3934/era.2022101
[8]	Junseok Kim . Maximum principle preserving the unconditionally stable method for the Allen–Cahn equation with a high-order potential. Electronic Research Archive, 2025, 33(1): 433-446. doi: 10.3934/era.2025021
[9]	Martin Bohner, Alexander Domoshnitsky, Oleg Kupervasser, Alex Sitkin . Floquet theory for first-order delay equations and an application to height stabilization of a drone's flight. Electronic Research Archive, 2025, 33(5): 2840-2861. doi: 10.3934/era.2025125
[10]	David Cheban, Zhenxin Liu . Averaging principle on infinite intervals for stochastic ordinary differential equations. Electronic Research Archive, 2021, 29(4): 2791-2817. doi: 10.3934/era.2021014

Abstract

1. Introduction

Waveform relaxation (WR) is an iterative method for systems of ordinary differential equations (ODEs). It is introduced for the first time by Lelarasmee et al. ^[1] for the time domain analysis of large-scale nonlinear dynamical systems. A lot of studies have been done on WR methods and most of them focus on convergence (see ^{[2,3,4,5,6,7,8,9]}). Zero-stability is the basic property of numerical methods of ODEs^[10]. However, to our best knowledge, so far there exists little work on zero-stability of WR methods.

We first of all propose two examples to show the fact that tiny perturbation may lead to huge change of WR approximation solutions, at the same time, to confirm that it is necessary to study zero-stability of WR methods. For this, it is enough to consider a simple method, and suppose only the initial value has tiny perturbations.

Example 1. Consider the WR method

$\begin{equation} x^{(k+1)}_{n+2}-(1+\alpha)x^{(k+1)}_{n+1}+\alpha x^{(k+1)}_{n} = -(1-\alpha)h(x^{(k+1)}_{n}+x^{(k)}_{n}) \end{equation}$

(1.1)

of solving numerically the initial value problem

$\dot{x}(t) = -2x(t), t\in[0, T]; x(0) = x_0,$

which is gotten by applying the two step method

$x_{n+2}-(1+\alpha)x_{n+1}+\alpha x_n = -2(1-\alpha)hx_n$

that is consistent for any $\alpha$ and zero-stable for $-1 < \alpha < 1$ to the iterative scheme

$\dot{x}^{(k+1)}(t) = -x^{(k+1)}(t)-x^{(k)}(t).$

Suppose that

$\begin{equation} N \;{\rm{ is\;a\;positive\;integer\;and }}\; h = \dfrac{T}{N}, \end{equation}$

(1.2)

$\begin{equation} x^{(0)}_n = x_0, n = 0, 1, \cdots, N, x^{(k)}_0 = x_0, k = 0, 1, \cdots \end{equation}$

(1.3)

and

$\begin{equation} x^{(k+1)}_1 = x^{(k+1)}_0-h (x^{(k+1)}_0+ x^{(k)}_0), k = 0, 1, \cdots. \end{equation}$

(1.4)

We can get the approximate solution $\{\{x^{(k)}_n, n = 0, 1, \cdots, N\}, k = 0, 1, \cdots\}$ by using Eqs (1.1)–(1.4). Suppose that tiny perturbations denoted by $\delta$ lead to the initial values become

$\begin{equation} \tilde{x}^{(0)}_n = x_0+\delta, n = 0, 1, \cdots, N, \tilde{x}^{(k)}_0 = x_0+\delta, k = 0, 1, \cdots \end{equation}$

(1.5)

and the resulting perturbed solution of Eq (1.1) is denoted by $\{\tilde{x}^{(k)}_n\}$ . Let $\delta = 1$ for simplicity and denote $\tilde{x}^{(k)}_n-x^{(k)}_n$ with $e^{(k)}_n$ . Then we have

$\begin{equation} \begin{aligned} &e^{(k+1)}_{n+2}-(1+\alpha)e^{(k+1)}_{n+1}+\alpha e^{(k+1)}_{n} = -(1-\alpha)h(e^{(k+1)}_{n}+e^{(k)}_{n})\\ &\qquad \qquad \qquad (n = 0, 1, \cdots, N-2, k = 0, 1, \cdots), \\ &e^{(k+1)}_0 = 1, e^{(k+1)}_1 = (1-2h)e^{(k+1)}_0(k = 0, 1, \cdots), e^{(0)}_n = 1(n = 0, 1, \cdots, N).\\ \end{aligned} \end{equation}$

(1.6)

By virtue of Eq (1.6) the following numerical results are obtained.

The data in and show that errors of the approximate solution of Eq (1.1) brought by perturbations of initial values are controllable for $\alpha = 0.5$ , but they are unbounded for $\alpha = 2$ when $h$ tends to zero.

Table 1. Results obtained by (1.6) with

$\alpha = 0.5$ .

$N$	10	100	250	500	750	1000
$e^{(5)}_N$	0.10621	0.1319	0.13343	0.13388	0.13402	0.13409
$e^{(10)}_N$	0.10621	0.13238	0.13422	0.13478	0.13497	0.13506
$e^{(100)}_N$	0.10621	0.13238	0.13422	0.13478	0.13497	0.13506

| Show Table

DownLoad: CSV

Table 2. Results obtained by (1.6) with

$\alpha = 2$ .

$N$	10	100	250	500	1000
$e^{(5)}_N$	268.74	6.2863e+028	3.7892e+073	3.4922e+148	5.7695e+298
$e^{(10)}_N$	268.74	6.2867e+028	3.7897e+073	3.4927e+148	5.7704e+298
$e^{(100)}_N$	268.74	6.2867e+028	3.7897e+073	3.4927e+148	5.7704e+298

| Show Table

DownLoad: CSV

In this numerical examples, taking $\delta = 1$ instead of tiny perturbations seem to be unreasonable. The main reason for this is that we are usually interested in the ratio $\frac{e^{(k)}_N}{\delta}$ instead of $\delta$ itself. In fact we have also performed some other experiments using smaller $\delta$ , in which the similar ratios are obtained for different $\delta$ .

Example 2. Consider the one-step WR method

$\begin{equation} \begin{aligned} &x^{(k+1)}_{n+1} = x^{(k+1)}_n+h\left(500(x^{(k+1)}_n)^{1.001}+500(x^{(k)}_n)^{1.001}\right), n = 0, 1, \cdots, \dfrac{2}{h}, \\ &x^{(k+1)}_0 = 0;\ x^{(0)}_n = 0, \;{ \rm{ for\;all } }\;n, \end{aligned} \end{equation}$

(1.7)

where $2/h$ is an integer. Suppose that the initial value $0$ becomes $\delta$ by the influence of perturbations and the solution $x_n^{(k)}$ becomes $\tilde{x}_n^{(k)}$ by the influence of $\delta$ , which satisfies

$\begin{equation} \begin{aligned} &\tilde{x}^{(k+1)}_{n+1} = \tilde{x}^{(k+1)}_n+h\left(500(\tilde{x}^{(k+1)}_n)^{1.001}+500(\tilde{x}^{(k)}_n)^{1.001}\right), n = 0, 1, \cdots, \dfrac{2}{h}, \\ &\tilde{x}^{(k+1)}_0 = \delta;\ \tilde{x}^{(0)}_n = \delta, \;{ \rm{ for\;all } }\;n. \end{aligned} \end{equation}$

(1.8)

Clearly the numerical solution generated by Eq (1.7) is the constant zero and Eq (1.8) can be regarded as the WR method of solving the following initial value problem

$\begin{equation} \dot{x}(t) = 1000(x(t))^{1.001}, t\in[0, 2]; x(0) = \delta. \end{equation}$

(1.9)

It is easy to derive that $t_0 = \dfrac{1}{\delta^{0.001}}$ is the blow-up time of the solution of Eq (1.9) and $t_0\in [0, 2]$ whenever $\delta > 10^{-301}$ . Consequently, $\sup\limits_{0 < t < 2}x(t) = +\infty$ for any $\delta > 10^{-301}$ . If the "2" in Eqs (1.8) and (1.9) is replaced by $t_0-\varepsilon$ , where $\varepsilon > 0$ is small enough, then Eq (1.8) converges to Eq (1.9), that is,

$\lim\limits_{h\rightarrow 0, k\rightarrow \infty}\sup\limits_{0\leq n\leq \frac{t_0-\varepsilon}{h}}|\tilde{x}^{(k)}_n-x(nh)| = 0,$

see ^[4]. Thus

$\lim\limits_{h\rightarrow 0, k\rightarrow \infty}\sup\limits_{0\leq n\leq \frac{2}{h}}|\tilde{x}^{(k)}_n| = +\infty, \;{ \rm{ for\;any } }\;\delta > 10^{-301},$

that is, the change of the solution of Eq (1.8) brought by the tiny change of initial values is unbounded as $h$ tends to zero.

The above two examples show that tiny perturbations can lead to huge change of WR approximation solutions, although the methods used in the examples are too simple to be practically useful.

A numerical method is said to be zero-stable if the change of its solution brought by the tiny perturbations is controllable when $h$ tends to zero (see ^[10]). In this paper we will explore what conditions guarantee the zero-stability of WR methods.

2. Preliminary

Consider the initial problem

$\dot{x}(t) = f(t, x(t)), t\in[0, T]; x(0) = x_0 (x(t)\in\mathbb{R}^d).$

Taking the splitting function $F(t, x, x) = f(t, x)$ we can construct the iterative scheme

$\begin{equation} \begin{aligned} &\dot{x}^{(k+1)}(t) = F(t, x^{(k+1)}(t), x^{(k)}(t)), t\in[0, T], k = 0, 1, \cdots;\\ & x^{(0)}(t)\equiv x_0, x^{(k+1)}(0) = x_0. \end{aligned} \end{equation}$

(2.1)

Using one-step methods with variable step-size to discrete Eq (2.1) one arrives at

$\begin{equation} \begin{aligned} &x^{(k+1)}_{n+1} = x^{(k+1)}_n+h_n \Phi(t_{n}, h_n, x^{(k+1)}_{n+1}, x^{(k+1)}_n, x^{(k)}_{n+1}, x^{(k)}_n);x^{(k+1)}_0 = x_0, \end{aligned} \end{equation}$

(2.2)

where $n = 0, 1, \cdots, N-1, k = 0, 1, \cdots$ , $\sum\limits_{n = 0}^{N-1}h_n = T$ and $x^{(0)}_n = x_0$ for all $n$ . Applying a class of multi-step methods with fixed step-size to Eq (2.1) we get

$\begin{equation} x^{(k+1)}_{n+1} = x^{(k+1)}_n+h\Psi(t_n, h, x^{(k+1)}_{n+1}, x^{(k+1)}_{n}, \cdots, x^{(k+1)}_{n-p}, x^{(k)}_{n+1}, x^{(k)}_{n}, \cdots, x^{(k)}_{n-p}), \end{equation}$

(2.3)

where $n = p, p+1, \cdots, N-1, k = 0, 1, \cdots$ , $h = T/N$ , the initial values except $x^{(0)}_n(n = 0, 1, \cdots, N)$ and $x^{(k)}_0(k = 0, 1, \cdots)$ are generated by a suitable one-step method. Here we take the initial values as

$\begin{equation} x^{(k)}_n = c^{(k)}_n, n = 0, 1, \cdots, p, k = 0, 1, \cdots(c^{(k)}_n = x_0, \;{\rm{ if }}\;n\cdot k = 0). \end{equation}$

(2.4)

A classical example of Eq (2.3) is Admas-type linear multi-step methods:

$x^{(k+1)}_{n+1} = x^{(k+1)}_n+h\sum\limits_{j = 0}^{p+1} \beta_j F(t_{n+1-j}, x^{(k+1)}_{n+1-j}, x^{(k)}_{n+1-j}), n = p, p+1, \cdots, N-1,$

where $\beta_j, j = 0, 1, \cdots, p+1$ are constants. The linear multi-step methods use generally the fixed step-size and therefore we only consider the Eq (2.3) with the fixed step-size.

Let $_ix^{(k)}$ denote the component of $x^{(k)}$ satisfying

$x^{(k)} = \left(\left(_1x^{(k)}\right)^{\rm{T}}, \left(_2x^{(k)}\right)^{\rm{T}}, \cdots, \left(_mx^{(k)}\right)^{\rm{T}}\right)^{\rm{T}},$

where $_ix^{(k)}\in \mathbb{R}^{d_i}$ , $x^{(k)}\in \mathbb{R}^{d}$ and $d_1+d_2+\cdots+d_m = d$ . The Eq (2.1) can be rewritten as

$_i\dot{x}^{(k+1)} = F_i(t, \sideset{_1}{}{\mathop{x}^{(k+1)}}, \sideset{_2}{}{\mathop{x}^{(k+1)}}, \cdots, \sideset{_m}{}{\mathop{x}^{(k+1)}}, \sideset{_1}{}{\mathop{x}^{(k)}}, \sideset{_2}{}{\mathop{x}^{(k)}}, \cdots, \sideset{_m}{}{\mathop{x}^{(k)}}),$

with $i = 1, 2, \cdots, m$ , that is, the large Eq (2.1) is divided into $m$ subsystems. When taking

$F_i(t, \cdots) = f_i(t, \sideset{_1}{}{\mathop{x}^{(k)}}, \cdots, \sideset{_{i-1}}{}{\mathop{x}^{(k)}}, \sideset{_i}{}{\mathop{x}^{(k+1)}}, \sideset{_{i+1}}{}{\mathop{x}^{(k)}}, \cdots, \sideset{_m}{}{\mathop{x}^{(k)}}), i = 1, 2, \cdots, m$

one arrives at Guass-Jacobi WR method used frequently in actual computation, which consists of $m$ independent subsystems and is hence parallel in nature(see ^[2,6,9]). In Eq (2.2) the unique mesh $0 = t_0 < t_1 < \cdots < t_{N} = T$ is applied to all subsystems. However the subsystems of Gauss-Jocobi WR methods are independent and may have distinct behaviors, and consequently it is better to use different meshes for the subsystems with different behaviors. Applying the mesh

$0 = \sideset{_i}{_0}{\mathop{t}} < \sideset{_i}{_1}{\mathop{t}} < \cdots < \sideset{_i}{_{N_i}}{\mathop{t}} = T, \sideset{_i}{_j}{\mathop{h}} = \sideset{_i}{_{j+1}}{\mathop{t}}-\sideset{_i}{_{j}}{\mathop{t}}, j = 0, 1, \cdots, N_i-1$

to $i$ th subsystem yields the following multi-rate WR method

$\begin{equation} \begin{aligned} \sideset{_i}{^{(k+1)}_{n+1}}{\mathop{x}} = &\sideset{_i}{^{(k+1)}_{n}}{\mathop{x}}+\sideset{_i}{_n}{\mathop{h}}\Phi_i\left(\sideset{_i}{_n}{\mathop{t}}, \sideset{_i}{_n}{\mathop{h}}, \sideset{_1}{^{(k)}}{\mathop{y}}(\sideset{_i}{_{n+1}}{\mathop{t}}), \cdots, \sideset{_{i-1}}{^{(k)}}{\mathop{y}}(\sideset{_i}{_{n+1}}{\mathop{t}}), \sideset{_i}{^{(k+1)}_{n+1}}{\mathop{x}}, \right.\\ &\quad \sideset{_{i+1}}{^{(k)}}{\mathop{y}}(\sideset{_i}{_{n+1}}{\mathop{t}}), \cdots, \sideset{_m}{^{(k)}}{\mathop{y}}(\sideset{_i}{_{n+1}}{\mathop{t}}), \sideset{_1}{^{(k)}}{\mathop{y}}(\sideset{_i}{_{n}}{\mathop{t}}), \cdots, \sideset{_{i-1}}{^{(k)}}{\mathop{y}}(\sideset{_i}{_{n}}{\mathop{t}}), \\ &\quad\left. \sideset{_i}{^{(k+1)}_{n}}{\mathop{x}}, \sideset{_{i+1}}{^{(k)}}{\mathop{y}}(\sideset{_i}{_{n}}{\mathop{t}}), \cdots, \sideset{_m}{^{(k)}}{\mathop{y}}(\sideset{_i}{_{n}}{\mathop{t}})\right), n = 0, 1, \cdots, N_i-1, \end{aligned} \end{equation}$

(2.5)

where $i = 1, 2, \cdots, m, k = 0, 1, \cdots$ . Here the initial values $\sideset{_i}{^{(k+1)}_0}{\mathop{x}} = \sideset{_i}{^{(0)}_n}{\mathop{x}} = \sideset{_i}{_0}{\mathop{x}}$ for all $k$ and $n$ , $\left(\sideset{_1}{_0}{\mathop{x}}^{\rm{T}}, \sideset{_2}{_0}{\mathop{x}}^{\rm{T}}, \cdots, \sideset{_m}{_0}{\mathop{x}}^{\rm{T}} \right)^{\rm{T}} = x_0$ and $\sideset{_j}{^{(k)}}{\mathop{y}}(t)$ is the interpolation function satisfying $\sideset{_j}{^{(k)}}{\mathop{y}}(\sideset{_j}{_n}{\mathop{t}}) = \sideset{_j}{^{(k)}_n}{\mathop{x}}$ for $n = 0, 1, \cdots, N_j$ .

Because of the presence of errors in actual computing, it is necessary to consider the following perturbed systems of Eqs (2.2), (2.3) and (2.5) generated by the perturbations $\{\delta^{(k)}_n, n = 0, 1, \cdots, N, k = 0, 1, \cdots\}$

$\begin{equation} \begin{aligned} &&\tilde{x}^{(k+1)}_{n+1} = \tilde{x}^{(k+1)}_n+h_n \Phi(t_{n}, h_n, \tilde{x}^{(k+1)}_{n+1}, \tilde{x}^{(k+1)}_n, \tilde{x}^{(k)}_{n+1}, \tilde{x}^{(k)}_n)+h_n\delta^{(k+1)}_{n+1}, \\ &&n = 0, 1, \cdots, N-1;\\ &&\tilde{x}^{(k+1)}_0 = x_0+\delta^{(k+1)}_{0}, \tilde{x}^{(0)}_n = x_0+\delta^{(0)}_{n}(n = 0, 1, \cdots, N), \end{aligned} \end{equation}$

(2.6)

$\begin{equation} \begin{aligned} &\tilde{x}^{(k+1)}_{n+1} = \tilde{x}^{(k+1)}_n+h \Psi\left(t_n, h, \tilde{x}^{(k+1)}_{n+1}, \tilde{x}^{(k+1)}_{n}, \cdots, \tilde{x}^{(k+1)}_{n-p}, \right.\\ &\qquad \qquad \qquad \qquad \left. \tilde{x}^{(k)}_{n+1}, \tilde{x}^{(k)}_{n}, \cdots, \tilde{x}^{(k)}_{n-p}\right)+h\delta^{(k+1)}_{n+1}, n = p, p+1, \cdots, N-1;\\ &\tilde{x}^{(k+1)}_n = c^{(k+1)}_n+\delta^{(k+1)}_n, n = 0, 1, \cdots, p , \tilde{x}^{(0)}_n = x_0+\delta^{(0)}_{n}(n = 0, 1, \cdots, N), \end{aligned} \end{equation}$

(2.7)

and

$\begin{equation} \begin{aligned} &\sideset{_i}{^{(k+1)}_{n+1}}{\mathop{\tilde{x}}} = \sideset{_i}{^{(k+1)}_{n}}{\mathop{\tilde{x}}}+\sideset{_i}{_n}{\mathop{h}}\Phi_i\left(\sideset{_i}{_n}{\mathop{t}}, \sideset{_i}{_n}{\mathop{h}}, \sideset{_1}{^{(k)}}{\mathop{\tilde{y}}}(\sideset{_i}{_{n+1}}{\mathop{t}}), \cdots, \sideset{_{i-1}}{^{(k)}}{\mathop{\tilde{y}}}(\sideset{_i}{_{n+1}}{\mathop{t}}), \sideset{_i}{^{(k+1)}_{n+1}}{\mathop{\tilde{x}}}, \right.\\ &\qquad \qquad \quad \sideset{_{i+1}}{^{(k)}}{\mathop{\tilde{y}}}(\sideset{_i}{_{n+1}}{\mathop{t}}), \cdots, \sideset{_m}{^{(k)}}{\mathop{\tilde{y}}}(\sideset{_i}{_{n+1}}{\mathop{t}}), \sideset{_1}{^{(k)}}{\mathop{\tilde{y}}}(\sideset{_i}{_{n}}{\mathop{t}}), \cdots, \sideset{_{i-1}}{^{(k)}}{\mathop{\tilde{y}}}(\sideset{_i}{_{n}}{\mathop{t}}), \\ &\qquad \qquad \quad \left.\sideset{_i}{^{(k+1)}_{n}}{\mathop{\tilde{x}}}, \sideset{_{i+1}}{^{(k)}}{\mathop{\tilde{y}}}(\sideset{_i}{_{n}}{\mathop{t}}), \cdots, \sideset{_m}{^{(k)}}{\mathop{\tilde{y}}}(\sideset{_i}{_{n}}{\mathop{t}})\right)+\sideset{_i}{_n}{\mathop{h}}\cdot\sideset{_i}{^{(k+1)}_{n+1}}{\mathop{\delta}}, \\ &\qquad \qquad\qquad \qquad \qquad \qquad \qquad\qquad \qquad \qquad n = 0, 1, \cdots, N_i-1;\\ &\sideset{_i}{^{(k+1)}_0}{\mathop{\tilde{x}}} = \sideset{_i}{_0}{\mathop{x}}+\sideset{_i}{^{(k+1)}_0}{\mathop{\delta}}, \sideset{_i}{^{(0)}_n}{\mathop{\tilde{x}}} = \sideset{_i}{_0}{\mathop{x}}+\sideset{_i}{^{(0)}_n}{\mathop{\delta}}(n = 0, 1, \cdots, N_i), \end{aligned} \end{equation}$

(2.8)

where $i = 1, 2, \cdots, m$ and $\sideset{_j}{^{(k)}}{\mathop{\tilde{y}}}(t)$ is the interpolation function satisfying $\sideset{_j}{^{(k)}}{\mathop{\tilde{y}}}(\sideset{_j}{_n}{\mathop{t}}) = \sideset{_j}{^{(k)}_n}{\mathop{\tilde{x}}}$ for $n = 0, 1, \cdots, N_j, j = 0, 1, \cdots, m$ .

Definition 2.1. (Page 32 of ^[10]) Let $\{\delta^{(k)}_n, n = 0, 1, \cdots, N, k = 0, 1, \cdots\}$ be any perturbation of Eq (2.2) or Eq (2.3) and Eq (2.4), and let $\{\tilde{x}^{(k)}_n, n = 0, 1, \cdots, N, k = 0, 1, \cdots\}$ be the solution of the resulting perturbed system Eq (2.6) or Eq (2.7). Then if there exist constants $C$ and $h_0$ such that, for all $h\in(0, h_0]$ ( $h = \max\limits_n h_n$ for Eq (2.2))

$\max\limits_{\begin{subarray}{c}0\leq k, 0\leq n\leq N\end{subarray}}\|\tilde{x}^{(k)}_n-x^{(k)}_n\|\leq C \varepsilon,$

whenever

$\max\limits_{\begin{subarray}{c}0\leq k, 0\leq n\leq N\end{subarray}}\|\delta^{(k)}_n\|\leq \varepsilon,$

we say that the Eq (2.2) or Eqs (2.3) and Eq (2.4) is zero-stable.

Here and hereafter $\|\cdot\|$ denotes any norm defined in $\mathbb{R}^d$ .

Definition 2.2. Let $\{_i\delta^{(k)}_n, i = 1, 2, \cdots, m, n = 0, 1, \cdots, N, k = 0, 1, \cdots\}$ be any perturbation of Eq (2.5), and let $\{_i\tilde{x}^{(k)}_n, i = 1, 2, \cdots, m, n = 0, 1, \cdots, N, k = 0, 1, \cdots\}$ be the solution of the resulting perturbed Eq (2.8). Then if there exist constants $h_0$ and $C_k$ depended on $k$ such that

$\max\limits_{\begin{subarray}{c}1\leq i\leq m\\0\leq l\leq k\end{subarray}} \max\limits_{0\leq n\leq N_i}\|_i\tilde{x}^{(l)}_n-\sideset{_i}{^{(l)}_n}{\mathop{x}}\|\leq C_k \varepsilon$

for all $h = \max\limits_{\begin{subarray}{c}1\leq i\leq m\end{subarray}} \max\limits_{0\leq n\leq N_i}\sideset{_i}{_n}{\mathop{h}}\in(0, h_0]$ , whenever

$\max\limits_{\begin{subarray}{c}1\leq i\leq m\\0\leq l\leq k\end{subarray}}\max\limits_{0\leq n\leq N_i}\|_i\delta^{(l)}_n\| \leq \varepsilon,$

we say that the Eq (2.5) is weakly zero-stable.

The following several lemmas are useful for studying zero-stability of the WR methods mentioned above.

Lemma 2.3. Let $a, b$ and $T$ be nonnegative constants, and $N$ be a positive integer. Suppose that the sets $\{h_n > 0, n = 0, 1, \cdots, N-1\}$ and $\{u_n > 0, n = 0, 1, \cdots, N\}$ satisfy

$\begin{equation} u_{n+1}\leq (1+ah_n)u_n+bh_n, n = 0, 1, \cdots, N-1. \end{equation}$

(2.9)

Then

$\max\limits_{0\leq n\leq N} u_n\leq \mathrm{e} ^{aT}u_0+\dfrac{b}{a}(\mathrm{e}^{aT}-1)$

provided that $\sum\limits_{n = 0}^{N-1}h_n = T$ .

Proof. By using Eq (2.9) repeatedly we have

$\begin{aligned} u_{n+1}\leq &(1+ah_n)(1+ah_{n-1})\cdots (1+ah_0)u_0\\ &+(1+ah_n)(1+ah_{n-1})\cdots (1+ah_1)bh_0\\ &+(1+ah_n)(1+ah_{n-1})\cdots (1+ah_2)bh_1\\ &+\cdots+(1+ah_n)bh_{n-1}+bh_n, \end{aligned}$

where $n = 0, 1, \cdots, N-1$ . Let $t_0 = 0, t_{n+1} = t_n+h_n, n = 0, 1, \cdots, N-1$ , then $t_N = T$ . Noting that $1+x < {\mathrm{e}}^x$ for $x > 0$ we can derive easily from above inequality

$\begin{aligned} u_{n+1}\leq &{\mathrm{e}}^{a(t_{n+1}-t_0)}u_0+{\mathrm{e}}^{a(t_{n+1}-t_1)}b(t_1-t_0)+{\mathrm{e}}^{a(t_{n+1}-t_2)}b(t_2-t_1)\\ &+\cdots+{\mathrm{e}}^{a(t_{n+1}-t_n)}b(t_n-t_{n-1})+{\mathrm{e}}^{a(t_{n+1}-t_{n+1})}b(t_{n+1}-t_{n})\\ \leq &{\mathrm{e}}^{a(T-t_0)}u_0+{\mathrm{e}}^{aT}b\left({\mathrm{e}}^{-at_1}(t_1-t_0)+\cdots+{\mathrm{e}}^{-at_n}(t_n-t_{n-1}) \right.\\ &\qquad \qquad \qquad \qquad \left.+\cdots+{\mathrm{e}}^{-at_N}(t_N-t_{N-1})\right) \end{aligned}$

for $n = 0, 1, \cdots, N-1$ . This together with monotonicity of the function ${\mathrm{e}}^{-at}$ imply

$\begin{aligned} u_{n+1}\leq {\mathrm{e}}^{a(T-t_0)}u_0+{\mathrm{e}}^{aT}b\int_{t_0}^T{\mathrm{e}}^{-at}\mathrm{d}t = {\mathrm{e}}^{aT}u_0+\dfrac{b}{a}\left({\mathrm{e}}^{aT}-1\right) \end{aligned}$

for all $n = 0, 1, \cdots, N-1$ .

Lemma 2.4. Let $a, b, c, d$ and $T$ be nonnegative constants, $N$ be a positive integer and $h_n, n = 0, 1, \cdots, N-1,$ be positive real numbers.

Suppose that the sequence of positive numbers $\{e_{n}^{(k)} > 0, n = 0, 1, \cdots N, k = 0, 1, \cdots\}$ satisfies

$\begin{equation} e^{(k+1)}_{n+1}\leq (1+ah_n)e^{(k+1)}_{n}+bh_ne^{(k)}_{n}+ch_ne^{(k)}_{n+1}+dh_n, n = 0, 1, \cdots N-1. \end{equation}$

(2.10)

Then for any $0 < h = \max\limits_{0\leq n\leq N-1} h_n < \dfrac{1}{2c}$

$\begin{equation} \max\limits_{\begin{subarray}{c} 0\leq k, 0\leq n\leq N\end{subarray}}e^{(k)}_n\leq \mathrm{e}^{2(a+b+c)T}\max\left\{\max\limits_{0\leq k}e^{(k)}_0, \max\limits_{0\leq n\leq N}e^{(0)}_n\right\}+ \dfrac{d}{a+b+c}(\mathrm{e}^{2(a+b+c)T}-1) \end{equation}$

(2.11)

provided that $\sum\limits_{n = 0}^{N-1}h_n = T$ .

Proof. By Eq (2.10) we have

$\begin{equation} \max\limits_{0\leq k}e^{(k+1)}_{n+1}\leq (1+ah_n)\max\limits_{0\leq k}e^{(k+1)}_{n}+bh_n\max\limits_{0\leq k}e^{(k)}_{n} +ch_n\max\limits_{0\leq k}e^{(k)}_{n+1}+dh_n \end{equation}$

(2.12)

for $n = 0, 1, \cdots, N-1$ . Let $\varepsilon = \max\limits_{0\leq n\leq N}e^{(0)}_n$ and $u_n = \max\left\{\max\limits_{0\leq k}e^{(k)}_n, \varepsilon\right\}, n = 0, 1, \cdots, N$ . Clearly $\max\left\{\max\limits_{0\leq k}e^{(k+1)}_n, \varepsilon\right\} = u_n$ . Hence we can get by Eq (2.12)

$\begin{aligned} u_{n+1}&\leq (1+ah_n)u_n+bh_nu_n+ch_nu_{n+1}+dh_n\\ &\leq \dfrac{1+ah_n+bh_n}{1-ch_n}u_n+\dfrac{dh_n}{1-ch_n}, n = 0, 1, \cdots, N-1. \end{aligned}$

Suppose that $0 < h < \dfrac{1}{2c}$ . Consequently, $\dfrac{1}{1-ch_n} < 2$ and the above inequality thus yield

$u_{n+1}\leq (1+2(a+b+c)h_n)u_n+2dh_n, n = 0, 1, \cdots, N-1.$

This together with Lemma 2.3 yield

$\max\limits_{0\leq n\leq N}u_n\leq \mathrm{e}^{2(a+b+c)T}u_0+\dfrac{d}{a+b+c}(\mathrm{e}^{2(a+b+c)T}-1).$

Thus Eq (2.11) holds true. The proof is complete.

Lemma 2.5. Let $a_j, b_j(j = 0, 1, \cdots, p), c, d$ and $T$ be nonnegative constants. Let $N$ be a positive integer and $h = T/N$ .Suppose that the sequence of positive numbers $\{e_{n}^{(k)} > 0, n = 0, 1, \cdots N, k = 0, 1, \cdots\}$ satisfies

$\begin{equation} e^{(k+1)}_{n+1}\leq e^{(k+1)}_{n}+h\sum\limits_{j = 0}^p \left(a_je^{(k+1)}_{n-j}+b_je^{(k)}_{n-j}\right)+che^{(k)}_{n+1}+dh, n = p, p+1, \cdots, N-1. \end{equation}$

(2.13)

Then for any $0 < h < \dfrac{1}{2c}$

$\max\limits_{\begin{subarray}{c} 0\leq k, p+1\leq n\leq N\end{subarray}}e^{(k)}_n\leq \mathrm{e}^{2(A+c)T}\varepsilon+\dfrac{d}{A+c}\left(\mathrm{e}^{2(A+c)T}-1\right),$

where $\varepsilon = \max\left\{ \max\limits_{0\leq n\leq N}e^{(0)}_n, \max\limits_{\begin{subarray}{c} 0\leq k, 0\leq n\leq p \end{subarray}}e^{(k)}_n\right\}$ , $A = \sum\limits_{j = 0}^p (a_j+b_j)$ .

Proof. Note that $e^{(k)}_{n}\leq \varepsilon$ for all $n\leq p$ and all $k\geq 0$ . By using Eq (2.13) repeatedly we have

$\begin{equation} \begin{aligned} e^{(k+1)}_{p+1}\leq &(1+ch+\cdots+(ch)^k)(1+Ah)\varepsilon +(ch)^{k+1}\varepsilon\\ &+(1+ch+\cdots+(ch)^k)dh, k = 0, 1, \cdots. \end{aligned} \end{equation}$

(2.14)

Let $0 < h < \dfrac{1}{2c}$ . With the notation $\alpha_h = \dfrac{1+Ah}{1-ch}$ and $\beta_h = \dfrac{dh}{1-ch}$ we can derive from Eq (2.14)

$e^{(k)}_{p+1}\leq \alpha_h\varepsilon+\beta_h, k = 0, 1, \cdots.$

Noting that $\varepsilon\leq \alpha_h\varepsilon+\beta_h$ and therefore $e^{(k)}_{n}\leq \alpha_h\varepsilon+\beta_h$ for all $n\leq p+1$ and all $k\geq 0$ we thus have by repeating the above process

$e^{(k)}_{p+2}\leq \alpha_h(\alpha_h\varepsilon+\beta_h)+\beta_h = (\alpha_h)^2\varepsilon+\beta_h(1+\alpha_h), k = 0, 1, \cdots.$

Consequently, we get

$e^{(k)}_{p+m}\leq (\alpha_h)^m\varepsilon+\beta_h(1+\alpha_h+(\alpha_h)^2+\cdots+(\alpha_h)^{m-1}), k = 0, 1, \cdots$

hold true for $m = 1, 2, \cdots, N-p$ . Note that $\alpha_h\geq 1$ . Thus

$e^{(k)}_n \leq (\alpha_h)^N\varepsilon+\beta_h(1+\alpha_h+(\alpha_h)^2+\cdots+(\alpha_h)^{N-1})$

for $k = 0, 1, \cdots$ , $n = 0, 1, \cdots, N$ . This together with $0 < ch < 1/2$ yield

$\begin{aligned} e^{(k)}_n \leq & (\alpha_h)^N\varepsilon+\dfrac{(\alpha_h)^N-1}{\alpha_h-1}\beta_h\\ \leq &(1+2(A+c)h)^{T/h}\varepsilon+\dfrac{d}{A+c}\left((1+2(A+c)h)^{T/h}-1\right)\\ \leq &\mathrm{e}^{2(A+c)T}\varepsilon+\dfrac{d}{A+c}\left(\mathrm{e}^{2(A+c)T}-1\right) \end{aligned}$

for $k = 0, 1, \cdots, n = 0, 1, \cdots, N$ . The proof is complete.

3. Main results

Theorem 3.1. Suppose that there exist constants $L_1$ , $L_2$ , $L_3$ and $L_4$ such that

$\begin{equation} \begin{aligned} &\|\Phi(t, h, x_1, x_2, x_3, x_4)-\Phi(t, h, y_1, y_2, y_3, y_4)\|\leq \sum\limits_{i = 1}^{4} L_i\|x_i-y_i\| \end{aligned} \end{equation}$

(3.1)

for any real numbers $t, h\in \mathbb{R}, x_i, y_i\in \mathbb{R}^{d}, i = 1, 2, 3, 4$ . Then the WR Eq (2.2) is zero-stable.

Proof. Let $\varepsilon^{(k)}_n = \|\tilde{x}^{(k)}_n-x^{(k)}_n\|$ and for any $\varepsilon > 0$ the perturbations in Eq (2.6) $\{\delta^{(k)}_n, k = 0, 1, \cdots, n = 0, 1, \cdots, N\}$ satisfies $\max\limits_{\begin{subarray}{c}0\leq k, 0\leq n\leq N\end{subarray}}\|\delta^{(k)}_n\| < \varepsilon$ . Then by virtue of Eqs (2.2) and (2.6) and the condition (3.1) we have

$\begin{equation} \varepsilon^{(k+1)}_{n+1}\leq \varepsilon^{(k+1)}_n +h_n\left( L_1\varepsilon^{(k+1)}_{n+1}+L_2 \varepsilon^{(k+1)}_n+L_3 \varepsilon^{(k)}_{n+1}+L_4\varepsilon^{(k)}_n\right)+h_n\varepsilon \end{equation}$

(3.2)

for $n = 0, 1, \cdots, N-1, k = 0, 1, \cdots$ and

$\begin{equation} \varepsilon^{(k+1)}_0\leq \varepsilon, k = 0, 1, \cdots, \varepsilon^{(0)}_n\leq \varepsilon, n = 0, 1, \cdots, N. \end{equation}$

(3.3)

Let $h_n < \dfrac{1}{2L_1}$ . Then $\dfrac{1}{1-h_nL_1} < 2$ . We therefore derive from Eq (3.2)

$\begin{equation} \varepsilon^{(k+1)}_{n+1}\leq (1+2h_n(L_1+L_2))\varepsilon^{(k+1)}_{n}+2h_nL_3\varepsilon^{(k)}_{n+1}+2h_nL_4\varepsilon^{(k)}_{n}+2h_n\varepsilon \end{equation}$

(3.4)

for $n = 0, 1, \cdots, N-1, k = 0, 1, \cdots$ . Consequently, by Eqa (3.3), (3.4) and Lemma 2.4 we have

$\varepsilon^{(k)}_{n}\leq \mathrm{e}^{4(L_1+L_2+L_3+L_4)T}\varepsilon+\dfrac{\varepsilon}{L_1+L_2+L_3+L_4}\left(\mathrm{e}^{4(L_1+L_2+L_3+L_4)T}-1\right)$

for any $k = 0, 1, \cdots, n = 0, 1, \cdots, N$ if $0 < h < \min\left\{\dfrac{1}{2L_1}, \dfrac{1}{4L_3}\right\}$ . The proof is therefore complete.

Theorem 3.2. Suppose that there exist constants $L_i, i = 1, 2, \cdots, 2p+4$ such that

$\begin{equation} \begin{aligned} &\|\Psi(t, h, x_1, x_2, \cdots, x_{2p+4})-\Psi(t, h, y_1, y_2, \cdots, y_{2p+4})\|\leq \sum\limits_{i = 1}^{2p+4}L_i\|x_i-y_i\| \end{aligned} \end{equation}$

(3.5)

for any real numbers $t, h\in \mathbb{R}, x_i, y_i\in \mathbb{R}^d, i = 1, 2, \cdots, 2p+4$ . Then the WR Eq (2.3) and Eq (2.4) is zero-stable.

Proof. Let $\varepsilon^{(k)}_n$ denote $\|\tilde{x}^{(k)}_n-x^{(k)}_n\|$ and the perturbations $\{\delta^{(k)}_n, k = 0, 1, \cdots, n = 0, 1, \cdots, N\}$ in Eq (2.7) satisfy $\max\limits_{\begin{subarray}{c}0\leq k, 0\leq n\leq N\end{subarray}}\|\delta^{(k)}_n\| < \varepsilon$ , where $\varepsilon$ is any positive number. Then by virtue of Eqs (2.3), (2.4), (2.7) and the condition (3.5) we have

$\begin{equation} \begin{aligned} \varepsilon^{(k+1)}_{n+1}\leq & \varepsilon^{(k+1)}_n+h(L_1\varepsilon^{(k+1)}_{n+1}+L_2\varepsilon^{(k+1)}_{n}+\cdots+L_{p+2}\varepsilon^{(k+1)}_{n-p}\\ &+L_{p+3}\varepsilon^{(k)}_{n+1}+L_{p+4}\varepsilon^{(k)}_{n}+\cdots+L_{2p+4}\varepsilon^{(k)}_{n-p})+h\varepsilon, \\ &\qquad\qquad\qquad\qquad\qquad\quad n = p, p+1, \cdots, N-1, \\ \varepsilon^{(k+1)}_{n}\leq & \varepsilon, n = 0, 1, \cdots, p, \varepsilon^{(0)}_n\leq \varepsilon, n = 0, 1, \cdots, N, \end{aligned} \end{equation}$

(3.6)

where $k = 0, 1, \cdots$ . Let $h < \dfrac{1}{2L_1}$ . Consequently, $\dfrac{1}{1-hL_1} < 2$ . We therefore derive from Eq (3.6)

$\begin{equation} \begin{aligned} \varepsilon^{(k+1)}_{n+1}\leq & (1+2h(L_1+L_2))\varepsilon^{(k+1)}_n+2h(L_3\varepsilon^{(k+1)}_{n-1}+\cdots+L_{p+2}\varepsilon^{(k+1)}_{n-p}+L_{p+3}\varepsilon^{(k)}_{n+1}\\ &+L_{p+4}\varepsilon^{(k)}_{n}+\cdots+L_{2p+4}\varepsilon^{(k)}_{n-p})+2h\varepsilon, n = p, p+1, \cdots, N-1, \\ \varepsilon^{(k+1)}_{n}\leq & \varepsilon, n = 0, 1, \cdots, p, \varepsilon^{(0)}_n\leq \varepsilon, n = 0, 1, \cdots, N \end{aligned} \end{equation}$

(3.7)

for $n = 0, 1, \cdots, N-1, k = 0, 1, \cdots$ . By Eq (3.7) and Lemma 2.5 we have

$\varepsilon^{(k)}_{n}\leq \mathrm{e}^{4(L_1+L_2+\cdots+L_{2p+4})T}\varepsilon+\dfrac{\varepsilon}{L_1+L_2+\cdots+L_{2p+4}}\left(\mathrm{e}^{4(L_1+L_2+\cdots+L_{2p+4})T}-1\right)$

for any $k = 0, 1, \cdots, n = 0, 1, \cdots, N$ if $0 < h < \min\left\{\dfrac{1}{2L_1}, \dfrac{1}{4L_{p+3}}\right\}$ . The proof is complete.

Theorem 3.3. For Eqs (2.5) and (2.8) let $\Phi = (\Phi_1^{\mathrm{T}}, \Phi_2^{\mathrm{T}}, \cdots, \Phi_m^{\mathrm{T}})^{\mathrm{T}}$ satisfy the Lipschitz condition:there exist constants $L_1$ and $L_2$ such that

$\begin{equation} \|\Phi(t, h, x_1, x_2)-\Phi(t, h, y_1, y_2)\|\leq \sum\limits_{i = 1}^{2} L_i\|x_i-y_i\|, \;for\;any\; x_1, x_2, y_1, y_2\in \mathbb{R}^d, \end{equation}$

(3.8)

and the interpolation functions $\sideset{_j}{^{(k)}}{\mathop{\tilde{y}}}(t)$ and $\sideset{_j}{^{(k)}}{\mathop{y}}(t)$ satisfy that

$\begin{equation} \sup\limits_{0\leq t\leq T}\|\sideset{_j}{^{(k)}}{\mathop{\tilde{y}}}(t)- \sideset{_j}{^{(k)}}{\mathop{y}}(t)\|\leq C\max\limits_{0\leq n\leq N_j}\|\sideset{_j}{^{(k)}_n}{\mathop{\tilde{x}}}-\sideset{_j}{^{(k)}_n}{\mathop{x}}\|, j = 1, 2, \cdots, m, \end{equation}$

(3.9)

where $C$ is the positive constant. Then the multi-rate Eq (2.5) is weakly zero-stable.

Proof. Define $\|x\|_a = \max\limits_{1\leq i\leq m}\|_ix\|$ , $\|x\|_b = \sum\limits_{i = 1}^m\|_ix\|$ for $x = (\sideset{_1}{^{\rm{T}}}{\mathop{x}}, \sideset{_2}{^{\rm{T}}}{\mathop{x}}, \cdots, \sideset{_m}{^{\rm{T}}}{\mathop{x}})^{\rm{T}}\in \mathbb{R}^d$ , where $_ix\in \mathbb{R}^{d_i}, i = 1, 2, \cdots, m, d_1+d_2+\cdots+d_m = d$ . It is easy to prove that $\|\cdot\|_a$ and $\|\cdot\|_b$ are the norm. Consequently, there exist constants $C_1$ and $C_2$ such that

$\begin{equation} C_1\|x\|_a\leq \|x\|\leq C_2\|x\|_b \;{ \rm{ for\;any } }\;x\in \mathbb{R}^d. \end{equation}$

(3.10)

Eqs (3.8) and (3.10) thus yield

$\begin{aligned} &C_1\|\Phi(t, h, x_1, x_2)- \Phi(t, h, y_1, y_2)\|_a \leq \|\Phi(t, h, x_1, x_2)- \Phi(t, h, y_1, y_2)\| \\ \leq &\sum\limits_{j = 1}^2 L_j\|x_j-y_j\|\leq \sum\limits_{j = 1}^2 L_jC_2\|x_j-y_j\|_b = \sum\limits_{j = 1}^2 L_jC_2\sum\limits_{i = 1}^m\|_ix_j-\sideset{_i}{_j}{\mathop{y}}\| \end{aligned}$

for any $x_1, x_2, y_1, y_2\in \mathbb{R}^d$ and therefore

$\begin{equation} \|\Phi_i(t, h, x_1, x_2)-\Phi_i(t, h, y_1, y_2)\|\leq \sum\limits_{j = 1}^2L_j\dfrac{C_2}{C_1}\sum\limits_{i = 1}^m\|_ix_j-\sideset{_i}{_j}{\mathop{y}}\|. \end{equation}$

(3.11)

With the notation $_ie^{(k)}_n = \|_i\tilde{x}^{(k)}_n-\sideset{_i}{^{(k)}_n}{\mathop{x}}\|$ we can derive from Eqs (2.5), (2.8), (3.9) and (3.11)

$\begin{equation} \begin{aligned} _ie^{(k+1)}_{n+1}\leq &\sideset{_i}{^{(k+1)}_n}{\mathop{e}}+_ih_n \dfrac{L_1C_2C}{C_1}\left(\max\limits_{0\leq j\leq N} \sideset{_1}{^{(k)}_j}{\mathop{e}}+\cdots+ \max\limits_{0\leq j\leq N_i}\sideset{_{i-1}}{^{(k)}_j}{\mathop{e}}+\sideset{_i}{^{(k+1)}_{n+1}}{\mathop{e}}\right.\\ &+\left.\max\limits_{0\leq j\leq N_i} \sideset{_{i+1}}{^{(k)}_j}{\mathop{e}}+\cdots+\max\limits_{0\leq j\leq N_i}\sideset{_m}{^{(k)}_j}{\mathop{e}}\right)+ \sideset{_i}{_n}{\mathop{h}} \dfrac{L_2C_2C}{C_1}\left(\max\limits_{0\leq j\leq N_i} \sideset{_1}{^{(k)}_j}{\mathop{e}}\right.\\ &\left.+\cdots+\max\limits_{0\leq j\leq N_i}\sideset{_{i-1}}{^{(k)}_j}{\mathop{e}}+\sideset{_i}{^{(k+1)}_{n}} {\mathop{e}}+\max\limits_{0\leq j\leq N_i} \sideset{_{i+1}}{^{(k)}_j}{\mathop{e}}+\cdots+\max\limits_{0\leq j\leq N_i}\sideset{_m}{^{(k)}_j}{\mathop{e}}\right)\\ &+\sideset{_i}{_n}{\mathop{h}} \|\sideset{_i}{^{(k+1)}_{n+1}}{\mathop{\delta}}\|. \end{aligned} \end{equation}$

(3.12)

Let $h = \max\limits_{\begin{subarray}{c}1\leq i\leq m \end{subarray}}\max\limits_{0\leq n\leq N_i} \sideset{_i}{_n}{\mathop{h}} < \dfrac{C_1}{2L_1C_2C}$ . Consequently, $_ih_n\dfrac{L_1C_2C}{C_1} < \dfrac12 < 1-\sideset{_i}{_n}{\mathop{h}}\dfrac{L_1C_2C}{C_1}$ which together with Eq (3.12) imply that

$\begin{equation} \begin{aligned} _ie^{(k+1)}_{n+1}\leq &\left(1+a \sideset{_i}{_n}{\mathop{h}}\right) \sideset{_i}{^{(k+1)}_n}{\mathop{e}}+\left(a\sum\limits_{\begin{subarray}{c}l = 1\\ l\neq i\end{subarray}}^m\max\limits_{0\leq j\leq N_i} \sideset{_l}{^{(k)}_j}{\mathop{e}}+2\|\sideset{_i}{^{(k+1)}_{n+1}}{\mathop{\delta}}\|\right)\sideset{_i}{_n}{\mathop{h}}\\ \leq &(1+a \sideset{_i}{_n}{\mathop{h}}) \sideset{_i}{^{(k+1)}_n}{\mathop{e}}+ \left(a\sum\limits_{i = 1}^m\max\limits_{0\leq j\leq N_i} \sideset{_i}{_j^{(k)}}{\mathop{e}}+2\max\limits_{\begin{subarray}{c}1\leq i\leq m \end{subarray}} \max\limits_{0\leq j\leq N_i}\|\sideset{_i}{_j^{(k+1)}}{\mathop{\delta}}\| \right)\sideset{_i}{_n}{\mathop{h}} \end{aligned} \end{equation}$

(3.13)

with $a = \dfrac{2(L_1+L_2)C_2C}{C_1}$ . By using Eq (3.13) with $k = 0$ and Lemma 2.3 we get

$\begin{aligned} _ie^{(1)}_n\leq &\mathrm{e}^{aT}\sideset{_i}{^{(1)}_0}{\mathop{e}}+\dfrac{1}{a}({\mathrm{e}}^{aT}-1) \left(a\sum\limits_{i = 1}^m\max\limits_{0\leq j\leq N_i} \sideset{_i}{_j^{(0)}}{\mathop{e}}+2\max\limits_{\begin{subarray}{c}1\leq i\leq m\end{subarray}} \max\limits_{0\leq j\leq N_i}\|\sideset{_i}{_j^{(1)}}{\mathop{\delta}}\| \right). \end{aligned}$

Thus

$\begin{aligned} \max\limits_{\begin{subarray}{c}1\leq i\leq m\end{subarray}}\max\limits_{0\leq n\leq N_i}\sideset{_i}{^{(1)}_n}{\mathop{e}} \leq a_1\max\left(\max\limits_{1\leq i\leq m} \sideset{_i}{^{(1)}_0}{\mathop{e}}, \sum\limits_{i = 1}^m\max\limits_{0\leq j\leq N_i}\sideset{_i}{_j^{(0)}}{\mathop{e}}, \max\limits_{\begin{subarray}{c}1\leq i\leq m\end{subarray}}\max\limits_{0\leq j\leq N_i}\|\sideset{_i}{_j^{(1)}}{\mathop{\delta}}\| \right) \end{aligned}$

with $a_1 = {\mathrm{e}}^{aT}+\dfrac{1}{a}({\mathrm{e}}^{aT}-1)(a+2)$ . Similarly, using the above inequality, Eq (3.13) with $k = 1$ and Lemma 2.3 we have

$\begin{aligned} \max\limits_{\begin{subarray}{c}1\leq i\leq m\end{subarray}}\max\limits_{0\leq n\leq N_i}\sideset{_i}{^{(2)}_n}{\mathop{e}} \leq &a_2\max\left(\max\limits_{ \begin{subarray}{c}1\leq i\leq m\\ 1\leq l\leq 2\end{subarray}} \sideset{_i}{^{(l)}_0}{\mathop{e}}, \sum\limits_{i = 1}^m\max\limits_{0\leq j\leq N_i}\sideset{_i}{_j^{(0)}}{\mathop{e}}, \max\limits_{\begin{subarray}{c}1\leq i\leq m\\1\leq l\leq 2\end{subarray}} \max\limits_{0\leq j\leq N_i}\|\sideset{_i}{_j^{(l)}}{\mathop{\delta}}\| \right) \end{aligned}$

with $a_2 = {\mathrm{e}}^{aT}+({\mathrm{e}}^{aT}-1)(ma_1+\dfrac{2}{a})$ . Repeating the above process yields that for any $k$ there exists the constant $a_k$ such that

$\begin{aligned} &\max\limits_{\begin{subarray}{c}1\leq i\leq m\\ 0\leq l\leq k \end{subarray}} \max\limits_{0\leq n\leq N_i}\|\sideset{_i}{^{(l)}_n}{\mathop{\tilde{x}}}-\sideset{_i}{^{(l)}_n}{\mathop{x}}\| \\ \leq & a_k\max\left(\max\limits_{\begin{subarray}{c}1\leq i\leq m\\ 0\leq l\leq k\\ \end{subarray}}\|\sideset{_i}{^{(l)}_0}{\mathop{\delta}}\|, \sum\limits_{i = 1}^m \max\limits_{0\leq n\leq N_i}\|\sideset{_i}{^{(0)}_n}{\mathop{\delta}}\|, \max\limits_{\begin{subarray}{c}1\leq i\leq m\\0\leq l\leq k \end{subarray}}\max\limits_{0\leq n\leq N_i}\|\sideset{_i}{^{(l)}_n}{\mathop{\delta}}\| \right) \end{aligned}$

for any mesh $\{_it_n, n = 0, 1, \cdots, N, i = 1, 2, \cdots, m\}$ . The proof is therefore complete.

Remark 1. In the above discussions we assume that the perturbation at one step is $\delta_n^{(k)} h$ , which depends on step-size $h$ . This assumption is necessary for the proof of the theorems mentioned above. However it seems to be more reasonable to regard $\delta_n^{(k)}$ , a small number, as the perturbation, if the step-size is very small. In this case the tiny perturbation at each step may bring out the huge change of the numerical solution. We will explore these in next section by some computer experiments.

Remark 2. Note that the convergence property of the WR methods is not be used in the proof of Theorem 3.1-3.3. Hence, a divergent WR method may be zero-stable. In other words, zero-stability does not imply convergence.

4. Numerical experiments

In this section some numerical experiments are performed to verify the theorems obtained in Section 3.

Consider the equation

$\begin{equation} x' = Ax = \left( \begin{array}{ccc} T_1 & I & 0 \\ I & T_2 & I \\ 0 & I & T_3 \\ \end{array} \right)x, t\in [0, 1], x(0) = x_0, \end{equation}$

(4.1)

where $x\in \mathbb{R}^{15}$ , $T_i = \left(\begin{array}{cccc} -2^i & 1 & & \\ 1 & -2^i & \ddots & \\ & \ddots & \ddots & 1\\ & & 1 & -2^i\\ \end{array} \right)\in \mathbb{R}^{5\times 5}$ and $I$ is the identity matrix. The special case of such system has been examined by Burrage ^[11], which is obtained by discretizing the heat equation in two variables.

Let $A_1 = \left(\begin{array}{ccc} T_1 & & \\ & T_2 & \\ & & T_3 \\ \end{array} \right)$ and $A_2 = A-A_1$ . First, choose the following WR methods of solving the Eq (4.1):

(i) Euler method with variable step-size

$\begin{aligned} &x^{(k+1)}_{n+1} = x^{(k+1)}_n+h_n(A_1x^{(k+1)}_n+A_2x^{(k)}_n), n = 0, 1, \cdots, N-1, k = 0, 1, \cdots, \\ &x^{(k+1)}_0 = x_0\;{ \rm{ for\;all } }\;k, x^{(0)}_n = x_0\;{ \rm{ for\;all } }\;n. \end{aligned}$

(ii) Two step methods with fixed step-size

$\begin{aligned} &x^{(k+1)}_{n+1} = x^{(k+1)}_n+\dfrac{h}{2}\left[3(A_1x^{(k+1)}_n+A_2x^{(k)}_n)-(A_1x^{(k+1)}_{n-1}+A_2x^{(k)}_{n-1})\right], \\ &\qquad \qquad \qquad \qquad \qquad \qquad n = 1, 2, \cdots, N-1, k = 0, 1, \cdots, \\ &x^{(k+1)}_0 = x_0\;{ \rm{ for\;all } }\;k, x^{(0)}_n = x_0\;{ \rm{ for\;all } }\;n, \end{aligned}$

where $x^{(k+1)}_1$ is obtained by a suitable one-step method. Here let

$x^{(k+1)}_1 = x_0+hAx_0+\dfrac{h^2}{2}A^2x_0.$

(iii) Multi-rate Jacobi methods based on Euler method

$\left. \begin{aligned} &&_ix^{(k+1)}_{n+1} = \sideset{_i}{_{n}^{(k+1)}}{\mathop{x}}+ \sideset{_i}{_{n}}{\mathop{h}}\left(T_i\cdot \sideset{_i}{_{n+1}^{(k+1)}}{\mathop{x}} +\sideset{_{i-1}}{_{n+1}^{(k)}}{\mathop{x}}+\sideset{_{i+1}}{_{n+1}^{(k)}}{\mathop{x}}\right), \\ && n = 0, 1, \cdots, N_i-1, \\ &&\sideset{_i}{_{0}^{(k+1)}}{\mathop{x}} = \sideset{_i}{_{n}^{(0)}}{\mathop{x}} = \sideset{_i}{_{0}}{\mathop{x}}, \sideset{_0}{_{n}^{(k)}}{\mathop{x}} = \sideset{_4}{_{n}^{(k)}}{\mathop{x}} = (0, 0, 0, 0, 0)^{\rm{T}}\;{ \rm{ for\;all} }\;k\;{ \rm{ and } }\;n \end{aligned}\right\} i = 1, 2, 3.$

Second, assume the initial values and the approximation solution at each calculation step have tiny perturbations as in Eqs (2.6)–(2.8) and let $\delta^{(k)}_n\equiv \Delta = \delta\cdot (1, 1, \cdots, 1)^{\rm{T}}\in \mathbb{R}^{15}$ , $_i\delta^{(k)}_n\equiv \Delta_p = \delta\cdot (1, 1, 1, 1, 1)^{\rm{T}}\in \mathbb{R}^5$ , $\tilde{x}^{(k)}_n$ denote the resulting perturbed solution and $u^{(k)}_n$ denote $\tilde{x}^{(k)}_n-x^{(k)}_n$ , which yields

$\begin{equation} \begin{aligned} &u^{(k+1)}_{n+1} = u^{(k+1)}_n+h_n(A_1u^{(k+1)}_n+A_2u^{(k)}_n)+h_n\Delta, n = 0, 1, \cdots, N-1, k = 0, 1, \cdots, \\ &u^{(k+1)}_0 = \Delta\;{ \rm{ for\;all } }\;k, u^{(0)}_n = \Delta\;{ \rm{ for\;all } }\;n, \end{aligned} \end{equation}$

(4.2)

$\begin{equation} \begin{aligned} &u^{(k+1)}_{n+1} = u^{(k+1)}_n+\dfrac{h}{2}\left[3(A_1u^{(k+1)}_n+A_2u^{(k)}_n)-(A_1u^{(k+1)}_{n-1}+A_2u^{(k)}_{n-1})\right]+\dfrac{h}{2}\Delta, \\ &\qquad \qquad \qquad \qquad \qquad \qquad n = 1, 2, \cdots, N-1, k = 0, 1, \cdots, \\ &u^{(k+1)}_0 = \Delta, u^{(k+1)}_1 = (I+hA+\dfrac{h^2}{2}A^2)\Delta, \;{ \rm{ for\;all } }\;k, u^{(0)}_n = \Delta\;{ \rm{ for\;all } }\;n, \end{aligned} \end{equation}$

(4.3)

and

$\begin{array}{l} \left. \begin{array}{r} _iu^{(k+1)}_{n+1} = \sideset{_i}{_{n}^{(k+1)}}{\mathop{u}}+ \sideset{_i}{_{n}}{\mathop{h}}\left(T_i\cdot \sideset{_i}{_{n+1}^{(k+1)}}{\mathop{u}} +\sideset{_{i-1}}{_{n+1}^{(k)}}{\mathop{u}}+\sideset{_{i+1}}{_{n+1}^{(k)}}{\mathop{u}}+\Delta_p\right), \\ n = 0, 1, \cdots, N_i-1, \\ \sideset{_i}{_{0}^{(k+1)}}{\mathop{u}} = \sideset{_i}{_{n}^{(0)}}{\mathop{u}} = \Delta_p, \sideset{_0}{_{n}^{(k)}}{\mathop{u}} = \sideset{_4}{_{n}^{(k)}}{\mathop{u}} = (0, 0, 0, 0, 0)^{\rm{T}}, \forall k, \forall n \end{array}\right\} i = 1, 2, 3. \end{array}$

(4.4)

Lastly, reach a reasonable conclusion by analysing data generated by Eqs (4.2)–(4.4).

We choose $A$ , $A_1$ , $A_2$ and $\delta^{(k)}_n$ as above to guarantee that $u^{(k)}_n$ increase with $k$ and $n$ that is the worst case of error propagation.

In our experiments the step-sizes $h_n$ in Eq (4.2) and $_ih_n$ in Eq (4.4) are taken as

$h_n = \min\left\{10h, \max\left\{0.1h, |\xi_n|h\right\}\right\}$

and

$_ih_n = \min\{10^ih, \max\{10^{i-2}h, 10^{i-1}|\xi_n|h\}\},$

where $h > 0$ , $\{\xi_0, \xi_1, \cdots\}$ is a sequence of independent and identically distributed random variables satisfying $\xi_i\sim N(0, 1)$ .

The data in , and show clearly the errors resulted by the perturbations $h_n\Delta(h\Delta/2)$ in Eq (4.2) (Eq (4.3) are controllable as $k$ and $n$ tend to infinity which support the theorems developed in Section 3). However the data in and show that the errors increase with $k$ and $1/h$ when the perturbations $h_n\Delta (h\Delta/2)$ are replaced by the constant perturbation $\Delta(\Delta/2)$ .

Table 3. The computing results of Eq (4.2) with

$\delta = 0.0001$ (

$\varepsilon = \max\limits_{0\leq l\leq k}\max\limits_{0\leq n\leq N}u^{(l)}_n$ ).

$h$	0.1		0.001
$k$	5	10	10	15
$\varepsilon$	3.1483e-04	3.1483e-04	3.1605e-04	3.1605e-04
$h$	0.00001
$k$	10	15
$\varepsilon$	3.1606e-04	3.1606e-04

| Show Table

DownLoad: CSV

Table 4. The computing results of Eq (4.2) with

$\delta = 0.0001$ (

$\varepsilon = \max\limits_{0\leq l\leq k}\max\limits_{0\leq n\leq N}u^{(l)}_n$ ) for the case of replacing perturbation

$h_n \Delta$ with the constant perturbation

$\Delta$ .

$h$	0.1		0.001		0.00001
$k$	5	10	10	15	10	15
$\varepsilon$	0.0018	0.0018	0.1824	0.1824	17.6723	17.6723
$h$	0.000001
$k$	10	15
$\varepsilon$	176.8079	176.8079

| Show Table

DownLoad: CSV

Table 5. The computing results of Eq (4.3) with

$\delta = 0.0001$ (

$\varepsilon = \max\limits_{0\leq l\leq k}\max\limits_{0\leq n\leq N}u^{(l)}_n$ ).

$h$	0.1		0.001
$k$	10	15	10	15
$\varepsilon$	3.1568e-04	3.1568e-04	3.1606e-04	3.1606e-04
$h$	0.00001
$k$	10	15
$\varepsilon$	3.1606e-04	3.1606e-04

| Show Table

DownLoad: CSV

Table 6. The computing results of Eq (4.3) with

$\delta = 0.0001$ (

$\varepsilon = \max\limits_{0\leq l\leq k}\max\limits_{0\leq n\leq N}u^{(l)}_n$ ) for the case of replacing perturbation

$h \Delta/2$ with

$\Delta/2$ .

$h$	0.1		0.001		0.00001
$k$	10	15	10	15	10	15
$\varepsilon$	0.1782	0.1782	0.1435	0.1435	14.1724	14.1724
$h$	0.000001
$k$	10	15
$\varepsilon$	141.7252	141.7252

| Show Table

DownLoad: CSV

Table 7. The computing results of Eq (4.4) with

$\delta = 0.0001$ (

$\varepsilon = \max\limits_{0\leq l\leq k}\max\limits_{0\leq n\leq N}u^{(l)}_n$ ).

$h$	0.1		0.001
$k$	5	10	10	15
$\varepsilon$	3.1085e-04	3.1085e-04	3.1567e-04	3.1567e-04
$h$	0.00001
$k$	10	15
$\varepsilon$	3.1606e-04	3.1606e-04

| Show Table

DownLoad: CSV

5. Further work

It is well known that a linear multi-step method with fixed step-size is convergent if, and only if, it is both consistent and zero-stable. In this paper we only present the sufficient conditions of zero-stability for some special methods. We will explore the relationship between convergence and zero-stability for WR methods based on general linear multi-step methods in the future. Moreover, we shall apply our methods to the numerical study of some inverse problems ^[12,13,14].

Acknowledgments

Research is supported by the Natural Science Foundation of Fujian Province, China (grant number 2021J011031).

Conflict of interest

The author declares that he has no conflicts of interest regarding the publication of this paper.

References

[1]	P. Vennapusa, D. White, S. Schram, Roller-integrated compaction monitoring for hot-mix asphalt overlay construction, J. Transp. Eng., 139 (2013), 1164–1173. https://doi.org/10.1061/(ASCE)TE.1943-5436.0000602 doi: 10.1061/(ASCE)TE.1943-5436.0000602
[2]	T. Chen, T. Ma, X. Huang, S. Ma, S. Wu, Microstructure of synthetic composite interfaces and verification of mixing order in cold-recycled asphalt emulsion mixture, J. Clean. Prod., 263 (2020), 121467. https://doi.org/10.1016/j.jclepro.2020.121467 doi: 10.1016/j.jclepro.2020.121467
[3]	J. Zhu, T. Ma, J. Fan, Z. Fang, T. Chen, Y. Zhou, Experimental study of high modulus asphalt mixture containing reclaimed asphalt pavement, J. Clean. Prod., 263 (2020), 121447. https://doi.org/10.1016/j.jclepro.2020.121447 doi: 10.1016/j.jclepro.2020.121447
[4]	Y. Ma, F. Chen, T. Ma, X. Huang, Y. Zhang, Intelligent compaction: An improved quality monitoring and control of asphalt pavement construction technology, IEEE T. Intell. Transp., 23 (2022), 14875–14822. https://doi.org/10.1109/TITS.2021.3134699 doi: 10.1109/TITS.2021.3134699
[5]	Q. Xu, G. Chang, Evaluation of intelligent compaction for asphalt materials, Autom. Constr., 30 (2013), 104–112. https://doi.org/10.1016/j.autcon.2012.11.015 doi: 10.1016/j.autcon.2012.11.015
[6]	Q. Xu, G. Chang, V. Gallivan, Development of a systematic method for intelligent compaction data analysis and management, Constr. Build. Mater., 37 (2012), 470–480. https://doi.org/10.1016/j.conbuildmat.2012.08.001 doi: 10.1016/j.conbuildmat.2012.08.001
[7]	X. Zhao, Study on Intelligent Compaction Control Technology of Subgrade, M.S. thesis, Chang'an University in Xi'an, 2016. https://doi.org/10710-2013521093
[8]	B. Chen, Research on Real-time Detection System of Subgrade Compaction Degree, M.S. thesis, Chang'an University in Xi'an, 2019. https://doi.org/10710-2016125023
[9]	R. Minchin, H. Thomas, Validation of vibration-based onboard asphalt density measuring system, J. Constr. Eng. M., 129 (2003), 1–7. https://doi.org/10.1061/(ASCE)0733-9364(2003)129:1(1) doi: 10.1061/(ASCE)0733-9364(2003)129:1(1)
[10]	H. Zhao, Study on the Evaluation Index of Roadbed Compaction Quality, M.S. thesis, Chang'an University in Xi'an, 2015. https://doi.org/10710-2013225045
[11]	Editorial Department of China Journal of Highway and Transport, Review on China's Pavement Engineering Research·2020, China J. Highw. Transp., 33 (2020), 1–66. https://doi.org/10.19721/j.cnki.1001-7372.2020.10.001
[12]	H. Cui, Based on Intelligent Compaction Technology of Filling Roadbed Compaction Degree Test Research, M.S. thesis, Hebei University in Shijiazhuang, 2017. https://doi.org/10075-20151780
[13]	F. Hao, Finite Element Analysis of "Vibration Wheel-Soil" Model, M.S. thesis, Chang'an University in Xi'an, 2007. https://doi.org/10710-20040455
[14]	Z. Wu, S. Zhang, L. Guo, Application of ABAQUS secondary development in rock breaking simulation of PDC cutter, J. Xi'an Shiyou Univ., 35 (2020), 104–109. https://doi.org/10.3969/j.issn.1673-064X.2020.01.015 doi: 10.3969/j.issn.1673-064X.2020.01.015
[15]	D. Liu, M. Lin, S. Li, Real-time quality monitoring and control of highway compaction, Autom. Constr., 62 (2016), 114–123. https://doi.org/10.3876 /j.issn.1000-1980.2018.04.005 doi: 10.3876/j.issn.1000-1980.2018.04.005
[16]	Q. Zhang, T. Liu, Z. Zhang, Z. Huangfu, Q. Li, Z. An, Unmanned rolling compaction system for rockfill materials, Autom. Constr., 100 (2019), 103–117. https://doi.org/10.1016/j.autcon.2019.01.004 doi: 10.1016/j.autcon.2019.01.004
[17]	Z. Wu, S. Zhang, L. Guo, W. Wang, Y. Pan, Application of ABAQUS secondary development in rock breaking simulation of PDC cutter, J. Xi'an Shiyou Univ., 35 (2020), 104–109. https://doi.org/10.3969/j.issn.1673-064X.2020.01.015 doi: 10.3969/j.issn.1673-064X.2020.01.015
[18]	F. Chen, C. Wang, W. Li, J. Yang, Application of Abaqus secondary development in shot peening strengthening of aerospace arc-shaped frame, Comp. Aided Eng., 29 (2020), 55–60. https://doi.org/10.13340/j.cae.2020.02.011 doi: 10.13340/j.cae.2020.02.011
[19]	W. Hu, X. Jia, X. Zhu, A. Su, B. Huang, Influence of moisture content on intelligent soil compaction, Autom. Constr., 113 (2020), 103141. https://doi.org/10.1016/j.autcon.2020.103141 doi: 10.1016/j.autcon.2020.103141
[20]	Y. Ma, Y. Luan, W. Zhang, Y. Zhang, Numerical simulation of intelligent compaction for subgrade construction, J. Cent. South Univ., 27 (2020), 2173–2184. https://doi.org/10.1007/s11771-020-4439-2 doi: 10.1007/s11771-020-4439-2
[21]	Y. Ma, Z. Fang, T. Han, S. Wang, B. Li, Dynamic simulation and evolution of key control parameters for intelligent compaction of subgrade, J. Cent. South Univ., 52 (2021), 2246–2257. https://doi.org/10.11817/j.issn.1672-7207.2021.07.012 doi: 10.11817/j.issn.1672-7207.2021.07.012
[22]	Y. Ma, Y. Zhang, W. Zhao, X. Ding, Z. Wang, T. Ma, Assessment of intelligent compaction quality evaluation index and uniformity, J. Transp. Eng. B-Pave., 2 (2022), 04022024. https://doi.org/10.1061/JPEODX.0000368 doi: 10.1061/JPEODX.0000368
[23]	X. Teng, Numerical Analysis and Quality Control of Dynamic Consolidation of Silty Soil Subgrade –in Yellow River Alluvial Plain, M.S. thesis, Shandong University in Jinan, 2017. https://doi.org/10422-201413217
[24]	X. Yan, Some Studies on Functional Linear Regression, Ph.D. thesis, East China Normal University in Shanghai, 2020. https://doi.org/10269-52164404001
[25]	Q. Xu, The Research on Non-Linear Regression Analysis Methods, M.S. thesis, Hefei University of Technology in Hefei, 2009. https://doi.org/10359-0631111370
[26]	D. Zhou, Research on Correlation Optimization of Differential Privacy Regression Analysis Based on Laplace Mechanism, M.S. thesis, Heilongjiang University in Heilongjiang, 2018. https://doi.org/10212-2151237
[27]	D. White, P. Vennapusa, H. Gieselman, Field assess uthorment and specification review for roller-integrated compaction monitoring technologies, Adv. Civ. Eng., 2011, (2011), 1–15. https://doi.org/10.1155/2011/783836 doi: 10.1155/2011/783836

Reader Comments

Your name:*

Email:*
© 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)