Phys. Scr. 99 (2024) 015247 https://doi.org/10.

1088/1402-4896/ad156e

PAPER

Offset boosting in a memristive hyperchaotic system

RECEIVED
15 October 2023
REVISED
Xin Zhang1 , Chunbiao Li1,2,∗ , Tengfei Lei3, Haiyan Fu3 and Zuohua Liu4
5 December 2023 1
School of Electronic and Information Engineering, Nanjing University of Information Science & Technology, Nanjing 210044, People’s
ACCEPTED FOR PUBLICATION Republic of China
2
13 December 2023 School of Artificial Intelligence, Nanjing University of Information Science & Technology, Nanjing 210044, People’s Republic of China
PUBLISHED
3
Collaborative Innovation Center of Memristive Computing Application (CICMCA), Qilu Institute of Technology, Jinan 250200, People’s
29 December 2023 Republic of China
4
State Key Laboratory of Coal Mine Disaster Dynamics and Control, Chongqing University, Chongqing, 400044, People’s Republic of
China
∗
Author to whom any correspondence should be addressed.
E-mail: [email protected] and [email protected]

Keywords: memristive, hyperchaotic, amplitude control, offset boosting

Abstract
In this article, an autonomous memristive hyperchaotic system with multi-dimensional offset
boosting is constructed and analyzed. Besides this, the oscillation can be rescaled by an independent
controller in the memristor. Two independent constants are obtained for offset boosting with one or
two variables, which provide two modes of offset boosting, including single control and synchronous
reverse control. In addition, the offset of the variables is also modified by the system bifurcation
parameters or combined with amplitude control. The multistability can also be identified according to
the offset boosting. Finally, circuit implementation based on PCB is proposed to confirm the
numerical simulations.

1. Introduction

Over the recent years, as a nonlinear two-terminal component with the intrinsic memory effort that connects
electric charge with magnetic flux [1], the memristor has aroused considerable interest in various applications
such as logic gates [2, 3], reservoir computing [4, 5], neuromorphic computing [6, 7]. In particular, the particular
advantages of this device offer additional flexibility for generating chaos [8, 9], hyperchaos [10, 11] and complex
dynamics [12, 13]. Many chaotic or hyperchaotic oscillators have been constructed based on the existing chaotic
system or circuit topology and then replaced some existing linear or nonlinear components with a memristor.
These studies about memristors can be divided into three categories as a whole. The first category is based on
the mathematical expression of differential equations, and then deduces the dynamics through numerical
simulation [14–16]. Bao et al [17] established a new memristive hyperchaotic system with hidden attractors and
then explained that the existence of extreme multistability closely depends on the initial condition of the
memristor. Lin et al [18] proposed a hyperchaotic HNN model by using a memristor as the synapse, where the
system dynamics show complex behaviors, changing from periodic motions to chaos and hyperchaos. Zhang
et al [19] designed a chaotic system by introducing a memristor, where the system shows complex dynamics,
such as amplitude control and coexisting attractors. The second category is to replace the components in the
circuit with the memristor built in the form of the equivalent circuit to obtain the dynamics of circuit oscillation
based on the existing classical circuit. Bao et al [20] designed a 5D hyperchaotic Chua’s circuit with double
memristors, and then analyzed the 3D dimensionality decreasing model, the complex and sensitive
multistability phenomenon was deeply explored. Kengne et al [21] proposed a memristive Duffing oscillator by
replacing the analog multipliers with a memristor and then found the multi-scroll attractors. Besides simulating
the memristor based on an equivalent circuit, chaotic dynamics can also be generated based on a physical
memristor device, confirming the practical feasibility of the approach. The third category is the introduction of
physical memristor devices into existing circuits. Minati et al [22] designed an autonomous chaotic oscillator
circuit suitable for generating chaos by using a physical memristor as the sole source of nonlinearity. Yuan et al

© 2023 IOP Publishing Ltd

Phys. Scr. 99 (2024) 015247 X Zhang et al

[23] effectuated a series of procedures from manufacturing and modeling of memristor to its application in
chaos-based circuits.
The latest research on memristor chaotic systems has shifted from dynamics to geometry control, including
amplitude control [24, 25], and offset boosting [26]. Further research shows that some chaotic systems with non-
bifurcation parameters can readjust the amplitude or polarity of signals without modifying the Lyapunov
exponents, which saves more modulators and meets some practical requirements of simplicity and adjustability
for engineering applications. In addition, geometric control can not only alter the signals but also detect the
multistability of the systems. Usually, when the system is rescaling or changing polarity, the initial conditions
need to change with the basin of attractions, fixed initial conditions can trigger different stable states in the
basin [27].
Geometric control is an effective method of realizing the arbitrary repositioning of the dynamics. There are
two methods to manipulate the attractor arbitrarily in phase space. The first type is the initial-condition-based
offset boosting. The introduced offset function in the dynamic system can trigger the multistability by the
different initial conditions. Such as doubling coexisting [28], dynamics editing [29], initial condition-dominated
offset boosting [30–32]. However, it is not easy to design a physical chaotic oscillator with such characteristics,
because the physical devices are evidently different from the mathematical model, so it is very hard to effectively
preconfigure the initial voltage on the capacitor in the chaotic circuit. Therefore, the arbitrary relocation of the
dynamics based on the initial values has great uncertainty in the circuit. Most systems are implemented based on
digital circuits or analog simulation [33–36]. The second type is system parameters-based offset boosting,
introducing an additional constant into the dynamic system that can move the attractors to any position in the
phase space without changing the dynamics. Compared with the infinitely many attractors triggered by the
initial values, the distance between boosting attractors generated by the parameter is continuous and
indistinguishable.
For these reasons, in this work, a 4D memristive hyperchaotic systems was designed with free geometric
control. Compared with most existing systems with only one-dimensional offset boosting [24, 26, 27, 33], the
newly designed model is the first to propose a multi-dimensional offset boosting with amplitude control, which
has never been achieved in previous literatures, and it also exhibits the characteristic of hyperchaos. More
features are highlighted in the following:

(1) A memristive hyperchaotic system is proposed, and it maintains the characteristics of hyperchaos.
(2) Partial amplitude control of all the proposed systems can be obtained through a single knob in the
memristor.
(3) The system has the property of offset boosting. It can be divided into two kinds, one can control the offset of
one-dimensional signal, and another can control the offset of multi-dimensional signals.
(4) An analog hardware experiment is constructed to show the example of boosting hyperchaotic attractors in a
physical environment.

The rest of this article is organized as follows. In section 2, the proposed memristive hyperchaotic system is
described, where hyperchaotic oscillations are presented. Section 3 analysis of basic dynamics. Section 4 studies
amplitude control and multi-dimensional offset boosting. In section 5, coexisting attractors diagnosed by offset
boosting are explored. An exemplary analog circuit is designed and implemented in section 6. Conclusions and
discussion are given in the last section.

2. Memristive hyperchaotic system description

Offset boosting has become a research hotspot recently. Hence, we are interested in extending this concept and
proposing a new memristive hyperchaotic oscillator with multi-dimensional offset boosting that has never been
reported before. Based on the equation proposed in [25], we see that the variable y appears only twice in the
equation, so we can construct a memristive chaotic system with a two-dimensional offset boosting using y as an
intermediate variable. A newly memristive hyperchaotic system is designed:

⎧ x = a (x - y ) - i - u
⎪ y = b (x - y )
(1)
⎨ z = - ez + x 2
⎪u = kx
⎩

2
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 1. Memristor restriction function and pinched hysteresis loops with different frequencies.

It can be seen from equation (1) that the properties of amplitude control are only affected by the current i of the
memristor. When memristor elements are introduced, the function of amplitude depends on the constraint
relation from the memristor. Therefore, the amplitude control characteristics and the simplicity of the
hyperchaotic circuit of the system can be maintained by reasonable construction of the internal function of the
memristor. Among the currently used memristor models, absolute value memristor are simpler in terms of
circuit structure and it is easy to construct hyperchaotic behaviors, so in this system, we select the classical
absolute value voltage-controlled memristor and modify the feedback terms to obtain the amplitude control
characteristics, the introduced memristor is defined:

⎧ i = W (z ) x
W (z ) = - 5 + 4∣ z ∣ (2)
⎨
⎩ z = - ez + x
2

In order to study the electrical characteristics of this memristor model, the parameter e is set to 0.4, and then
suppose a voltage signal v = Vm sin(2πft) with amplitude Vm and frequency f is introduced into the embedded
memristor. When fixing the amplitude Vm = 1 and the initial state z(0) = 0, the pinched hysteresis loops
obtained by adjusting different frequencies f are shown in figure 1.
As can be seen from the figure, all three hysteresis loops cross the origin of the v-i plane. In addition, when
the frequency increases, the area of the hysteresis loop of the memristor gradually shrinks and eventually
converges to a straight line. It can be seen that the model exhibits the three fingerprint characteristics of a
memristor, indicating that the mathematical model (2) is a memristor.

3. Basic dynamics

3.1. Analysis of equilibria

The equilibria of equation (1) can be obtained by:

⎧ a (x - y ) - x ( - 5 + 4∣ z ∣) - u = 0
⎪ b (x - y ) = 0
(3)
⎨- ez + x 2 = 0
⎪
⎩ kx = 0
Then we can solved that equation (3) only has an origin equilibrium point S0 (0, 0, 0, 0), In fact, all the proposed
eight systems have only one origin equilibrium point. By linearizing equation (1) at zero equilibrium S0, we can
obtain the Jacobian matrix:

⎡a +5 - a 0 - 1⎤
J=⎢ b -b 0 0 ⎥ (4)
⎢ 0 0 -e 0 ⎥
⎢
⎣ k 0 0 0⎥ ⎦
S
We get the characteristic equation at 0 :
det (lI - J ) = (l + e )(l3 + a1 l2 + a2 l + a3) = 0 (5)

3
Phys. Scr. 99 (2024) 015247 X Zhang et al

Where:

a1 = b - a - 5
a2 = k - 5b
a3 = bk (6)

If the Routh-Hurwitz conditions needs to be satisfied, then the coefficients of equation (6) should be:

a1 > 0
a3 > 0
a1 a2 - a3 > 0 (7)

We can get:

b>a+5
bk > 0
(k - 5b)(b - a - 5) - bk > 0 (8)

It is found that when the parameters of equation (1) are all positive, a1 a2 - a3 > 0 is always negative, so the
Routh- Hurwitz conditions are not satisfied. When set a = 4.5, b = 5.5, e = 0.4, and k = 0.2, the eigenvalues at S0
are calculated as:
l1 = - 0.4
l2 = - 3.6217 + 8.5265*10-14i
l3 = 0.0401 - 1.1369*10-13i
l 4 = 7.5817 + 2.8422*10-14i (9)
which implies that S0 is an unstable saddle-focus.

3.2. Bifurcation diagrams and lyapunov exponents spectra

When fixed a = 4.5, b = 5.5, e = 0.4, k = 0.2, the equation (1) can be solved with the algorithm of ode45 in
software MATLAB, and the step size is fixed as 0.005 based on one million calculations. Then the calculated
Lyapunov exponents were (0.1889, 0.0278, 0, −1.9304) and a Kaplan-Yorke dimension DKY = 3.1122 under
the initial conditions [1, 0, 1, 0], the hyperchaotic attractors are given in figure 2, and the frequency of the
hyperchaotic signals are given in figure 3. It can be seen that the hyperchaotic signals are not very high in
frequency and the signal z have a lower frequency in comparison and are in the effective operating region of the
memristor.
Fixed the parameters of equation (1) as b = 5.5, e = 0.4, k = 0.2 under the initial conditions [1, 0, 1, 0],
figure 4(a) shows the corresponding Lyapunov exponent spectrum based on parameter a in [0, 6], and the
bifurcation diagram is given in figure 4(b). When the control parameter a ä [0, 1.8], equation (1) begin with
period and then entered a large range of hyperchaotic state with two positive Lyapunov exponents at a = 1.8,
when a ä [4.56, 4.86], the system was break into chaotic orbits. Finally, the orbits returns to period nearby
a = 5.22. According to the bifurcation behaviors in figure 4, some typical phase portraits in the x-z plane are
drawn in figure 5. The change in phase portraits also corroborates the change in dynamics demonstrated in
figure 4.
The parameter k also introduce too many dynamics, fixed a = 4.5, b = 5.5, e = 0.4, changing the parameter k
from 0 to 1, and setting the initial conditions [1, 0, 1, 0], we can obtain the first three Lyapunov exponents of
parameter k drawn in figure 6(a), and the bifurcation diagrams in figure 6(b), the system begin with hyperchaotic
and goes to be chaotic at k = 0.27, then when k varies in [0.71, 0.78] and [0.88, 1], the trajectory break into quasi-
period and period.
From the above discussion, it can be concluded that the proposed nonlinear model can generate various
types of oscillations under the nonlinear action of memristors, including chaotic, periodic, quasi-periodic, and
hyperchaotic behaviors. The reason for this is that the memristor has a memory property, which greatly
enhances the complexity of equation (1).

4. Two-dimensional offset boosting and hybrid control

4.1. Two-dimensional offset boosting

Offset boosting is an interesting feature that can provides the possibility of polarity control, a constant
introduced in the nonlinear dynamic system will not change the dynamic exhibits. Take the equation (1) as an
example. Introducing two constants d1 and d2 in the first and second dimension of the system, obtaining,

4
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 2. Hyperchaotic attractors in different plane of equation (1) with a = 4.5, b =5.5, e = 0.4, and k = 0.2 under the initial
conditions [1, 0, 1, 0]: (a) x-y, (b) x-z, (c) y-z, (d) x-u.

Figure 3. Frequency spectra of equation (1) with a = 4.5, b =5.5, e = 0.4, and k = 0.2 under the initial conditions [1, 0, 1, 0].

5
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 4. Dynamical behavior of equation (1) with b = 5.5, e = 0.4, k = 0.2 under initial conditions [1, 0, 1, 0]: (a) Lyapunov exponents,
(b) bifurcation diagram.

Figure 5. Typical phase portraits of equation (1) with fixed b = 5.5, e = 0.4, and k = 0.2 under the initial condition [1, 0, 1, 0]: (a) a = 1,
(b) a = 1.6, (c) a = 4, (d) a = 4.56, (e) a = 4.7, (f) a = 5.3, (g) a = 5.4, (h) a = 5.5, (i) a = 6.

⎧ x = a (x - y ) - x ( - 5 + 4∣ z ∣) - u + d1
⎪ y = b (x - y ) + d2
(10)
⎨ z = - ez + x 2
⎪
⎩u = kx

Theorem . In system (10), the introduced constants d1 and d2 generate independent offset boosting.

d2 ad
Proof. With the replacements of y  y + , u  u + d1 - 2 , system (10) reduces to system (1). This
b b
indicates that the introduced constant d1 and d2 does not modify the dynamics of equation (1), but it facilitates
the offset boosting for the variable y and variable u.

6
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 6. Dynamical behavior of equation (1) with a = 4.5, b = 5.5, and e = 0.4 under the initial conditions [1, 0, 1, 0]: (a) Lyapunov
exponents of parameter k, (b) bifurcation diagram of variable x.

Figure 7. Offset boosting of equation (10) with a = 4.5, b = 5.5, e = 0.4, k = 0.2, d2 = 0 under the initial condition [1, 0, 1, d1]: (a) u-w
plane, (b) waveform u(t).

(i) When d1 ¹ 0, d2 = 0, the replacement is u  u + d1, indicating that the introduced constant d1 produce
an independent offset boosting on the variable u, holding other variables unchanged.
d2 ad
(ii) When d1 = 0, d2 ¹ 0, the replacement is y  y + , u  u - 2 , which means that the single
b b
constant d2 embedded in the second dimension of system (10) generates synchronous reverse control of
variable y and variable u.
d2 ad
(iii) When d1 ¹ 0, d2 ¹ 0, the replacements are y  y + , u  u + d1 - 2 , which indicates that the
b b
constants d1 and d2 in system (10) can achieve different control.

When a = 4.5, b = 5.5, e = 0.4, k = 0.2, the phase orbits and waveforms in different modes of the system are
obtained as shown in figures 7–9. In figure 7, the introduced constant d1 in the first dimension realizes positive
offset boosting of variable u. Negative constant pull the attractor in the negative direction of the u-axis. It should
be noted that the bifurcation parameters a, and b can affect the offset distance of the variables. In figure 8, the
introduced constant d2 in the second dimension realizes positive offset boosting with y but negative offset
boosting with u. It is shown that the constant d2 makes the attractor shift both in the y-axis and u-axis direction.
However, due to the existence of offset coefficient, the initial condition should be [1, d2/b, 1, -ad2/b], the
waveforms are shown in figures 8(c), (d). In figure 9, differential offset boosting is captured when
d1 ¹ 0, d2 ¹ 0. This time larger offset constant d1 does not make the attractor a relatively bigger step in the u-
axis for the hedging of d2.
For observing the boosting influence of attractors under different constants d1, and d2 conveniently,
figure 10 shows the average value of the system variables under different parameter configurations. Positive
offset boosting of variable u under different parameter d1 can be seen in figures 10(a), (b) shows the reverse offset
effect on the variable u. Cross-dimensional differential control is shown in figure 10(c), when d1 = d2, the

7
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 8. Reverse control offset boosting of equation (10) with a = 4.5, b = 5.5, e = 0.4, k = 0.2, d1 = 0 under the initial condition [1,
d2/b, 1, -ad2/b]: (a) y-u, (b) y-z, (c) y(t), (d) u(t).

Figure 9. Differential offset control of equation (10) with a = 4.5, b = 5.5, e = 0.4, k = 0.2 under the initial condition [1, d2/b, 1, d1-
ad2/b]: (a) y-u plane, (b) z-u. d1 = d2 = 0(dark turquoise), d1 = 10, d2 = 15(lilac), d1 = −10, d2 = −15(orange).

ad2
variable y and variable u are all boosted in a positive direction. In particular, when fixed d1 = b
, only variable y
was boosted, positive single offset control is shown in figure 10(d).

4.2. Hybrid control of amplitude and offset

More interestingly, in the proposed system, the introduction of the memristor leaves a knob for controlling the
amplitude of variables. When it is combined with the offset control constant, the system is also shown the
properties of hybrid control.

Theorem 2. In system (11), the introduced constants m in the third dimension can generate independent local
amplitude control.

8
Phys. Scr. 99 (2024) 015247 X Zhang et al

d2 ad
Figure 10. Different offset control modes in equation (10): (a) IC = [1, 0, 1, d1], d1 Î [−15, 15], d2 = 0, (b) IC = [1, b
, 1, - b 2 ], d2 Î
d2 ad2 d ad2
[−15, 15], d1 = 0, (c) IC = [1, b
, 1, d1 - b
], d3 Î [−5, 5], d1 = d2 Î [−15, 15], (d) IC = [1, b2 , 1, 0], d1 = b
, d2 Î [−15, 15].

⎧ x = a (x - y ) - x ( - 5 + 4∣ z ∣) - u + d1
⎪ y = b (x - y ) + d2
(11)
⎨ z = - ez + mx 2
⎪
⎩u = kx

x y d u ad
Proof. With the replacements of x  ,y + 2 , z  z, u  + d1 - 2 , equation (11)
m m b m b
is equivalent to equation (1), it is indicated that the constant d1, d2, and m do not modify the dynamics.

Figures 11–13 illustrate the utilization of coefficients m, d1, and d2 in a hybrid approach of amplitude and
offset control to realign the geometric control of the attractor. In figure 11, the introduced constant m and d1
realize positive offset boosting of variable u and amplitude control of variable x, y, and u. Negative constant d1
pull the attractor to the negative u-axis direction, but with the amplitude no change. As can be seen from the
figure, the level of offset boosting will not be affected by the amplitude control parameter. Figure 12 gives the
phase orbits variation of the system when the parameter d1 = 0 under different parameters m and d2. When
parameter d2 = 0, the system will provide only amplitude control. Figure 13 gives the phase orbits under
different parameters d1, d2 and m. We can also observe a similar phenomenon like figures 11 and 12. The phase
trajectories show that the amplitude control and offset boosting remain independent of one another. It is
noteworthy that the signal z remains unaltered regardless of any parameter adjustments.

5. Multistability analysis

Offset boosting analyzed above can not only provide chaotic signals with different polarities, but also be used for
identifying multistability. By linearly transforming the basins of attraction for fixed initial conditions, it
generates a dynamical dispersion that shows the coexisting attractors through different average values of
variables. When a constant d is introduced in the first dimension of equation (1),

9
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 11. Hybrid control of equation (11) with d2 = 0 under the initial condition [1, 0, 1, d1]: (a) x-u, (b) y-z. d1 = 0, m = 1 (dark
turquois), d1 = 2, m = 2 (orange), d1 = −2, m = 4 (lilac).

Figure 12. Hybrid control of equation (11) with d1 = 0 under the initial condition [1, d2/b, 1, -ad2/b]: (a) x-u, (b) y-z. (a) x-z plane, (b)
x-z plane. d2 = 0, m = 1 (dark turquois), d2 = 15, m = 2 (orange), d2 = 0, m = 2 (lilac).

Figure 13. Hybrid control of equation (11) under the initial conditions [1, d2/b, 1, d1-ad2/b]: (a) x-y, (b) x-u. d1 = 0, d2 = 0, m = 1
(dark turquois), d1 = 10, d2 = 15, m = 2 (orange), d1 = −10, d2 = −15, m = 4 (lilac).

⎧ x = a (x - y ) - x ( - c + d∣ z ∣) - (u - d )
⎪ y = b (x - y )
(13)
⎨ z = - ez + mx 2
⎪
⎩u = kx
When fixed parameters as a = 4.5, b = 5.5, e = 1.4, k = 0.2, figure 14 gives the variation of the system dynamics as
the parameter d is varied, and since the basin of attraction also moves with the parameter, the fixed initial
conditions will fall into different parts of the basin, forming a wavy line in the figure. This means that, when
fixing the initial values, a change in the parameter d induces a jump in the dynamical state of the system, in terms
of the phase trajectory, that is the occurrence of coexisting attractors.

10
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 14. Dynamical behaviors of equation (13) with a = 4.5, b = 5.5, e = 1.4, k = 0.2 and initial conditions [1, 0, 1, 0]: (a) average
values of all variables, (b) Lyapunov exponents.

Figure 15. Coexistence symmetric attractors when a = 4.5, b = 5.5 with initial condition IC1= [1, 0, 1, 0] (dark turquoise), IC2 = [−1,
0, 1, 0] (orange). (a) (b) e = 1.4, k = 0.2, (c) (d) e = 0.4, k = 1.

Figures 15(a), (b) shows typical hyperchaotic phase trajectories of the symmetric attractor of equation (13)
for different initial conditions, there also exist coexistence attractors in the periodic state, as shown in
figures 15(c), (d). The phase trajectories of the different states are also mirrored in the fluctuations of the mean
value in figure 14.
The basin of attraction under different parameters and initial conditions are presented in figure 16,
illustrating the existence of two differently colored areas. The basin of attraction correspond to the attractors in
figure 15 with the same colors are under the same parameters. Due to the symmetry of the proposed system, the
basin of attractions also exhibits a symmetric fractal structure. The respective Lyapunov exponents of the two
hyperchaotic attractors are (0.0827, 0.0291, 0, −2.8472), while that of the two periodic attractors are (0.0827,
0.0291, 0, −2.8472).

11
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 16. Basins of attraction of equation (1) with a = 4.5, b = 5.5. (a) e = 1.4, k = 0.2 under cross section z = 1, u = 0, (b) e = 0.4,
k = 1 under cross section z = 1, u = 0.

Figure 17. The simulation circuit of the memristive hyperchaotic oscillator.

6. Circuit implementation

The objective of this section is to verify theoretical results through a practical experimental analysis. For this
purpose, the chaotic system has been built on a PCB layout containing potentiometer, monolithic ceramic
capacitor, op-amp LM741CN, and multiplier AD633JN are made to verify dynamical behaviors of the chaotic
oscillator.
The element values used during the numerical investigation were preserved to facilitate meaningful
comparisons. The analog circuit of equation (1) is depicted in figure 17, based on Kirchhoff’s law, the
corresponding circuit state equations are written as:

12
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 18. Circuit simulation of typical phase trajectories of equation (12) with a = 4.5, b = 5.5, c = 5, d = 4, e = 0.4, m = 1 under
initial condition [1, 0, 1, 0]: (a) x- y plane, (b) x- z plane, (c) y-z plane, (d) x-u plane.

⎧ x = 1 1 1 1
x- y- xW (z ) - u
⎪ R1 C1 R2 C1 R3 C1 R 4 C1
⎪ 1 1
⎪ y = x- y
⎪ R5 C 2 R6 C 2
(12)
⎨ 1 2 1
⎪ z = R9 C4
x -
R8 C 4
z
⎪
⎪u = 1
⎪ x
⎩ R7 C3

Where W (z ) is realized by a separate module. The circuit is powered by ±15V. State variables x, y and u in
system (1) correspond to the state voltages of the capacitors in three channels. The inner variable z represents the
internal state of the memristor but in its simulation model corresponds to the capacitor voltage too. Here
introducing a common time scale 1000 is for better display in oscilloscope. We take the resistor 100 k W as the
base unit, using the parameters a = 4.5, b = 5.5, e = 0.4, k = 0.2, calculate the corresponding circuit element
parameters: C1 = C2 = C3 = 10nF , R1 = R2 = 100/a = 100/4.5 » 22.22k W, R3 = R13 = R14 = 10k W,
R 4 = 100k W, R5 = R 6 =100/b = 100/5.5 » 18.18k W, R7 = 100/k = 100/0.2 = 500k W. The corresp-
onding equivalent circuit parameters of the memristive are: C4 = 10nF , R8 = 100/e =100/0.4 = 250k W,
R 9 = R10 = R12 = 10k W, R11 = 2.5k W. To achieve absolute value function, we select R15 = R16 = 10k W. When
the circuit elements are set to the above parameters, we observe a typical hyperchaotic attractor similar to the
numerical simulation on the digital oscilloscope, as shown in figure 18.
By comparing equation (12) with equation (1), we know that voltage V2 corresponds to the offset parameter
d1 in the mathematical model, voltage V3 corresponds to the offset parameter d2 in the mathematical model. By
adjusting the voltage V2 and V3, we can observe attractors at different offset levels on the oscilloscope. Figure 19
shows the real boosting behaviors of u(t) under different V2. Two-dimensional offset boosting of single u(t) and
y(t) under different V3 are shown in figure 20.

7. Discussion and conclusions

A 4D memristive system with amplitude control and offset-boostable hyperchaotic attractors is proposed and
analyzed. One of the coefficients of the memristor can rescale all the variables, except the internal variable of the
memristor. The proposed systems can directly control the offset of one-dimensional variable through the
additional constants or meet the demand of multiple signal conditioning by two independent constants. Two
groups of hyperchaotic signals with the same or opposite polarity can be obtained by selecting different values of

13
Phys. Scr. 99 (2024) 015247 X Zhang et al

Figure 19. Offset boosting of equation (12) for real-time circuit implementation. (a) z-u plane, (b) signal of u(t). (V1 = 5 V. V3 = 0 V.
V2 = 1 V(blue), V2 = 0 V (pink), V2 = −1 V(gold)).

Figure 20. Reverse control offset boosting of equation (12) photographed on the digital oscilloscope. (a) y–u plane, (b) y–z plane, (c)
signal of y(t), (d) signal of u(t). (V1 = 5 V. V2 = 0 V. V3 = 1 V(green), V3 = 0 V(pink), V3 = −1 V(gold)).

constants, reflecting the arbitrary relocation of the dynamics under different constants. In addition, symmetric
attractors were also detected by variable offset boosting. The circuit experiment verifies the complex dynamic
behaviors of the proposed system as an example.
The proposed hyperchaotic systems provide a simple but effective model for generating multi-dimensional
controllable hyperchaotic signals. Different from the circuit implementation difficulties of homogenous
multistability, the free offset of the attractors from the initial space to the parameter space can achieve more
stable outputs in the circuit implementation. With the development of neuroscience, future work will focus on
non-bifurcation modulation in memristive neurons or neural networks, and then use the modulation of
neuronal firing to understand the working mechanism of the brain and the means of preventing some
physiological diseases such as epilepsy. More possible applications based on amplitude control and offset
boosting will be explored in the future, which could bring practical value to the electronic applications such as
secure communications [37], electrolysis of metals [38], and reservoir computing [39].

14
Phys. Scr. 99 (2024) 015247 X Zhang et al

Acknowledgments

This work was supported financially by the National Natural Science Foundation of China (Grant No. 62371242,
51974045) and a Project Funded by the Priority Academic Program Development of Jiangsu Higher Education
Institutions.

Data availability statement

No new data were created or analysed in this study.

ORCID iDs

Xin Zhang https://orcid.org/0000-0002-6006-8731

Chunbiao Li https://orcid.org/0000-0002-9932-0914

References
[1] Chua L 1971 Memristor-the missing circuit element IEEE Trans. Circuit Theory 18 507–19
[2] Singh J, Singh S and Paras N 2023 Design and Integration of vertical TFET and memristor for better realization of logical functions
Silicon 15 783–92
[3] Grant B et al 2023 Cascaded logic gates with printed electrochemical memristors Adv. Mater. Technol. 8 2202091
[4] Ma M, Zhu Y, Zhu Z, Yuan R, Liu J, Xu L, Yang Y and Wang Y 2022 Efficient in-memory aes encryption implementation using a general
memristive logic: surmounting the data movement bottleneck IEEE Nanatechnol. Mag. 16 24–C3
[5] Cao J et al 2022 Emerging dynamic memristors for neuromorphic reservoir computing Nanoscale 14 289–98
[6] Park S O, Jeong H, Park J, Bae J and Choi S 2022 Experimental demonstration of highly reliable dynamic memristor for artificial
neuron and neuromorphic computing Nat. Commun. 13 2888
[7] Deng Z et al 2023 A memristive spiking neural network circuit with selective supervised attention algorithm IEEE Trans. Comput. Aided
Des. Integr. Circuits Syst. 42 2604–17
[8] Ramadoss J, Kengne J, Tanekou S T, Rajaopal K and Kenmoe G D 2022 Reversal of period doubling, multistability and symmetry
breaking aspects for a system composed of a van der pol oscillator coupled to a duffing oscillator Chaos, Solitons Fractals 159 112157
[9] Zhang X et al 2023 A memristive autapse-synapse neural network: application to image encryption Phys. Scr. 98 035222
[10] Lin H, Wang C, Cui L, Sun Y, Xu C and Yu F 2022 Brain-like initial-boosted hyperchaos and application in biomedical image
encryption IEEE Trans. Ind. Inf. 18 8839–50
[11] Fozin T F et al 2022 Coexistence of hyperchaos with chaos and its control in a diode-bridge memristor based MLC circuit with
experimental validation Phys. Scr. 97 075204
[12] Xu Q et al 2023 Neuromorphic behaviors of a symmetric LAM-based electronic neuron circuit: Numerical simulation and
experimental measurement AEU-International Journal of Electronics and Communications 162 154594
[13] Lai Q et al 2023 Adaptive finite-time projective synchronization of complex networks with nonidentical nodes and quantized time-
varying delayed coupling Inf. Sci. 119891
[14] Li X et al 2023 Dynamical analysis and circuit implementation of a memristive chaotic system with infinite coexisting attractors Phys.
Scr. 98 115212
[15] Zhao Y et al 2023 Fully fixed-point integrated digital circuit design of discrete memristive systems AEU-International Journal of
Electronics and Communications 161 154522
[16] Lai Q, Wan Z and Kuate P D K 2022 Generating grid multi-scroll attractors in memristive neural networks IEEE Trans. Circuits Syst. I
70 1324–36
[17] Bao B, Bao H, Wang N, Chen M and Xu Q 2017 Hidden extreme multistability in memristive hyperchaotic system Chaos, Solitons
Fractals 94 102–11
[18] Lin H and Wang C 2020 Influences of electromagnetic radiation distribution on chaotic dynamics of a neural network Appl. Math.
Comput. 369 124840
[19] Zhang X et al 2020 Broken symmetry in a memristive chaotic oscillator IEEE Access 8 69222–9
[20] Chen M, Sun M, Bao H, Hu Y and Bao B 2019 Flux–charge analysis of two-memristor-based Chua’s circuit: dimensionality decreasing
model for detecting extreme multistability IEEE Trans. Ind. Electron. 67 2197–206
[21] Njimah O M, Ramadoss J, Telem A N K, Kengne J and Rajagopal K 2023 Coexisting oscillations and four-scroll chaotic attractors in a
pair of coupled memristor-based duffing oscillators: theoretical analysis and circuit simulation Chaos, Solitons Fractals 166 112983
[22] Minati L, Gambuzza L V, Thio W J, Sprott J and Frasca M 2020 A chaotic circuit based on a physical memristor Chaos, Solitons Fractals
138 109990
[23] Yuan F, Li S, Deng Y, li Y and Chen G 2022 Cu-Doped TiO2−x nanoscale memristive applications in chaotic circuit and true random
number generator IEEE Trans. Ind. Electron. 70 4120–7
[24] Zhang X, Li C, Dong E, Zhao Y and Liu Z 2022 A conservative memristive system with amplitude control and offset boosting Int. J.
Bifurcation Chaos 32 2250057
[25] Nguyen N T, Bui T, Gagnon G, Giard P and Kaddoum G 2021 Designing a pseudorandom bit generator with a novel five-dimensional-
hyperchaotic system IEEE Trans. Ind. Electron. 69 6101–10
[26] Du C, Liu L, Zhang Z and Yu S 2022 A mem-element wien-bridge circuit with amplitude modulation and three kinds of offset boosting
Chaos, Solitons Fractals 165 112832
[27] Li C, Wang X and Chen G 2017 Diagnosing multistability by offset boosting Nonlinear Dyn. 90 1335–41
[28] Li C et al 2021 Embedding any desired number of coexisting attractors in memristive system Chin. Phys. B 30 120511
[29] Li C, Lei T, Wang X and Chen G 2020 Dynamics editing based on offset boosting Chaos 30 063124
[30] Li C, Lei T and Liu Z 2022 Offset parameter cancellation produces countless coexisting attractors Chaos 32 121104

15
Phys. Scr. 99 (2024) 015247 X Zhang et al

[31] Lai Q, Yang L and Chen G 2023 Design and performance analysis of discrete memristive hyperchaotic systems with stuffed cube
attractors and ultraboosting behaviors IEEE Trans. Ind. Electron.
[32] Zhang S, Li C, Zheng J, Wang X and Zeng Z 2022 Memristive autapse-coupled neuron model with external electromagnetic radiation
effects IEEE Trans. Ind. Electron. 70 11618 –27
[33] Bao H et al 2023 Offset-control plane coexisting behaviors in two-memristor-based Hopfield neural network IEEE Trans. Ind. Electron.
70 10526–35
[34] Mohamed S M, Sayed W S, Radwan A G and Said L A 2022 FPGA implementation of reconfigurable CORDIC algorithm and a
memristive chaotic system with transcendental nonlinearities IEEE Trans. Circuits Syst. I 69 2885–92
[35] Yang Y, Huang L, Xiang J and Guo Q 2021 Three-dimensional sine chaotic system with multistability and multi-scroll attractor IEEE
Trans. Circuits Syst. Express Briefs 69 1792–6
[36] Bao B et al 2023 Memristor-induced mode transitions and extreme multistability in a map-based neuron model Nonlinear Dyn. 111
3765–79
[37] Ai Y and Wang H 2022 Fixed-time anti-synchronization of unified chaotic systems via adaptive backstepping approach IEEE Trans.
Circuits Syst. Express Briefs 70 626–30
[38] Xie Z et al 2020 Electrochemical behaviors of MnO2 on lead alloy anode during pulse electrodeposition for efficient manganese
electrowinning ACS Sustainable Chemistry & Engineering 8 15044–54
[39] Tanaka G et al 2019 Recent advances in physical reservoir computing: a review Neural Netw. 115 100–23

16