CN100547944C - Quantum nerve network is used for the method for Multiuser Detection - Google Patents
Quantum nerve network is used for the method for Multiuser Detection Download PDFInfo
- Publication number
- CN100547944C CN100547944C CNB2006100387221A CN200610038722A CN100547944C CN 100547944 C CN100547944 C CN 100547944C CN B2006100387221 A CNB2006100387221 A CN B2006100387221A CN 200610038722 A CN200610038722 A CN 200610038722A CN 100547944 C CN100547944 C CN 100547944C
- Authority
- CN
- China
- Prior art keywords
- quantum
- feedback
- user detector
- neuron
- nerve network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Abstract
The method of quantum nerve network Multiuser Detection relates to this method of The Realization of Simulation on classic computer, this method constitutes multi-user detector with quantum nerve network, network core adopts the feedback-type quantum neuron to simplify the structure of multi-user detector, network evolution utilizes quantum parallel computation characteristic to carry out quick optimizing, reduces the complexity of multi-user detector; Be specially: design a feedback-type quantum neuron, a kind ofly show the method for multiuser receiver received signal, a quantum nerve network multi-user detector, a parallel evolutionary operator F with the quantum register tables
1With parallel evolutionary operator F
1Act on the output quantum state of quantum nerve network multi-user detector, it is upgraded evolution; Repeat previous step suddenly the output quantum state after upgrading with upgrade before no change; Design the operator F that develops at random
2Substitute parallel evolutionary operator F
1Repeat previous step suddenly the output quantum state after upgrading with upgrade before no change.
Description
Technical field
Quantum nerve network (QNN-Quantum Neural Networks) is a kind of new intelligence computation normal form that conventional artificial neural net (ANN) is combined with the quantum calculation theory.The present invention relates to that quantum nerve network is used for the method for Multiuser Detection and on classic computer this method of The Realization of Simulation, research contents belongs to the technical field that signal of communication is handled.
Background technology
Nineteen ninety-five, professor Kak of the U.S. proposed the neural notion of calculating of quantum first, and some scholars have proposed quantum nerve network models such as quantum derivative neural net, quantum dot neural net, quantum associative memory model, quantum entanglement neural net afterwards.Studies show that, quantum nerve network (QNN) is owing to utilized quantum estimated performances such as quantum is parallel, quantum entanglement, be better than ANN in aspect performances such as memory capacity and processing speeds, have the parallel processing capability stronger and also can handle more large data collection than ANN, and can solve the insurmountable problem of some ANN, can't find the solution linear inseparable problem etc. as catastrophe memory loss problem, monolayer neural networks.In recent years, the research day of quantum nerve network interest is active, and in pattern recognition, tangle aspects such as calculatings, approximation to function and obtain Preliminary Applications, report still rare but quantum nerve network is used for the research that signal of communication handles.
The multi-user communication of large user's amount is the basic communication form of following broadband high-speed multimedia mobile communication, is that 3G reaches the main mode of next-generation mobile communications, and Multiuser Detection (MUD-MutiuserDetection) technology is one of its key technology in real time.Verdu proposed and had analyzed best multi-user detector in 1986, it detects performance can approach single user's receptivity, can effectively overcome near-far interference, and improve power system capacity greatly, but its computation complexity of finding the solution the optimal solution process is the exponential function of number of users, along with the increase of number of users is that a NP is difficult to resolve problem.Therefore many scholars have proposed linear and nonlinear sub-optimal algorithm, comprise classical neural net detection method, more typically mainly contain the multi-user test method that adopts the Hopfield neural net.
Hopfield neural net multi-user test method has the hardware complexity that is directly proportional with number of users, be easy to large scale integrated circuit and realize, but its iterative process is owing to adopt gradient descent algorithm generally to be difficult for converging to global optimum's point.In addition, the complexity of this network is pressed square increase of number of users, and when the user who activates simultaneously in the channel is a lot, network will be very huge, be unsuitable for a lot of channel of user.
Summary of the invention
Technical problem: the object of the present invention is to provide a kind of on classic computer the method for The Realization of Simulation quantum nerve network Multiuser Detection, thereby solve classical neural net multi-user detector hardware complexity height, be difficult for converging to global optimum's point, be not suitable for the problem of the multi-user communication of large user's amount.
Technical scheme: the implementation method that quantum nerve network of the present invention is used for Multiuser Detection is that quantum nerve network is constituted multi-user detector, network core adopts the feedback-type quantum neuron to simplify the structure of multi-user detector, network evolution utilizes quantum parallel computation characteristic to carry out quick optimizing, reduces the complexity of multi-user detector; The specific implementation method is:
1.) feedback-type quantum neuron of design: its input and output are n position quantum bit, it is n position quantum register, output is imported as the feedback of feedback-type quantum neuron simultaneously, the threshold value of feedback-type quantum neuron also is a n position quantum bit, the connection weight of feedback-type quantum neuron is the matrix of a K * K, the evolution operator of feedback-type quantum neuron
Be an operator that acts on the quantum register of n position, the evolution of Control and Feedback type quantum neuron state;
2.) a kind of method of showing the multiuser receiver received signal with the quantum register tables of design: prepare the quantum register of a n position quantum bit, use | y〉expression, select quantum register-bit number
K is the number of users of multiuser receiver, order | y 〉=y, y is the received signal information bit vector of multiuser receiver;
3.) quantum nerve network multi-user detector of design: its core is the feedback-type quantum neuron, with quantum register | y〉as the input of this multi-user detector, |
bBe its output;
4.) parallel evolutionary operator of design
Wherein sign () is a sign function;
5.) with the parallel evolutionary operator
Act on the output quantum state of quantum nerve network multi-user detector, it is upgraded evolution;
6.) repeating step 5), no change before output quantum state after upgrading and the renewal, promptly network state is stable, and the pairing transmission information sequence of output quantum state of this moment is the testing result of quantum nerve network multi-user detector;
7.) operator that develops at random of design
Substitute the parallel evolutionary operator
Function f () is defined as f (x)=sign (u-|x| in the formula
2), wherein sign () is a sign function, u one belongs to the random number of [0,1], | x| is the probability amplitude of quantum nerve network output quantum bit;
8.) the operator that will develop at random
Act on the output quantum state of quantum nerve network multi-user detector, it is upgraded evolution;
9.) repeating step 8), no change before output quantum state after upgrading and the renewal, promptly network state is stable, and the pairing transmission information sequence of output quantum state of this moment is the optimal solution of quantum nerve network multi-user detector.
Beneficial effect: the present invention constitutes the quantum nerve network multi-user detector by adopting single quantum neuron to substitute traditional Hopfield neural net, and the Multiuser Detection network configuration is simple, and hardware complexity is low.The quantum nerve network multi-user detector can utilize quantum parallel computation characteristic that optimizing speed is doubly improved by index in the evolution of network state.Computer artificial result shows that the designed quantum nerve network multi-user detector performance of the present invention is better than classical neural net multi-user detector; It detects the quantum nerve network multi-user detector that performance is better than adopting parallel operator the improved quantum nerve network multi-user detector of employing random operator, more near the performance of optimal detector, and when same detector performance, can hold more users, can solve the multi-user communication problem of large user's amount preferably.
Description of drawings
Fig. 1 is quantum nerve network (QNN) multi-user detector structure,
Fig. 2 is the flow chart that quantum nerve network is used for the implementation method of Multiuser Detection,
Fig. 3 is feedback-type quantum neuron (FQN) model,
Fig. 4 is the curve that 8 user's synchro system users' 1 bit error rate changes with signal to noise ratio,
Fig. 5 is the curve that synchro system signal to noise ratio user's 1 when being 9dB bit error rate changes with excited user number,
Fig. 6 is the curve that 8 user's asynchronous system users' 1 bit error rate changes with signal to noise ratio,
Fig. 7 is the curve that asynchronous system signal to noise ratio user's 1 when being 9dB bit error rate changes with excited user number,
Fig. 8 is the curve that 8 user's synchro system users' 1 bit error rate changes with signal to noise ratio,
Fig. 9 is the curve that synchro system signal to noise ratio user's 1 when being 9dB bit error rate changes with excited user number,
Figure 10 is the curve that 8 user's asynchronous system users' 1 bit error rate changes with signal to noise ratio,
Figure 11 is the curve that asynchronous system signal to noise ratio user's 1 when being 9dB bit error rate changes with excited user number.
Embodiment
The present invention designed a kind of adopt quantum nerve network to carry out the method for Multiuser Detection and on classic computer The Realization of Simulation this method.Designed a kind of feedback-type quantum neuron model and shown with the quantum register tables and the method for multiuser receiver received signal designed the multi-user detector based on quantum nerve network on this basis, its structure as shown in Figure 1.FQN is the feedback-type quantum neuron among the figure, and the output of classical multi-user detector matched filter is produced (Preparing) and is quantum register | y〉and, as the input of QNN multi-user detector, R is the cross-correlation matrix of user characteristics waveform.
Quantum nerve network be used for Multiuser Detection implementation method concrete steps as shown in Figure 2.
The QNN form of best Multiuser Detection criterion is
The output quantum state replacement criteria of QNN multi-user detector is
|
b′(t+1)>=W|
b(t)>-|θ>=|y>-R|
b(t)>(2)
Subsequently to quantum state |
b' (t+1) 〉=[
b'
1(t+1),
b'
2(t+1), L,
b'
K(t+1)]
TDevelop, evolution attitude as a result is
The present invention has designed a series of emulation experiments and has detected the performance based on the multi-user detector of quantum nerve network that is proposed.Making quantum nerve network multi-user detector desired user when different signal to noise ratios and different user are counted by emulation (is without loss of generality, be user 1) bit error rate (BER-Bit Error Ratio) change curve, the performance when increasing to the anti-multiple access interference performance of QNN and other detection algorithms with number of users compares.Fig. 4---Figure 11 is respectively the simulation result under the different situations.
From figure simulation result as can be seen, no matter synchronous situation and asynchronous condition, QNN-MUD have all preferably that bit error rate detects performance, are better than conventional detector and HNN-MUD, and along with the increase of excited user number, the performance of QNN-MUD still is better than HNN-MUD.And the improved QNN-MUD that adopts the designed random operator of the present invention has the bit error rate detection performance better than QNN-MUD, and can hold more user under the same detection performance.
This method constitutes multi-user detector with quantum nerve network, network core adopts the feedback-type quantum neuron to simplify the structure of multi-user detector, network evolution utilizes quantum parallel computation characteristic to carry out quick optimizing, reduces the complexity of multi-user detector, and its method is:
1.) feedback-type quantum neuron of design: its input and output are n position quantum bit, it is n position quantum register, output is imported as the feedback of feedback-type quantum neuron simultaneously, the threshold value of feedback-type quantum neuron also is a n position quantum bit, the connection weight of feedback-type quantum neuron is the matrix of a K * K, the evolution operator of feedback-type quantum neuron
Be an operator that acts on the quantum register of n position, the evolution of Control and Feedback type quantum neuron state;
2.) a kind of method of showing the multiuser receiver received signal with the quantum register tables of design: prepare the quantum register of a n position quantum bit, use | y〉expression, select quantum register-bit number
K is the number of users of multiuser receiver, order | y 〉=y, y is the received signal information bit vector of multiuser receiver;
3.) quantum nerve network multi-user detector of design: its core is the feedback-type quantum neuron, with quantum register | y〉as the input of this multi-user detector, |
bBe its output;
5.) with the parallel evolutionary operator
Act on the output quantum state of quantum nerve network multi-user detector, it is upgraded evolution;
6.) repeating step 5), no change before output quantum state after upgrading and the renewal, promptly network state is stable, and the pairing transmission information sequence of output quantum state of this moment is the testing result of quantum nerve network multi-user detector;
7.) operator that develops at random of design
Substitute the parallel evolutionary operator
Function f () is defined as f (x)=sign (u-|x| in the formula
2), wherein sign () is a sign function, u one belongs to the random number of [0,1], | x| is the probability amplitude of quantum nerve network output quantum bit;
8.) the operator that will develop at random
Act on the output quantum state of quantum nerve network multi-user detector, it is upgraded evolution;
9.) repeating step 8), no change before output quantum state after upgrading and the renewal, promptly network state is stable, and the pairing transmission information sequence of output quantum state of this moment is the optimal solution of quantum nerve network multi-user detector.
Concrete grammar of the present invention is:
Feedback-type quantum neuron model of first step design
Model as shown in Figure 3 for the designed feedback-type quantum neuron (FQN-Feedback Quantum Neuron) of the present invention.The output of FQN among the figure | v〉be n position qubit, i.e. n position quantum register (Qregister), then
Output | v〉import as the feedback of FQN simultaneously, | θ〉be threshold value, also be n position qubit, the column vector form that is expressed as the Hilbert space is
|θ>=[θ
1θ
2Lθ
K]
T,K=2
n (5)
W is the connection weight matrix of one K * K,
It is an operator that acts on the quantum register of n position.
A kind of method of showing the multiuser receiver received signal with the quantum register tables of second step design
The quantum register for preparing a n position qubit | y 〉,
| y〉be 2
nThe column vector in dimension Hilbert space makes 2
n=K, K are number of users, then the quantum register figure place
Further can make | y 〉=y, with quantum register | y〉as the input of quantum multi-user detector.It should be noted that n position quantum register | y〉can store from 0 to 2 simultaneously
nAll K=2 of-1
nNumber, they respectively exist simultaneously with certain probability.According to quantum parallel computation characteristic, any conversion that acts on the quantum register all is simultaneously all K numbers to be operated, thereby the once-through operation of quantum computer can produce 2
nIndividual operation result is equivalent to classic computer 2
nInferior computing.
Quantum nerve network multi-user detector of the 3rd step design
The present invention designed based on the multi-user detector structure of quantum nerve network as shown in Figure 1.
FQN is the feedback-type quantum neuron among the figure, and the output of classical multi-user detector matched filter is produced (Preparing) and is quantum register | y〉and, as the input of QNN multi-user detector, R is the cross-correlation matrix of user characteristics waveform.The QNN form of best Multiuser Detection criterion is
The output quantum state replacement criteria of QNN multi-user detector is
|
b′(t+1)>=W|
b(t)>-|θ>=|y>-R|
b(t)>(8)
Subsequently to quantum state |
b' (t+1) 〉=[
b'
1(t+1),
b'
2(t+1), L,
b'
K(t+1)]
TDevelop, evolution attitude as a result is
The present invention has designed two kinds of evolution operators---parallel operator
And random operator
Operator
Wherein sign () is a sign function.Parallel operator
To constituting the K=2 of quantum neuron stack attitude in the evolutionary process of quantum state
nThe conversion of individual ground state is carried out synchronously.
Operator
Function f () is defined as f (x)=sign (u-|x| in the formula
2), wherein sign () is a sign function, u one belongs to the random number of [0,1].Operator
Act on quantum state shown in the formula (5) |
b' (t+1) on, if random number u greater than the quantum bit probability amplitude square |
b'
i(t+1) |
2, attitude as a result then |
b(t+1)〉i element in
b i(t+1) value 1; Otherwise, value-1.So the attitude as a result of quantum neuron is that probability according to quantum bit obtains in the evolutionary process, utilized quantum stack attitude in the process of measuring, to collapse to principle on the some ground state with certain probability.The present invention defines operator
Be random operator.
The 4th step The Realization of Simulation quantum nerve network multi-user test method on classic computer
The present invention has designed a series of emulation experiments and has detected the performance based on the multi-user detector of quantum nerve network that is proposed.The DS-CDMA system of BPSK modulation in the Gaussian channel is adopted in emulation, and frequency expansion sequence adopts 31 Gold sequences, and maximum normalized crosscorrelation coefficient is 9/31.As a comparison, we are carrying out emulation to following multi-user detector respectively synchronously and under the asynchronous condition:
Conventional detector
Hopfield neural net multi-user detector (HNN-MUD)
Adopt random operator
Improved quantum nerve network multi-user detector (improved QNN-MUD)
Making above multi-user detector desired user when different signal to noise ratios and different user are counted by emulation (is without loss of generality, be user 1) bit error rate (BER-Bit Error Ratio) change curve, the performance when increasing to the anti-multiple access interference performance of QNN and other detection algorithms with number of users compares.Fig. 4---Figure 11 is respectively the simulation result under the different situations.
Claims (1)
1. a quantum nerve network is used for the implementation method of Multiuser Detection, it is characterized in that this method constitutes multi-user detector with quantum nerve network, network core adopts the feedback-type quantum neuron to simplify the structure of multi-user detector, network evolution adopts the parallel evolutionary operator, utilize quantum parallel computation characteristic to carry out quick optimizing, reduce the complexity of multi-user detector, the specific implementation method is:
1.) feedback-type quantum neuron of design: its input and output are n position quantum bit, it is n position quantum register, output is imported as the feedback of feedback-type quantum neuron simultaneously, the threshold value of feedback-type quantum neuron also is a n position quantum bit, the connection weight of feedback-type quantum neuron is the matrix of a K * K, the evolution operator of feedback-type quantum neuron
Be an operator that acts on the quantum register of n position, the evolution of Control and Feedback type quantum neuron state;
2.) a kind of method of showing the multiuser receiver received signal with the quantum register tables of design: prepare the quantum register of a n position quantum bit, use | y〉expression, select quantum register-bit number
K is the number of users of multiuser receiver, order | y 〉=y, y is the received signal information bit vector of multiuser receiver;
3.) quantum nerve network multi-user detector of design: its core is the feedback-type quantum neuron, with quantum register | y〉as the input of this multi-user detector, |
bBe its output;
4.) parallel evolutionary operator of design
Wherein sign () is a sign function; The number of sign function is 2
n
5.) with the parallel evolutionary operator
Act on the output quantum state of quantum nerve network multi-user detector, it is upgraded evolution, the parallel evolutionary operator
To in the evolutionary process of output quantum state to constituting the K=2 of quantum neuron stack attitude
nIndividual ground state is carried out conversion synchronously;
6.) repeating step 5), no change before output quantum state after upgrading and the renewal, promptly network state is stable, and the pairing transmission information sequence of output quantum state of this moment is the testing result of quantum nerve network multi-user detector.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2006100387221A CN100547944C (en) | 2006-03-09 | 2006-03-09 | Quantum nerve network is used for the method for Multiuser Detection |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2006100387221A CN100547944C (en) | 2006-03-09 | 2006-03-09 | Quantum nerve network is used for the method for Multiuser Detection |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1862982A CN1862982A (en) | 2006-11-15 |
CN100547944C true CN100547944C (en) | 2009-10-07 |
Family
ID=37390324
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB2006100387221A Expired - Fee Related CN100547944C (en) | 2006-03-09 | 2006-03-09 | Quantum nerve network is used for the method for Multiuser Detection |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100547944C (en) |
Families Citing this family (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101464965B (en) * | 2009-01-16 | 2011-08-17 | 北京航空航天大学 | Multi-nuclear parallel ant group design method based on TBB |
PL406171A1 (en) * | 2013-11-21 | 2015-05-25 | Wojciech Burkot | Method and the device data processing |
CN103795436B (en) * | 2014-02-24 | 2016-01-27 | 哈尔滨工程大学 | Based on the robust multi-user test method of Quantum Hopfield Neural Network and quantum fish-swarm algorithm |
CN108713206B (en) * | 2016-02-18 | 2022-12-27 | 微软技术许可有限责任公司 | Randomized gap and amplitude estimation |
CN106127300A (en) * | 2016-07-04 | 2016-11-16 | 哈尔滨理工大学 | A kind of rotating machinery health status Forecasting Methodology |
CN107231214B (en) * | 2017-06-12 | 2020-07-28 | 哈尔滨工程大学 | Optimal multi-user detection method based on evolutionary chaotic quantum neural network |
CN108194143B (en) * | 2017-12-25 | 2019-10-11 | 陕西煤业化工技术研究院有限责任公司 | Masonry beam DEFORMATION MONITORING SYSTEM and monitoring and pre-alarming method based on Hopfield neural network |
CN110727462B (en) * | 2018-07-16 | 2021-10-19 | 上海寒武纪信息科技有限公司 | Data processor and data processing method |
CN109063359A (en) * | 2018-08-16 | 2018-12-21 | 燕山大学 | A kind of dynamic modelling method of Process of Circulating Fluidized Bed Boiler |
CN113360747B (en) * | 2020-03-04 | 2024-01-12 | 阿里巴巴集团控股有限公司 | Data processing method and device based on neural network model |
CN116458872B (en) * | 2023-06-13 | 2023-09-05 | 汶上县人民医院 | Method and system for analyzing respiratory data |
-
2006
- 2006-03-09 CN CNB2006100387221A patent/CN100547944C/en not_active Expired - Fee Related
Non-Patent Citations (4)
Title |
---|
量子神经网络及其在CDMA多用户检测中的应用. 李飞,赵生妹,郑宝玉.信号处理,第21卷第6期. 2005 |
量子神经网络及其在CDMA多用户检测中的应用. 李飞,赵生妹,郑宝玉.信息处理,第21卷第6期. 2005 |
量子神经网络及其在CDMA多用户检测中的应用. 李飞,赵生妹,郑宝玉.信号处理,第21卷第6期. 2005 * |
量子神经网络及其在CDMA多用户检测中的应用. 李飞,赵生妹,郑宝玉.信息处理,第21卷第6期. 2005 * |
Also Published As
Publication number | Publication date |
---|---|
CN1862982A (en) | 2006-11-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN100547944C (en) | Quantum nerve network is used for the method for Multiuser Detection | |
CN107743103A (en) | The multinode access detection of MMTC systems based on deep learning and channel estimation methods | |
CN109617584A (en) | A kind of mimo system beamforming matrix design method based on deep learning | |
Bai et al. | Deep learning based fast multiuser detection for massive machine-type communication | |
CN108334945A (en) | The acceleration of deep neural network and compression method and device | |
CN108921298A (en) | Intensified learning multiple agent is linked up and decision-making technique | |
CN107294659B (en) | One kind being based on cross validation non-orthogonal multiple system multi-user detection method | |
CN109787715A (en) | The DNN coding/decoding method and decoded communications equipment of SCMA system | |
CN107248902A (en) | A kind of block compressed sensing non-orthogonal multiple system multi-user detection method | |
CN104850653A (en) | Short-term tourist traffic and trend prediction system based on streaming data extraction | |
Wang et al. | A deep learning-based intelligent receiver for improving the reliability of the MIMO wireless communication system | |
CN106130689A (en) | A kind of non-linear self-feedback chaotic neural network signal blind checking method | |
CN102780542A (en) | Gain factor adjustment method for Hopfield neural network signal blind detection | |
Dehuri et al. | Genetic feature selection for optimal functional link artificial neural network in classification | |
CN103793747B (en) | A kind of sensitive information template construction method in network content security management | |
CN107181705A (en) | A kind of half-blind channel estimating method and system | |
CN105916197A (en) | Power adaptive method of social credit driving in D2D network | |
CN107273970B (en) | Reconfigurable platform of convolutional neural network supporting online learning and construction method thereof | |
CN109525598A (en) | A kind of fault-tolerant compression method of wireless sense network depth and system based on variation mixing | |
Zhang et al. | Application of convolutional neural network in multi-channel scenario D2D communication transmitting power control | |
CN114048774A (en) | Se-block-based resnet communication radiation source identification method and system | |
CN108282199A (en) | Power distribution method in wireless energy supply MIMO relay system and wireless energy supply MIMO relay system | |
Chen et al. | Tasks-oriented joint resource allocation scheme for the Internet of vehicles with sensing, communication and computing integration | |
CN113869238A (en) | Cognitive Internet of vehicles intelligent frequency spectrum sensing method and system | |
CN101739565A (en) | Large-capacity pattern recognition method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20091007 Termination date: 20150309 |
|
EXPY | Termination of patent right or utility model |