ARGO实验中“膝”区原初宇宙线成分分辨方法研究
|
摘要
宇宙线(来自宇宙空间的质子和原子核)的能谱遵循负幂律:dN╱dE∝E~γ,但在能量约为4 PeV附近,宇宙线的全粒子能谱有明显的拐折,幂指数由γ≈-2.7变为γ≈-3.1,形成所谓的“膝”结构,对应的能区称为“膝”区。虽然实验上已经观测到这一结构,但是它的成因还不能确定。为解释宇宙线能谱的这一特殊结构,人们提出了多个理论模型。某些模型认为,“膝”结构起因于宇宙线的加速和传播过程,是宇宙线能谱的内禀特性;而另外一些模型认为,在超高能下新的强相互作用机制的出现导致了“膝”区的形成。大多数的理论模型可很好地解释实验所观测到的宇宙线全粒子能谱,但所预言的“膝”区宇宙线各个分成分的能谱却存在着很大的差异。因此在实验上精确测量“膝”区原初宇宙线的分成分能谱对于了解甚高能宇宙线的起源、加速和传播机制具有重要的意义。
由于宇宙线流强随能量急剧下降,超高能的宇宙线粒子已经不能通过高空气球实验或卫星实验来进行直接观测,只能利用位于地面的探测器阵列通过广延空气簇射(Extensive Air Shower,EAS)过程进行间接的测量。如何有效地区分由不同的原初宇宙线粒子所引起的空气簇射是利用地面实验进行“膝”区物理研究的关键。这一方面要求探测器阵列本身能够提供尽可能多的有关EAS的实验信息,另一方面需要在数据分析过程中采用有效的分辨由各种原初粒子所引起的空气簇射的方法。
中意合作ARGO-YBJ实验是位于我国西藏羊八井宇宙线观测站(大气深度606g/cm~2,海拔高度4300m)的广延大气簇射地面探测实验,该实验采用了由阻性板计数器(Resistive Plate Chamber,RPC)所组成的“地毯式”探测器阵列,具有高的覆盖率(90%以上)和很好的时间、空间分辨率的特性,可以对到达阵列的EAS粒子的时间和空间分布进行较为细致的测量,探测器信号的模拟读出使得阵列的观测能区可以涵盖“膝”区;另外ARGO-YBJ实验阵列处于由“膝”区原初宇宙线粒子所引起的空气簇射的纵向发展极大区域,簇射大小的涨落小且几乎与原初粒子的种类无关,利用该实验的观测信息可以较精确地测量原初粒子的能量。这些特点使得ARGO-YBJ实验在分辨“膝”区原初粒子种类和重建原初粒子能量上具有优势,如果再配以有效的数据分析方法,有望得到较精确的“膝”区原初粒子的分成分能谱,进而为解决“膝”区物理问题提供可靠的实验数据。本文的目的就是针对ARGO-YBJ实验阵列的特点,利用Monte Carlo模拟数据寻找可有效地区分由不同的原初粒子所引起的空气簇射的方法,并研究测量“膝”区原初粒子分成分能谱的可行性。
在本工作中,首先利用CORSIKA程序进行了EAS模拟。为了估计强相互作用模型对分辨结果的影响,在EAS数据模拟中选用了两种最具代表性且与实验结果符合较好的强相互作用模型:QGSJET-Ⅱ和SIBYLL。利用这两种强相互作用模型,细致地模拟了能量在100TeV-10PeV之间的不同原初宇宙线粒子(质子,He核,CNO核,Mg-Si核,铁核)所引起的EAS过程。利用以GEANT3为基础的探测器模拟程序模拟了EAS次级粒子在ARGO探测器中的传输过程和探测器的响应,获得了具有足够统计量的Monte Carlo模拟数据。
利用这些模拟数据分析了由不同的原初粒子所引起的簇射中次级粒子的时间、空间分布特征,得到了平均横向分布宽度、80%半径R_(80)、芯位区域与R_(80)区域粒子密度之比Ratio_(80)、簇射前锋面斜率S_(front)等几个可以描述不同簇射的次级粒子时间-空间分布的差异的特征量。将这些特征量作为人工神经网络的输入量,进行了多参数分析。训练所得的神经网络可以对原初质子成分做出有效识别。为了研究不同强相互作用模型对网络分辨效率的影响,分别利用两种强相互作用模型数据训练、测试了神经网络,并进行了交叉检验(利用一种模型数据训练所得的神经网络对另一种模型的数据进行分辨)。结果表明:人工神经网络挑选效果对强相互作用模型的依赖性较弱,在保存约40%质子事例的情况下可以排除约94%的其他强子事例。利用所挑选出来的质子事例,对“膝”区质子的能谱进行了重建,所得结果与模拟中实际使用的能谱可以很好地符合。
另外,本工作还使用了多尺度分析方法研究了区分“膝”区原初铁核的可行性。对在探测平面上的簇射次级粒子的分布进行多重分形分析和小波变换分析,得到了两个表征簇射次级粒子靶图分形特性的特征量:多重分形矩指数和小波变换矩指数。分别取它们在q=4,6,8时的值作为人工神经网络的输入值,得到了可以有效分辨“膝”区原初铁核成分的人工神经网络。将数据按击中数分段分别进行了多参数分析,分辨结果表明,所得人工神经网络对次级带电粒子数＞20000的事例分辨效果最好,对铁核事例的判选率为56.4%,对其他强子事例的排除率为98.3%,品质因子可达到4.36。
综上,通过对ARGO-YBJ实验条件的细致模拟和对簇射事例的仔细分析,本工作得到了可以用来区分原初粒子种类的特征量。结合人工神经网络进行了多参量分析,并检验了人工神经网络对强相互作用模型的依赖性。结果表明,所得的人工神经网络的分辨效果对强相互作用模型的依赖性较弱,可对“膝”区各原初成分引起广延大气簇射事例做出有效分辨,从而重建出“膝”区质子能谱。同时,本工作还研究了应用多尺度分析方法区分“膝”区原初铁核的可行性,得到了表征簇射次级粒子靶图分形特性的特征量,并结合人工神经网络方法进行了多参数分析。结果表明此网络可有效分辨“膝”区铁核成分。
The energy spectrum of cosmic rays (proton and nuclei from the universe) follows a negative power law: dN/dE∝E~r, but the all-particle energy spectrum ofcosmic rays shows a distinctive feature around 4 PeV, where the spectral index of the power-law changes from -2.7 to approximately -3.1. This feature is commonly called the the "knee", and the corresponding energy region is called the "knee" region. Existence of this feature has been well established experimentally, but there still remain controversial arguments on its origin. To explain this feature, several mechanisms have been proposed. In some of these theoretical models, it is believed that the knee is an intrinsic property of the energy spectrum, related to the acceleration and propagation of the cosmic ray. While in other models, the knee is explained as a new type of interaction at very high energy. Most of these models can describe the obsvered all-particle energy spectrum very well, but the predictions of the individual element spectra in some modles are quite different. So the precise measurements of the individual element spectra at the knee energy region are important for understanding the origin, acceleration and propagation of the cosmic rays.
Since the flux of the cosmic rays decreases rapidly with energy according to a negative power law, the cosmic ray particles with very high energy (VHE) can't be directly detected by balloon- and satellite-borne detectors. These VHE particles can only be indirectly detected by the ground-based detector arrays, which can record the secondary particles in the extensive air shower (EAS) induced by them. The key point in the study of the physics at the knee region in the groud based experiments is to efficiently identify the EAS's induced by different primary particles. To achieve this aim, it requires that the detector array can records sufficient information of EAS, and an effective method to discriminate primary cosmic rays is needed.
The ARGO-YBJ experiment, a collaboration between Chainese and Intalian institutions, is a groud-based EAS detector array, located at the Yangbajing Cosmic Ray Observatory (atmospheric detpth: 606 g/cm2,4300m a.s.1) in Tibet (P.R. China). This experiment utlize a full coverage detector array consisting of Resistive Plate Chambers (RPC). The high coverage (above 90%), good time resolution and fine space granularity makes it able to measure the time and lateral distributions of the secondary particles in a EAS with sufficient precision.With the analog read-out of the RPCs charge, the array can measure the cosmic rays with energies up to the knee region. In addition, at the observation level of this experiment, the EAS induced by cosmic ray particles with energies at "knee" region reach a maximum development, irrespective of the primary mass, so that the shower size is less fluctuated and the energy determination is more precise and less dependent upon the unknown composition. These advantages, together with an effective data analysis method, make it possible to get individual element spectrum in the "knee" region. These experimental information could be helpful for the study of the origin of "knee". In this work we try to find a method for identifying the primary cosmic ray particles with energies at the "knee" region by using Monte Carlo generated data, and study the feasibility of measuring the individuale element spectra in the "knee" region.
The EASs induced by different primary particles are simulated using CORSIKA program. In order to estimate the influence of hadronic interaction models on the identification results, we chose two interaction models for the Monte Carlo simulation: QGSJET-II and SIBYLL, which are representative and have good agreement with the experimental results. We have initiated the EAS with high energy protons, helium, CNO nuclei, Mg-Si nuclei and iron nuclei with energies ranging from 100TeV to 10PeV. The transportation of the EAS particles in ARGO detector and the detector response are simulated indetail with a detector simulation program based on GEANT3. Monte carlo events with sufficient statistics are thus obtained.
After examining the space-time information of the EAS induced by different primaries by using the selected Monte Carlo events, we found that following 5 parameters can be used to characterize the difference between proton and other nuclei induced showers: N_(Hits),, R_(80), S_(front), Ratio_(80). With these parameters as inputs to an artificial neural network (ANN), a multi-parameter analysis is performed. To check the influence of hadronic interaction models, two ANN is trained and tested by QGSJET-II data and SIBYLL data respectively, and also the cross-examination is done (test the QGSJET-II ANN with SIBYLL data and vice versa). The dependence on hadronic interaction models is very weak, and primary proton events can be identified effectively from other events by use of ANN method. The ANN can pick out about 40% proton events and the rejection ratio for other nuclei is about 94%. Using the selected proton events, the energy spectrum of proton in the knee region is reconstructed. The reconstructed result is in good agreement with the assumed spectrum in the simulation.
Meanwhile, multi-scale analysis method is used to study the feasibility of identifying the primary iron nucleus with energies at the knee region. After the multi-fractal analysis and wavelet transformation analysis are applied to the distribution of the secondary particles on the detecting surface, two characteristic parameters are got: the exponents of multi-scale moment and wavelet transform moment. Using the exponents at the order q equal to 4, 6, 8 as the inputs, an artificial neural network that can be used to identify the component of primary iron is constructed and trained and tested using Monte Carlo data sample. The data sample is divided into groups according to different hits number, and multi-parameter analysis is applied to each group. The best result is gotten in the group with hit number greater than 20000, It can pick out about 56.4% iron events and the rejection ratio for other nuclei is about 98.3%, the quality factor is 4.36.
In Summary, based on Monte Carlo simulation of ARGO-YBJ experiment, the space-time information of the charged particles in Extensive Air Showers is used to study the difference between showers induced by different primaries. Multi-parameter analysis is done by using an ANN method with several parameters which can efficiently pick out primary proton induced showers as inputs. With weak model dependence the ANN can efficiently pick out the proton induced events from others. The proton spectrum from 100 Tev to 10 PeV can be obtained using the proton events selected by ANN. Meanwhile, multi-scale analysis method is used to study the feasibility of identifying the primary iron nucleus with energies at the knee region. With several parameters which can describe the fractal characteristics of the event image as inputs, a multi-parameter analysis is done by using ANN method. The result shows that this ANN could efficiently discriminate the primary iron nucleus in "knee" region.
由于宇宙线流强随能量急剧下降,超高能的宇宙线粒子已经不能通过高空气球实验或卫星实验来进行直接观测,只能利用位于地面的探测器阵列通过广延空气簇射(Extensive Air Shower,EAS)过程进行间接的测量。如何有效地区分由不同的原初宇宙线粒子所引起的空气簇射是利用地面实验进行“膝”区物理研究的关键。这一方面要求探测器阵列本身能够提供尽可能多的有关EAS的实验信息,另一方面需要在数据分析过程中采用有效的分辨由各种原初粒子所引起的空气簇射的方法。
中意合作ARGO-YBJ实验是位于我国西藏羊八井宇宙线观测站(大气深度606g/cm~2,海拔高度4300m)的广延大气簇射地面探测实验,该实验采用了由阻性板计数器(Resistive Plate Chamber,RPC)所组成的“地毯式”探测器阵列,具有高的覆盖率(90%以上)和很好的时间、空间分辨率的特性,可以对到达阵列的EAS粒子的时间和空间分布进行较为细致的测量,探测器信号的模拟读出使得阵列的观测能区可以涵盖“膝”区;另外ARGO-YBJ实验阵列处于由“膝”区原初宇宙线粒子所引起的空气簇射的纵向发展极大区域,簇射大小的涨落小且几乎与原初粒子的种类无关,利用该实验的观测信息可以较精确地测量原初粒子的能量。这些特点使得ARGO-YBJ实验在分辨“膝”区原初粒子种类和重建原初粒子能量上具有优势,如果再配以有效的数据分析方法,有望得到较精确的“膝”区原初粒子的分成分能谱,进而为解决“膝”区物理问题提供可靠的实验数据。本文的目的就是针对ARGO-YBJ实验阵列的特点,利用Monte Carlo模拟数据寻找可有效地区分由不同的原初粒子所引起的空气簇射的方法,并研究测量“膝”区原初粒子分成分能谱的可行性。
在本工作中,首先利用CORSIKA程序进行了EAS模拟。为了估计强相互作用模型对分辨结果的影响,在EAS数据模拟中选用了两种最具代表性且与实验结果符合较好的强相互作用模型:QGSJET-Ⅱ和SIBYLL。利用这两种强相互作用模型,细致地模拟了能量在100TeV-10PeV之间的不同原初宇宙线粒子(质子,He核,CNO核,Mg-Si核,铁核)所引起的EAS过程。利用以GEANT3为基础的探测器模拟程序模拟了EAS次级粒子在ARGO探测器中的传输过程和探测器的响应,获得了具有足够统计量的Monte Carlo模拟数据。
利用这些模拟数据分析了由不同的原初粒子所引起的簇射中次级粒子的时间、空间分布特征,得到了平均横向分布宽度
另外,本工作还使用了多尺度分析方法研究了区分“膝”区原初铁核的可行性。对在探测平面上的簇射次级粒子的分布进行多重分形分析和小波变换分析,得到了两个表征簇射次级粒子靶图分形特性的特征量:多重分形矩指数和小波变换矩指数。分别取它们在q=4,6,8时的值作为人工神经网络的输入值,得到了可以有效分辨“膝”区原初铁核成分的人工神经网络。将数据按击中数分段分别进行了多参数分析,分辨结果表明,所得人工神经网络对次级带电粒子数＞20000的事例分辨效果最好,对铁核事例的判选率为56.4%,对其他强子事例的排除率为98.3%,品质因子可达到4.36。
综上,通过对ARGO-YBJ实验条件的细致模拟和对簇射事例的仔细分析,本工作得到了可以用来区分原初粒子种类的特征量。结合人工神经网络进行了多参量分析,并检验了人工神经网络对强相互作用模型的依赖性。结果表明,所得的人工神经网络的分辨效果对强相互作用模型的依赖性较弱,可对“膝”区各原初成分引起广延大气簇射事例做出有效分辨,从而重建出“膝”区质子能谱。同时,本工作还研究了应用多尺度分析方法区分“膝”区原初铁核的可行性,得到了表征簇射次级粒子靶图分形特性的特征量,并结合人工神经网络方法进行了多参数分析。结果表明此网络可有效分辨“膝”区铁核成分。
The energy spectrum of cosmic rays (proton and nuclei from the universe) follows a negative power law: dN/dE∝E~r, but the all-particle energy spectrum ofcosmic rays shows a distinctive feature around 4 PeV, where the spectral index of the power-law changes from -2.7 to approximately -3.1. This feature is commonly called the the "knee", and the corresponding energy region is called the "knee" region. Existence of this feature has been well established experimentally, but there still remain controversial arguments on its origin. To explain this feature, several mechanisms have been proposed. In some of these theoretical models, it is believed that the knee is an intrinsic property of the energy spectrum, related to the acceleration and propagation of the cosmic ray. While in other models, the knee is explained as a new type of interaction at very high energy. Most of these models can describe the obsvered all-particle energy spectrum very well, but the predictions of the individual element spectra in some modles are quite different. So the precise measurements of the individual element spectra at the knee energy region are important for understanding the origin, acceleration and propagation of the cosmic rays.
Since the flux of the cosmic rays decreases rapidly with energy according to a negative power law, the cosmic ray particles with very high energy (VHE) can't be directly detected by balloon- and satellite-borne detectors. These VHE particles can only be indirectly detected by the ground-based detector arrays, which can record the secondary particles in the extensive air shower (EAS) induced by them. The key point in the study of the physics at the knee region in the groud based experiments is to efficiently identify the EAS's induced by different primary particles. To achieve this aim, it requires that the detector array can records sufficient information of EAS, and an effective method to discriminate primary cosmic rays is needed.
The ARGO-YBJ experiment, a collaboration between Chainese and Intalian institutions, is a groud-based EAS detector array, located at the Yangbajing Cosmic Ray Observatory (atmospheric detpth: 606 g/cm2,4300m a.s.1) in Tibet (P.R. China). This experiment utlize a full coverage detector array consisting of Resistive Plate Chambers (RPC). The high coverage (above 90%), good time resolution and fine space granularity makes it able to measure the time and lateral distributions of the secondary particles in a EAS with sufficient precision.With the analog read-out of the RPCs charge, the array can measure the cosmic rays with energies up to the knee region. In addition, at the observation level of this experiment, the EAS induced by cosmic ray particles with energies at "knee" region reach a maximum development, irrespective of the primary mass, so that the shower size is less fluctuated and the energy determination is more precise and less dependent upon the unknown composition. These advantages, together with an effective data analysis method, make it possible to get individual element spectrum in the "knee" region. These experimental information could be helpful for the study of the origin of "knee". In this work we try to find a method for identifying the primary cosmic ray particles with energies at the "knee" region by using Monte Carlo generated data, and study the feasibility of measuring the individuale element spectra in the "knee" region.
The EASs induced by different primary particles are simulated using CORSIKA program. In order to estimate the influence of hadronic interaction models on the identification results, we chose two interaction models for the Monte Carlo simulation: QGSJET-II and SIBYLL, which are representative and have good agreement with the experimental results. We have initiated the EAS with high energy protons, helium, CNO nuclei, Mg-Si nuclei and iron nuclei with energies ranging from 100TeV to 10PeV. The transportation of the EAS particles in ARGO detector and the detector response are simulated indetail with a detector simulation program based on GEANT3. Monte carlo events with sufficient statistics are thus obtained.
After examining the space-time information of the EAS induced by different primaries by using the selected Monte Carlo events, we found that following 5 parameters can be used to characterize the difference between proton and other nuclei induced showers: N_(Hits),
Meanwhile, multi-scale analysis method is used to study the feasibility of identifying the primary iron nucleus with energies at the knee region. After the multi-fractal analysis and wavelet transformation analysis are applied to the distribution of the secondary particles on the detecting surface, two characteristic parameters are got: the exponents of multi-scale moment and wavelet transform moment. Using the exponents at the order q equal to 4, 6, 8 as the inputs, an artificial neural network that can be used to identify the component of primary iron is constructed and trained and tested using Monte Carlo data sample. The data sample is divided into groups according to different hits number, and multi-parameter analysis is applied to each group. The best result is gotten in the group with hit number greater than 20000, It can pick out about 56.4% iron events and the rejection ratio for other nuclei is about 98.3%, the quality factor is 4.36.
In Summary, based on Monte Carlo simulation of ARGO-YBJ experiment, the space-time information of the charged particles in Extensive Air Showers is used to study the difference between showers induced by different primaries. Multi-parameter analysis is done by using an ANN method with several parameters which can efficiently pick out primary proton induced showers as inputs. With weak model dependence the ANN can efficiently pick out the proton induced events from others. The proton spectrum from 100 Tev to 10 PeV can be obtained using the proton events selected by ANN. Meanwhile, multi-scale analysis method is used to study the feasibility of identifying the primary iron nucleus with energies at the knee region. With several parameters which can describe the fractal characteristics of the event image as inputs, a multi-parameter analysis is done by using ANN method. The result shows that this ANN could efficiently discriminate the primary iron nucleus in "knee" region.
引文
[1] V.F. Hess, Phys. Z. 13(1912)1084.
[2] Haungs. A et al, Acta Phys.Polon.B. 35(2004)331
[3] G.V. Kulikov et al., J. Exp. Theor. Phys. 35 (1958) 635
[4] M. Nagano, A.A. Watson, Rev. Mod. Phys. 72 (2000) 689.
[5] P.O.Lagage,C.J.Cesarsky,Astron.Astrophys.,l 18,(1983)223;125,(1983) 249
[6] H.J.Volk,P.L.Biermann,Ap.J,L 65,(1988)33
[7] S.Yoshida,et al,Astropart.Phys.,3(1995) 105
[8] R.J.Protheroe,A.P.Szabo,Phys.Rev.Lett.,69(l 992)2885
[9] 冯存峰,博士毕业论文,山东大学,(1999)
[10] D. Kazanas, A. Nikolaidis, Gen. Rel. Grav. 35(2001)1117.
[11] Jorg R. Horandel, Astroparticle Physics 21 (2004) 241.
[12] R.Ong, IntJ.Mod.Phys., A15S1(2000)740
[13] M.Ichimura et al., Phys.Rev., D48(1993) 1949
[14] T. Antoni, et al.,Astroparticle Physics 24(2005)1
[15] G. Aielli, et al.,Nuclear Physics B (Proc. Suppl.) 166 (2007) 96
[16] M.Amenomori,et al.,Phys.Rev.Let,69(1992)246
[17] M.Amenomori,et al.,Astrophys.J., (1994)
[18] M .Amenomoriet al., Phys.Rev., D47(1993)267
[19] M.Amenomoriet al., Astrophys.J.Let.,415(1993)L14
[20] M .Amenomoriet al .,Astrophys.J.,461(1996)40
[21] M .Amenomoriet al .SCIENCE,314(2006)439
[22] R.Cardaralli, et al.,Nucl.Instr.and Meth.,A333(1993)39
[23] M. Iacovacci, et al. Proceedings of the 28th International Cosmic Ray Conference.(2003)
[24] G. Aielli, et al. Nuclear Instruments and Methods in Physics Research A 562 (2006)92
[25] P. Cretia, et al. 29th International Cosmic Ray Conference Pune (2005) 8, 97-100
[26]http://argo.na.infn.it/
[27]K.Kasahara,Proc.24th ICRC,1(1995)39
[28]D.Heck et al.,Proc.25th ICRC,6(1997)24
[29]D'Etorri Piazzoli and G.Di Sciascio,Astrop.Phys.2(1994)19
[30]D.Heck et al.,Report FZKA 6019,Forschungszentrum Karlsruhe,1998.
[31]Gaisser.T.K,Honda.M.Ann.Rev.Nucl.Part.Sci.52(2002)153.
[32]Jorg R.Horandel,Astroparticle Physics 19(2003)193.
[33]R.Brunet al.,Geant 3,CERN program Lib,(1987)
[34]D.Martello et al.,Internal Note,ARGO-YBJ,1(2000)
[35]G.D.Sciascio,Internal Note,ARGO-YBJ,2(2002)
[36]G.D.Sciascio,E.Rossi Internal Note,ARGO-YBJ,4(2003)
[37]孔繁敏,博士毕业论文,山东大学,(1999)
[38]P.J.Werbas,Beyond Regression,New Tools for Prediction and Analysis in the Behavioral Sciences,Thesis,Harvard University,(1974)
[39]张乃尧,阎平凡,神经网络与模糊控制,清华大学出版社,(2000)。
[40]Hocker.A et al,.E-print arXiv:physics/0703039
[41]J.Kempa,J.Phys.G.,20(1994)215
[42]J.Kempa,Proc.25th ICRC,1(1995)317
[43]I.De Mitri,Nucl.Instr.Meth.A525(2004)132.
[44]B.B.Mandelbrot,The Fractal Geometry of Nature,Freeman,S an Francisco.(1982)
[45]王东升,曹磊,《混沌、分形及其应用》,中国科学技术大学出版社,(1995)
[46]A.Haungs et al.,Nucl.Instr.Meth.A 372(1996)515
[47]J.W.Kantelhardth et al..Physica A.220(1995)219.
[48]Kenneth.R.Castleman,朱志刚等译,Digital Image Processing,电子工业出版社,(1998)
[49]S.K.Aditya et al.,SPIE,(1996)2762
[50]J.Morlet et al.,Geophysics.,47(1982)203-236
[51]J.Morlet et al.,Springer-Verlag,1(1983)233-261
[52]Paul S.Addison,The Illustrated Wavelet Transform Handbook,Institute of Physics,(2002)
[53]Mladen Victor Wickerhauser,Adapted Wavelet Analysis From Theory to Software,AK Peters Ltd,(1994)
[54]Ingrid Daubechies,Ten Lectures on Wavelets,Society for Industrial and Applied Mathematics,(1992).
[55]马丽娜,博士毕业论文,山东大学,(2005)。
[2] Haungs. A et al, Acta Phys.Polon.B. 35(2004)331
[3] G.V. Kulikov et al., J. Exp. Theor. Phys. 35 (1958) 635
[4] M. Nagano, A.A. Watson, Rev. Mod. Phys. 72 (2000) 689.
[5] P.O.Lagage,C.J.Cesarsky,Astron.Astrophys.,l 18,(1983)223;125,(1983) 249
[6] H.J.Volk,P.L.Biermann,Ap.J,L 65,(1988)33
[7] S.Yoshida,et al,Astropart.Phys.,3(1995) 105
[8] R.J.Protheroe,A.P.Szabo,Phys.Rev.Lett.,69(l 992)2885
[9] 冯存峰,博士毕业论文,山东大学,(1999)
[10] D. Kazanas, A. Nikolaidis, Gen. Rel. Grav. 35(2001)1117.
[11] Jorg R. Horandel, Astroparticle Physics 21 (2004) 241.
[12] R.Ong, IntJ.Mod.Phys., A15S1(2000)740
[13] M.Ichimura et al., Phys.Rev., D48(1993) 1949
[14] T. Antoni, et al.,Astroparticle Physics 24(2005)1
[15] G. Aielli, et al.,Nuclear Physics B (Proc. Suppl.) 166 (2007) 96
[16] M.Amenomori,et al.,Phys.Rev.Let,69(1992)246
[17] M.Amenomori,et al.,Astrophys.J., (1994)
[18] M .Amenomoriet al., Phys.Rev., D47(1993)267
[19] M.Amenomoriet al., Astrophys.J.Let.,415(1993)L14
[20] M .Amenomoriet al .,Astrophys.J.,461(1996)40
[21] M .Amenomoriet al .SCIENCE,314(2006)439
[22] R.Cardaralli, et al.,Nucl.Instr.and Meth.,A333(1993)39
[23] M. Iacovacci, et al. Proceedings of the 28th International Cosmic Ray Conference.(2003)
[24] G. Aielli, et al. Nuclear Instruments and Methods in Physics Research A 562 (2006)92
[25] P. Cretia, et al. 29th International Cosmic Ray Conference Pune (2005) 8, 97-100
[26]http://argo.na.infn.it/
[27]K.Kasahara,Proc.24th ICRC,1(1995)39
[28]D.Heck et al.,Proc.25th ICRC,6(1997)24
[29]D'Etorri Piazzoli and G.Di Sciascio,Astrop.Phys.2(1994)19
[30]D.Heck et al.,Report FZKA 6019,Forschungszentrum Karlsruhe,1998.
[31]Gaisser.T.K,Honda.M.Ann.Rev.Nucl.Part.Sci.52(2002)153.
[32]Jorg R.Horandel,Astroparticle Physics 19(2003)193.
[33]R.Brunet al.,Geant 3,CERN program Lib,(1987)
[34]D.Martello et al.,Internal Note,ARGO-YBJ,1(2000)
[35]G.D.Sciascio,Internal Note,ARGO-YBJ,2(2002)
[36]G.D.Sciascio,E.Rossi Internal Note,ARGO-YBJ,4(2003)
[37]孔繁敏,博士毕业论文,山东大学,(1999)
[38]P.J.Werbas,Beyond Regression,New Tools for Prediction and Analysis in the Behavioral Sciences,Thesis,Harvard University,(1974)
[39]张乃尧,阎平凡,神经网络与模糊控制,清华大学出版社,(2000)。
[40]Hocker.A et al,.E-print arXiv:physics/0703039
[41]J.Kempa,J.Phys.G.,20(1994)215
[42]J.Kempa,Proc.25th ICRC,1(1995)317
[43]I.De Mitri,Nucl.Instr.Meth.A525(2004)132.
[44]B.B.Mandelbrot,The Fractal Geometry of Nature,Freeman,S an Francisco.(1982)
[45]王东升,曹磊,《混沌、分形及其应用》,中国科学技术大学出版社,(1995)
[46]A.Haungs et al.,Nucl.Instr.Meth.A 372(1996)515
[47]J.W.Kantelhardth et al..Physica A.220(1995)219.
[48]Kenneth.R.Castleman,朱志刚等译,Digital Image Processing,电子工业出版社,(1998)
[49]S.K.Aditya et al.,SPIE,(1996)2762
[50]J.Morlet et al.,Geophysics.,47(1982)203-236
[51]J.Morlet et al.,Springer-Verlag,1(1983)233-261
[52]Paul S.Addison,The Illustrated Wavelet Transform Handbook,Institute of Physics,(2002)
[53]Mladen Victor Wickerhauser,Adapted Wavelet Analysis From Theory to Software,AK Peters Ltd,(1994)
[54]Ingrid Daubechies,Ten Lectures on Wavelets,Society for Industrial and Applied Mathematics,(1992).
[55]马丽娜,博士毕业论文,山东大学,(2005)。