The Real-time Data Monitoring In DELPHI - CERN
DELPHI Collaboration Bi oY DELPHI 95-31 DAS 164 21 March, 1995 The real-time data monitoring in DELPHI V. Chorowicz CERN-PPE, on leave from LPNHE, CNRS/Paris VLVII Abstract This documents describes the general architecture of the online monitoring of DELPHI data : real-time data analysis program, automatic histograms analysis package and histogram/Ntuple presenter. The reader must be familiar with the DELPHI nomenclature about file organisation. This frame is used both for central data monitoring and sub-detector monitoring. Postscript : DELPHISONLINE:[LOCAL CONTROL.MONITOR.DOCJONL MONITOR.PS URL: http://delonline.cern.ch/delphi online/local control/monitor/doc/www manual / onl monitor.html
CONTENTS THE REAL-TIME 1.1 Spying mechanism . SS MONITORING PROGRAM .« a e 8 e ee & hl hl we & hl hl hl hl ew @ eS Chapter 1 a SS Be Preface. a ee ee SS ee eee ee fF oe NO i eee ee ee ee tt tw tll tl hl tl hl hl tl tll i Chapter 3 THE AUTOMATIC 3.1 Histogram analysis. 3.2 Ntuple production. SS HISTOGRAM - * oe ee eee ee ee eee ee ANALYSIS eee ee ee hell eee eee . lel el hl ell hl lhl hl a 7 Chapter 4 ee THE HISTOGRAM Se FF hl hl hl lhl tl lhl lhl lhl thle hl tl hl lhl hl tl hl hl hl hl hl hl tlt hl hl hl tl hl hl hl hl hl hl hl hl hl tlt i Sa tl tml SS ee ee 8 ee eee mellem lhl hl hl lhl hl ml hl Se hl hl ml hl hl hl 10 hl hl hl ml hl hl UC 10 Olt AND NTUPLE PRESENTER. 11 4.1 Access to histograms. 4.2 Access to Ntuples . 11 oo. eo 8 # ee « ee oe 6 8 . ee e& 4.3 Configuration file . 4.4 Running the presenter . 4.5 Histogram help files . Appendix A REFERENCES c . 8 e& 6 @ & 6 ee & ee ee 8 ee ee ee ee & ee ee e& ee & @ @ e@ @ & e@ e@ ee ee e@ e@ @ & @ 09 & ee oe oe &@ ee 7 oe oe ee ee ee we we ee 2» ew he ee we ew hl hl we hl ee ee hl hl hl hl hl hl ml hl hl Ml hl hl eh hl rhlhl hl hl hl hl hl hl hl hl hl tls hl hl Ohl hl hl hl hl hl lt ee he ell 11 tl 12 12 ew ee @ 0@ 8 6 w@ @ @ 12 15
Preface This document describes the general frame for quality checking inside DELPHI. Other documents listed in the Appendix A give more details on the real implementations of that system in different cases: central monitoring program [3], test of a single detector code in the central monitoring frame [4], local monitoring skeleton for detectors [5]. The Chapter 1 describes the structure of the software including the external libraries and the include files and the mechanisms used to access in real time the raw data and the way to manipulate the histograms. The two main implementations of that system and the differences between them are are roughly described in Chapter 2. The automatic analysis of the histograms is treated in Chapter 3 together with the production of Ntuples from this analysis. Finally the Chapter 4 describes the histogram and Ntuple presenter used to display the monitoring and analysis results. Preface v
CHAPTER 1 THE REAL-TIME MONITORING PROGRAM A copy of the data that are being readout on Fastbus is sent to a VAX or VAX-station on the surface. As many events as possible are spied from the VAX memory (Section 1.1), and a Fortran program analyses them in order to produce histograms or Ntuples (Section 1.2). The program structure is described in Section 1.3. The external libraries are listed in Section 1.5. 1.1 Spying mechanism The data are stored in the VAX memory under the control of MBM (Model Buffer Manager, CERN-ECP-DS product). The Section 2.1 and Section 2.2 give some details on the transport of the data to the surface according to the implementation. Three home-made routines provide access to the data in MBM. shareable library MJUMBM [1] (Monitor Job Utilities). They are integrated in the Function MJU MBM INIT This function will initialize MBM. Function MJU EVENT SELECTION This function is used to set up the event selection criteria. The first argument specifies the event type that should be sampled on an ’as many as possible basis’ the second argument specifies the event selection criteria which should be sampled on a 7100 %’ basis. By default the selection criteria will always give 100% of the so called control records. The second argument should never be used for ordinary monitoring tasks but has been added for the use in data loggers (which requires a 100% sample) and for event displays (which do not have to be data driven). e Function MJU MBM DATA This routine will release the currently locked buffer in MBM and request a new buffer. The return code of this routine indicates if an event was present or not. If not the event flag associated with the MBM service is cleared, and will be set again when data becomes available, upon which the MJU MBM DATA 1.2 should be called again. Production of histograms The HST package [2] is used to manipulate the histograms both for global sections and files. This package provides a lot of routines for initialisation, filling and saving. The routines with the name HST xxxxx are contained in the library HST LIB DELPHI ONLINE:[HISTOGRAM HANDLING.HST]. (file HST.OLB in the directory The real-time monitoring program 1
At initialisation time (routine HST INIT HST) the global section is created with the same name as the process (e.g. MONITOR CP) and several files are created in the directory defined by the logical name HST RZ DIR (all extensions are HST HIST): LOCK for locking, CURRENT which is not used, and MONITOR to save temporary files (see below). Up to the year 1994, there was a problem when trying to create the global section if the presenter or AHA is mapped onto it! A modification is being done to allow the restarting of the monitor even though the presenter or AHA is already mapped onto it: the global section name will include the numeric process identificator relative to the monitoring task. This library provides in particular a locking mechanism to avoid messing up the histograms in the global section. There are several possibilities to save the contents of the global sections. The related files are located in the directory with logical name HST RZ DIR (in general it is DETECTOR HISTOS or one of its sub-directories). HST OUTPUT HISTO(RUN”.) HST OUTPUT HISTO?CMONITOR.”.) saves to the file HST MONITOR (which is usually MONITOR.HST HIST) overwritting the previous contents. saves in the file described by the logical name HST RUN which is usually RUNxxxxxx.HST HIST, HST OUTPUT HISTO(REF’.) is used to update one histogram in the reference (logical name HST REF which is usually REFHST HIST). The HBOOK files produced have a sub directory with the same name as the process (for instance /MONITOR CP). 1.3. Program structure The monitoring program takes one event at a time in the MBM and processes it. It accu- mulates the statistics into histograms stored in a global section (GS) on the same machine. The code is organized in files which contain several routines. For historical reasons, the file organisation is slightly different depending on the implementation. The routines are described here after. e MONITOR The main program is called MONITOR. name, SMI domain, EMU, It does the necessary initalisations (process event flags, signals, HST package and communication with AHA and DIM) and is data driven: whenever an event is available in the MBM, an event flag (EF) is received, the event is fetched and analysed. Other EFs are used in order to be informed of the status of LEP and data logger, or to receive commands from the presenter (to reset the histograms in the GS) or to receive timer interrupts. The program will stop whenever there are more than 2 errors within N seconds (N is programmable). Before exiting, it automatically calls the routine e END RUN CHECK Exit handler. It has been declared with MJU INIT. Routines called at initalisation (on top of MJU *, e 2 USER INIT Used to load calibration constants for example. BOOK HISTOGRAMS Used to book histograms or Ntuples. The real-time monitoring program SMI *, HST *):
The routines called upon reception of an EF are presented below: DECODE LEP It decodes the LEP status. called: e NEW FILL e END FILL May contain user code. If we are at a begin or end of fill the relevant routine is May contain user code. HANDLE PRES COMMAND It’s called when the presenter requests a reset of one or all histograms. The GS will be cleared. The name of the signal (which is the same as the monitor process name) is declared in the presenter configuration file. MJU MBM DATA and XANAL XANAL is called by MONITOR after an event has been fetched with MJU MBM DATA. Calls to the routines MJU TMOUT and MJU TMOUT RSET avoid to go into infinite loop. XANAL first calls UXREAD(?**PMB’,LINK) in order to know the LINK to the top bank where information about date and event type is stored. Depending on the event type it calls the relevant routines: e NEWRUN Called when the first event is analysed after the startup of the program and at each beginning of run. May contain user code. e ANALYS CHKPNT It’s called at each check point and it usually saves the GS to a temprorary file. May contain user code. e ENDRUN Self explanatory! May contain user code. It is the main event analysis routine, it contains the user code. Other important routines (these ones are not used with the old implementation of the local monitoring. See Section 2.2) are: SEND AHA COMM to send commands to AHA for instance to wake it up in order to analyse a file. SAV HIST and UPDATE FILL FILE to save histograms in files: saves the GS to the RUN file with name RUNxxxxxx.HST HIST and also updates the fill statistics file (FILLxxxxxx.HST HIST) by adding the contents of the GS. May contain user code. The real-time monitoring program 3
1.4 Include files A lot of include files are needed, the one used by all the monitoring tasks are listed below: DELPHI ONLINE:[LOCAL CONTROL.MONITOR]ZEBRADEF INC DELPHI ONLINE:[LOCAL CONTROL.MONITORJACCNTDEFINC DELPHI ONLINE:[LOCAL CONTROL.UTILITIESIMJUMSGDEFINC DELPHI ONLINE:[LOCAL CONTROL.CFI.CFILSERVER.SOURCE]DT TRANS.INC DELPHI ONLINE:[COMMUNICATIONS.DIM]DIC.INC DELPHI ONLINE:[HISTOGRAM HANDLING.AHA.SOURCE]JAHA DIM.INC DELPHI ONLINE:[LEP.LEPCOMM.ELEMENTARY PROCESS.SOURCE]L STATING EP DELPHI SPECIFIC:[(CP.MONITOR.SOURCE]JHISTINC 1.5 Libraries and shareable images Here is the list of all the external libraries linked with the monitoring tasks: e DELPHI ONLINE:[LOCAL CONTROL.UTILITIES]UTILITIES e DELPHI SPECIFIC:[LEP.LEPCOMM.ELEMENTARY PROCESS]LEP SIGNAL DELPHI SPECIFIC:[(CP.MONITOR.DETECTOR]VD LIB The VD is the only detector which uses a specific library in central monitoring. HST LIB e CERN libraries in CERN ROOT([LIB] PAWLIB, MATHLIB, PHTOOLS, GRAFLIB, GRAFGKS, PACKLIB, KERNLIB, GKS Also shareable images are used: e SYS SHARE:DECW XLIBSHR, e SMIRTL e DIMRTL MJUMBM, e EMU SHARE UXCOM 1.6 MBM CODE SHARE, Command the environment. EMMBM CODE SHARE files The command ONLSETUP implementation. VAXCRTL, DECW DWTLIBSHR The [LOCAL CONTROL.MONITOR] symbols and logical names might invokes a setup file to define be different depending on the Nevertheless four command files are used to make, execute and stop the monitoring process: 4 MONITOR.MAKE to make the executable. The real-time monitoring program
MONITOR.START to start the process; either in batch or in debug mode and analysing events in real time or in playback on any kind of raw data files. By default it is started in batch in the queue USER CONTROL of the dedicates machine (detector workstation or central nodes) and runs in real time on the MBM data. Qualifiers are used to specify a different running mode: /NODE /DEBUG to run interactively in debug mode. /PLAYBACK /HELP displays information on how to use the command. to run in batch on a different node. to run on raw data files instead of real time from MBM. MONITOR.JOB is executed by the previous one and it defines some logical names: * monitor image : name of the executable. * delta time : minimum execution time for image execution retry. endrun procedure : command procedure to be called from the monitor task. MONITOR.STOP to stop the process, it uses FORCEX. The monitoring task must not be killed with STOPP. The real-time monitoring program 5
CHAPTER 2 IMPLEMENTATIONS Two main implementations are used in DELPHI: the central monitoring for quality checking [3] and the detectors local monitoring [5]. Local is used in opposition to central in the same way as for the data acquisition system: while central concerns the whole of DELPHI, local is relative to a specific partition (i.e. more or less one detector). Each of these implementations use all the routines above but they are not organised in the same way inside the files. The user code is of course completely different. 2.1 Implementation for central monitoring The aim of the central monitoring processes is to provide feedback on the data quality for the person on central shift. General figures such as data sizes, beam pickup timing, trigger informations, and detector related quantities are analysed (detector means one DELPHI sub-detector). More information may be found in [3]. The copy of the raw data events from the CHI Fastbus module into the MBM is done through the DRB32 interface of VXDEAQ (RPC protocol). Only the events which have passed the third level trigger are transfered. The data are also copied on VXDEZO after event selection. The same executable is used. The routines described in Section 1.3 are grouped in 4 files: MONITOR.FOR contains the main program MONITOR, XANAL.FOR contains the routine XANAL, ANALYS.FOR contains the routine ANALYS plus some utility routines, and HANDLERS.FOR contains all the rest. All these routines are compiled and grouped in the library MONITOR.OLB DELPHI SPECIFIC:[CP MONITOR]. in the directory On top of that, there is one routine per detector: e xx CP MON The detector xx provides a fortran file xx CP MON.FOR which is copied in DELPHI SPECIFIC:[CP MONITOR.DETECTOR.SOURCE] together with the necessary data files (calibrations for example). There must be at least the 2 routines FILL xx. All the detectors are included (Trigger Partition as well). BOOK xx and These routines are compiled and put together in the library DET CP MON.OLB directory DELPHI SPECIFIC:[(CP MONITOR.DETECTOR]. in the Implementations 7
2.2 Implementation for local monitoring The local monitoring provides detailed feedback to the detector groups on their own data accepted by the second level trigger. While in the central monitoring, detector s can not provide too many histograms so that the shifter is not flooded with informa tion, in the local monitoring the informations can be very detailed. The spying mechanism is slightly different from the central one: the data are sent from the detector LES FIP to the detector workstation (MBM) via TCP/IP by the FIP SERVER process. Old implementation In the “old” version (up to 1993) the whole program was supposed to be build from the standard library MONITOR.OLB in DELPHI ON LINE:[LOCAL CONTROL.MONITOR] and only the routine ANALYS in file ANALYS.FOR was supposed to be exported to local directories. But the routines of the type NEWRUN , ENDRUN where included in the library MONITOR so most of the detectors have copied to their own area (DETECTOR SPECIFIC:[MONITOR]) the whole code. This system is very difficult to maintain that’s why we have changed the organisation of the files for the local monitoring. Thus this old version is not maintained anymore. New implementation A new directory has been assigned: DELPHI ONLINE:[LOCAL CONTROL.MONITOR.DETECTOR] and only the standard routines which never need to be modified are in the library MONITOR. All the routines which may need user modification are in the file USER. ROUTINES.FOR which will be copied to the detector area. This new version takes bebefit from the developments done over the years in the central monitoring: e Managing the fill statistics Communication with AHA (via DIM) and the presenter (for the reset of the global section). Emu messages including automatic beeping. More information may be found in [5]. 8 Implementations
CHAPTER 3 THE AUTOMATIC HISTOGRAM ANALYSIS During the first years of data taking, the correctness of the histogr ams has been checked by eyes by comparing the histograms to their references. The AHA package has been developped in DELPHI to analyse automatically the histograms according to several methods described in the following sections. AHA can also fill Ntuples with the histogram fit parameters as well as merge several histogram files. It contains facilities to communicate histogram presenter via DIM [6]. AHA works 3.1 Histogram analysis in an automatic way with the other packages like monitoring tasks, the by means of a configuration file and it uses the HST package to manipulate the histograms. A lot of refined possibilities are available but only the main features are reported here. The reader should refer to [7] for more details. Each histogram may be compared to its reference according to 3 methods: Bin by bin: each bin (or group of bins) is compared to the related reference bin. Peak: assumes that there is a single peak in the histogram. Shape: either by chi square metod or Kolmogorov-Smirnov method. Another method used is the 3.2 Hole-spike finder: no reference histograms is needed for this method. The bins are grouped and checked. Known faulty bins may be specified such that they are not taken into account in the analysis. Ntuple production AHA may fill Ntuples as the result of the histogram standard DELPHI structure (QC Ntuples [8]). analysis. These Ntuples have the AHA performs some fitting on the histograms to extract rms or mean value or can provide any kind of information on a histogram (e.g. number of entries or averaging of bins contents) and stores them in an Ntuple which can be later displayed by the presenter. Having information in the form of Ntuples allows to display the evolution of the quantities as a function of time in order to spot either slow drifts of the variables or to know the exact time of occurence of sudden changes. The automatic histogram analysis 9
3.3 Running AHA Specific symbols are defined in the DCL HANDLING]. table by executing ONLSETUP [HISTOGRAM Once the configuration file is set up, it’s straightforward to run AHA: AHA START conf file (/BATCH). AHA START/HELP gives all the relevant informations on that. AHA can run as a background task and be awaken regularly to analyse the global section in which case DIM is ussed to send the result to the presenter. It can COMM) 3.4 be instructed to spawn itself to analyse an histogram file (routine and the diagnostic file (see below) is used to inform the presenter. SEND AHA Error reporting Several ways of reporting the errors or passing the comparisons results have been implemented in AHA: e the diagnostic file (defined with ERR MESS) provides a summary of the comparisons. the LOG FILE provides detailed information on the comparisons. the MAIL utility allows to send E-mail messages. 10 It is then used by the presenter to display the results. the DIM services are used to communicate with other tasks. the SMI states (domain QUALITY) reports on the comparisons. EMU messages may be sent and seen by the users on EMU displays (including automatic beeping). The automatic histogram analysis
CHAPTER 4 THE HISTOGRAM AND NTUPLE PRESENTER The presenter [9] [10] [11] [12] has evolved continuously along the years includi ng all the new features of the system as they were available: global sections (GS), MOTIF, AHA results, Ntuples, DIM, WWW Xmosaic browser. The presenter can access either global sections or HBOOK files (histograms and Ntuples), it displays the AHA results and uses the Web to give specific informations on the histogram contents and DELPHI running. It is also used to update the reference histogra ms: every single histogram may be updated. It is fully driven by a configuration file (see Section 4.3) and it uses the HST package to manipulate the histograms. The histogram 140 used to store information relative to the run number and the number of events analysed is decoded by the presenter and the information are displayed in the header. 4.1. Access to histograms The two access modes are quite different: ONLINE: the presenter maps the global section (for reading) and displays in real time the statistics which are updated regularly (when AHA sends its results or when you go back to the first page or on request). The presenter accesses the remote GS Ge. if GS and presenter are not on the same node) by decnet and it needs a valid user-id and password to do so. FROM FILE: the contents of the HBOOK file (histo or Ntuples) are displayed. directory and the RZ subdirectory are defined in the configuration file. The A reference set of histograms can be displayed as well and there is the choice between the default reference file and any other histogram file (all the relevant informations are defined in the configuration file). 4.2 Access to Ntuples A specific Ntuple structure is used (QC Ntuples [8]) The Ntuple contents may then be plotted as a function of time (trace plot). or run/fill number. Also two dimensional plots can be displayed. These Ntuples have a specific header containing information such as data taking time or run/fill number. A selection panel allows to choose the entries which will be displayed. The selection may be performed on: e Run number, The histogram and Ntuple presenter 11
Fill number, Data taking time, e Last N days. 4.3 Configuration file For histograms the configuration file contains all the information relative to the e allowed access modes, the reference file, the location of histograms and help files, the place where are the AHA results, e the list of histograms to be displayed together with their characteristics (color, zones, stat, ref, etc). For Ntuples, the configuration file is different. It specifies the location of the Ntuple file the booking of histograms with the Ntuple variables e the list of histograms to be displayed together with their characteristics (color, zones, 4.4 stat, ref, etc). Running the presenter A symbol is defined at the DELPHI level: DELPHI PRES It may will start the presenter in interactive mode. be ran in batch in which case proxy logins are used to access Hence no user-id nor password is asked for remote GS. the global section. All subsequent commands to the presenter are issued from the presenter windows itself using the mouse buttons and popup menus. Several configuration files may be opened in parallel and one can switch from one to the other. 4.5 Histogram help files There might be one help file per histogram (that is the case since 1994 for the central monitoring) containing information on what the plot must not be and what to do in case of trouble. Since we now use WWW to display them is very convenient for several reasons. It in case of trouble. Unfortunately only group annotations (it would allow every any other DELPHI member). the help files are in HTML format. Using the Web allows for annotations in the help files themselves personnal annotations are available and not the DELPHI member to see the annotations made by It is also important to scan informative files like detector descriptions or how the experiment is currently running. 12 The histogram and Ntuple presenter
A standard skeleton has been setup for the central monitoring histogr ams in order to make the life easier to the shifter. The histogram and Ntuple presenter 13
APPENDIX A REFERENCES Most of these documentations are also available in the bookreader format Wide Web from the DELPHI home or online pages: URLs of the type http://delonline.cern.ch/ delphi online/local control/xxxx/doc/www manual/file and on World htm]. [1] MJU routines: DELPHI ONLINE:[LOCAL CON TROL.UTILITIES.DOC]UTILITIES.PS [2] HST package documentation: DELPHI ONLINE:[HISTOGRAM HANDLING.HSTIHST DOC [3] The DELPHI central data quality monitoring. V.Chorowicz. Delphi note 95-32, DAS-165. DELPHI SPECIFIC:[(CP.MONITOR.DOCICENTRAL MON.PS. [4] Preparing a detector code in central monitoring . V.Chorowicz. Delphi note 95-34, DAS-167. DELPHI ONLIN :[LOCAL CONTROL.MONITOR.DETECTOR.DOC]TEST DET INCP.PS [5] The DELPHI skeleton for detector online monitoring. V.Chorowicz. Delphi note 95-33, DAS-166. DELPHI ONLINE:[LOCAL CONTROL.MONITOR.DETE CTOR.DOC]JLOCAL MON.PS [6] DIM: A distributed Information Management system for the DELPHI C.Gaspar. DELPHI note 94-4 DAS-148. DELPHI ONLINE:[COMMUNICATIONS.DIM.DOC]DIM.PS experiment, [7] The DELPHI online automatic histogram analysis, C.Lacasta. DELPHI note 94-38 DAS-156. DELPHI ONLINE:[HISTOGRAM HANDLING.AHA.DOC]AHA.PS [8] Standard Ntuples for the DELPHI quality checking: Y.Belokopytov DELPHI ONLINE:[HISTOGRAM HANDLING.QC NTUPLE.DOCINTPS [9] The DELPHI histogram presenter, L.Beneteau, T.Camporesi. DELPHI note 92-100 DAS-129. [10] The DELPHI histogram presenter reference manual, P.Giacomelli, L.Beneteau, T.Camporesi. DELPHI ONLINE:[HISTOGRAM HANDLING.PRESENTER.DOC]DELPHI PRESENTER REF.PS [11] The DELPHI histogram presenter user manual, L.Beneteau and T.Camporesi. DELPHI ONLINE:[HISTOGRAM HANDLING.PRESENTER.DOC]JDELPHI PRESEN TER USER.PS REFERENCES 15
[12] The DELPHI histogram presenter upgrade: displaying Ntuples, Y. Belokopytov. DELPHI ONLIN E:[HISTOGRAM HANDLIN G.PR ESENTER.DOC]DP UPGRADE V3.PS:6 16 REFERENCES
DELPHI Collaboration Bi DELPHI 95-31 DAS 164 oY 21 March, 1995 The real-time data monitoring in DELPHI V. Chorowicz CERN-PPE, on leave from LPNHE, CNRS/Paris VLVII Abstract This documents describes the general architecture of the online monitoring of DELPHI data : real-time data analysis program, automatic histograms analysis .
May 02, 2018 · D. Program Evaluation ͟The organization has provided a description of the framework for how each program will be evaluated. The framework should include all the elements below: ͟The evaluation methods are cost-effective for the organization ͟Quantitative and qualitative data is being collected (at Basics tier, data collection must have begun)
Silat is a combative art of self-defense and survival rooted from Matay archipelago. It was traced at thé early of Langkasuka Kingdom (2nd century CE) till thé reign of Melaka (Malaysia) Sultanate era (13th century). Silat has now evolved to become part of social culture and tradition with thé appearance of a fine physical and spiritual .
On an exceptional basis, Member States may request UNESCO to provide thé candidates with access to thé platform so they can complète thé form by themselves. Thèse requests must be addressed to esd rize unesco. or by 15 A ril 2021 UNESCO will provide thé nomineewith accessto thé platform via their émail address.
̶The leading indicator of employee engagement is based on the quality of the relationship between employee and supervisor Empower your managers! ̶Help them understand the impact on the organization ̶Share important changes, plan options, tasks, and deadlines ̶Provide key messages and talking points ̶Prepare them to answer employee questions
Dr. Sunita Bharatwal** Dr. Pawan Garga*** Abstract Customer satisfaction is derived from thè functionalities and values, a product or Service can provide. The current study aims to segregate thè dimensions of ordine Service quality and gather insights on its impact on web shopping. The trends of purchases have
Chính Văn.- Còn đức Thế tôn thì tuệ giác cực kỳ trong sạch 8: hiện hành bất nhị 9, đạt đến vô tướng 10, đứng vào chỗ đứng của các đức Thế tôn 11, thể hiện tính bình đẳng của các Ngài, đến chỗ không còn chướng ngại 12, giáo pháp không thể khuynh đảo, tâm thức không bị cản trở, cái được
Le genou de Lucy. Odile Jacob. 1999. Coppens Y. Pré-textes. L’homme préhistorique en morceaux. Eds Odile Jacob. 2011. Costentin J., Delaveau P. Café, thé, chocolat, les bons effets sur le cerveau et pour le corps. Editions Odile Jacob. 2010. Crawford M., Marsh D. The driving force : food in human evolution and the future.
Le genou de Lucy. Odile Jacob. 1999. Coppens Y. Pré-textes. L’homme préhistorique en morceaux. Eds Odile Jacob. 2011. Costentin J., Delaveau P. Café, thé, chocolat, les bons effets sur le cerveau et pour le corps. Editions Odile Jacob. 2010. 3 Crawford M., Marsh D. The driving force : food in human evolution and the future.
