forked from chunshen1987/iEBE
-
Notifications
You must be signed in to change notification settings - Fork 0
/
readme.txt
129 lines (93 loc) · 11.9 KB
/
readme.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
===================
iEBE Readme
===================
The iEBE package is a convenience package to automate event-by-event hybrid calculations. It divide calculatons into "jobs", where each job consists multiple "ebe-calculations". Each "ebe-calculation" is a complete hybrid calculation that in execution order performs: heavy-ion event generation (superMC), hydrodynamics simulation (VISHNew), particle emission sampling (iSS), hadron rescattering simulation (osc2u and urqmd), flow calculation (binUtilities), and finally collect important results to databases (EbeCollector). Each "job" runs the given number of "ebe-calculations" sequentially, and "jobs" are run in parallel. The package has utility scripts that can combine the generated SQLite database files from different jobs into one that can be analyzed later.
The main programs are contained in the subfolder "EBE-Node", which is used to perform one job, and which will be duplicated when multiple jobs are desired. The package needs two locations to perform multi-job calculations: one folder is used to store duplications of "EBE-Node" and intermediate results generated during the calculation (refer to as "working folder" in the following), and another folder is used to store final results (refer to as "result folder" in the following). By default the working folder is named as "PlayGround" and the result folder is named as "RESULTS", both in the root directory of the package.
------------------------------------------------------------------
<<1>> How to use the package to perform multi-job calculations
------------------------------------------------------------------
This section explains how to use the highest-level scripts provided by the package to perform event-by-event hybrid calculations. This section should be the only section requires reading for user who are not interested in modifying the package.
*VERY IMPORTANT* Make sure you have Python 2.7+ (or Python 3) installed before proceeding.
In the following all path are relative to the root directory of the package.
Step 1) Generate jobs using the ./generateJobs.py script.
To generate jobs use the generateJobs.py script in the root directory. Most of the runnable scripts in this package provide the feature that if you run it without additional arguments it will print the usage echo, for example:
$ ./generateJobs.py
And you should see the output:
Usage: generateJobs.py number_of_jobs number_of_events_per_job [working_folder="./PlayGround"] [results_folder="./RESULTS"] [walltime="03:00:00" (per event)] [compress_results_folder="yes"]
The echo says that the 1st argument for the script should specify the number of jobs you want to generate; the 2nd argument specifying the number of ebe-calculations for each job; the 3rd argument points to the result folder; the 4th argument specifies the "wall time" (used in torque system, explain later); the 5th argument points to the working folder; and the 6th argument is for whether to compress final results. Except for the first two, all other arguments have default values. The simplest way to generate jobs is just to accept the default values, and as an example to generate 2 jobs, eaching performing 5 ebe-calculations, simply do the following:
$ ./generateJobs.py 2 5
This script will first check required libraries, then compile all the programs if not existing already, the generate the actual folders for jobs.
After you see the echo "Jobs generated.", you should see the working folder "PlayGround" and the result folder "RESULTS" in the root directory if not existing previously.
Step 2) Submit jobs.
The way to submit jobs depends on the system. For a cluster that has "torque" scheduling system (therefore the "qsub" command is availible), submit jobs use the submitJobs_qsub.py script; for a local computation use the submitJobs_local.py script. The difference is that the local computation is only parallelled for the local CPUs and calculation on cluster, via the torque system, will be distributed to multiple nodes. To submit a local calculation simply do (The script knows how to get the location of the working folder automatically):
$ ./submitJobs_local.py
You should see some feedbacks listing the jobs that have been submitted.
Step 3) Checking progress.
Progresses for each job can be checked by the progressReport.py script in the root directory (The script knows how to get the location of the working folder automatically):
$ ./progressReport.py
It will list the current progress for all jobs.
Step 4) Combining databases.
Once all calculations are finished, the generated database files from all events will be combined automatically, and a single file "collected.db" will be generated in the results folder.
---------------------------------------
<<2>> How to analyze generated data
---------------------------------------
The "collected.db" generated from previous steps is the SQLite database file that can be analyzed by any desired means. The recommended way is to use the uhg.py script in the utilities folder. This script not only reads the database, but also performs additional analysis like interpolation along pT, calculation of mean, and etc. It can either be run from command line to evaluate a single expression, or interactively from a shell. To evaluate a single expression run the uhg.py script in the utilities folder:
$ ./uhg.py database_filename "expression to evaluate"
A more convenient way to evaluate multiple expressions as well as perform additional analysis, is to run the uhg.py script interactively. For example:
$ python -ic "from uhg import *"
The interactive mode will also print out a simple help showing recognizable symbols that can be included in the expression.
Another way is to use the databaseQuery.py script to evaluate a single piece of SQL query from command line:
# ./databaseQuery.py database_filename "SQL_query"
A third way is to use the ./unpackDatabase.py script located under /EBE-Node/EbeCollector/ to dump the whole database into separated space-separated text files, each for individual table. For example, running the following under /RESULTS/:
$ ../EBE-Node/EbeCollector/unpackDatabase.py ./collected.db .
will generate several ".dat" files, each containing data for the corresponding type. Each file have a one-line header to indicate what data each column records, and the rest are data separated by spaces.
For more details about the structure of the database and the uhg.py script see /EBE-Node/EbeCollector/EbeCollector_readme.txt.
--------------------------------
<<3>> How to tune parameters
--------------------------------
After you familiarized yourself about how to perform multi-job hybrid calculations, finally the time comes to the question about how to tune parameters for the simulations. The most commonly tuned parameters are in the ParameterDict.py file in the directory, which should be the only file use to direct the simluations. This file will be copied to the result folder for record-keeping purpose when generating jobs.
------------------------------------
<<4>> File map of the package
------------------------------------
/-------------------------
LICENSE # the license of the code package
ParameterDict.py # parameter script
check_prerequisites.py # python script to check pre-requisites packages in the computer
examples/ # this folder stores one example for a full simulation
generateJobs_local.py # python script to generate running jobs (By default, PlayGround and RESULTS folders will be generated)
progressReport.py # python script to check the running status of the jobs
readme.txt # readme file
saved_configs.py # a copy of the running configuration
submitJobs_local.py # python script to submit jobs to local computer
submitJobs_qsub.py # python script to submit jobs to qsub system in cluster
web_frontend/ # this folder stores scripts to build up web front server for online database querying
EBE-Node/ # this folder stores all the source codes
---EbeCollector/ # this folder stores python scripts to collect observable results into sqlite database
---VISHNew/ # this folder stores the source code (in Fortran) for VISH2+1, a (2+1)-d viscous hydrodynamic code
---binUtilities/ # this folder provides a collection of python scripts to collect individual particle information output from UrQMD into desired pT bins
---centrality_cut_tables/ # this folder contains tables for centrality selection in different collision systems
---crank/ # this folder contains python scripts to link all code components together and perfrom full simulation
---fs/ # this folder stores the source code (in C++) to perfrom free-streaming for the pre-equilibrium evolution. It generates initial inputs for viscous hydrodynamics
---iS/ # this folder contains source code (in C++) to conduct Cooper-Frye freeze-out for particle momentum distribution
---iSS/ # this folder stores the source code (in C++) to generate Monte-Carlo samples of particle spatial and momentum information according to Cooper-Fyre Formula
---osc2u/ # this folder stores the source code (in Fortran) to convert the OSCAR output of the particle Monte-Carlo sample from (iSS) into the initial conditions for UrQMD simulation
---photonEmission/ # this folder stores the source code (in C++) to compute thermal photon radiation with hydrodynamic medium generated from VISH2+1
---superMC/ # this folder contains the source code (in C++) to generate fluctuating initial conditions for pre-equilibrium model or viscous hydrodynamics
---urqmd/ # this folder contains the source code (in Fortran) for UrQMD
PlayGround/ # this folder is generated by generateJobs_local.py. All the simulation will be done in this folder
---watcher/ # this folder stores the running script and outputs for a watcher process, which regularly checks the status of the running jobs. Once all the jobs are finished, it will zip all the finalResults in each job folder and move them to RESULTS.
---job-*/ # this folder contains all the simulations scripts and outputs. The substructure in this folder is similar to EBE-Node/. finalResults folder will be generated once the job is running. All the screen outputs will to direct to RunRecord.txt and ErrRecord.txt in the crank/ folder.
RESULTS/ # this folder is generated by generateJobs_local.py. It stores the final results after all the simulations are finished.
---ParameterDict.py # a copy of ParameterDict.py from the root directory
---job-*.zip # zipped file that contains all final results from the simulations
---collected.db # SQLite database that stores final results from all the simulations (all jobs combined)
utilities/ # this folder stores scripts for multi-purposes
---uhg.py # a python script which provides a user interface for SQLite database querying with collected.db in the RESULTS folder
------------------------------------
<<5>> Trouble shooting
------------------------------------
(1) Compile errors in generateJobs_local.py
In some circumstances, one or more executables will not be found after compile. In this case, one needs to go to the corresponding code folder in the EBE-Node/ and types “make” to see exact compile error on the screen.
(2) Jobs finished in few seconds
One typical full event requires about one hour running time. If the submitted jobs finish in only few seconds, it usually means some error occurs during the running time. In this case, one can go to PlayGround/job-*/crank folder and read the RunRecord.txt and ErrRecord.txt files for details about the running errors.
Enjoy!