blackdynamite 1.0.4

Scientific Parametric Study Tool

Quick start

Documentation

The complete documentation of the project is on readthedocs

Installation

The easiest is through pip:

pip install blackdynamite

For a user scope installation (recommended):

pip install --user  blackdynamite

Or directly for the GitLab repository:

pip install  https://gitlab.com/ganciaux/blackdynamite.git

Getting the sources

You can clone the GIT repository:

git clone https://gitlab.com/ganciaux/blackdynamite.git

Installing completion

To benefit the autocompletion for BlackDynamite you need to activate the global completion as described in the argcomplete website: Howto activate global completion.

Introduction and philosophy

Blackdynamite is merely a tool to help managing parametric studies. In details it comprises:

1. Launching a program repeatedly with varying parameters, to explore the chosen parametric space.

2. Collect and sort results of Small sizes benefiting from the power of modern databases.

3. Analyze the results by making requests to the associated databases.

Launching is made simple by allowing any executable to be launched. The set of directories will be generated and managed by BlackDynamite to prevent errors. Requests of any kind will then be made to the underlying database through friendly commands of BlackDynamite.

Collecting the results will be possible thanks to the Blackdynamite C/C++ and python API which will let you send results directly to the database and thus automatically sort them. This is extremely useful. However heavy data such as Paraview files or any other kind of data should not be pushed to the database for obvious performance issues.

Analysis of the results can be made easy thanks to Blackdynamite which can retrieve data information in the form of Numpy array to be used, analyzed or plotted thanks to the powerful and vast Python libraries such as Matplotlib and Scipy.

The construction of a BlackDynamite parametric study follows these steps:

• Describing the parametric space
• Creating jobs (specific points in the parametric space)
• Creating runs (instances of the jobs)
• Launching runs
• Intrumenting the simulation to send results
• Analyzing the results

Setting up a parametric study

The parametrization of a study is done in a YAML file, labelled bd.yaml. It contains the information of the parametric space, spanned exploration and configuration of your simulations. An example of a working study is provided in the example directory.

A study description starts with a provided name in the YAML format:

---

study: bd_study

Choose the parameters of the study

Job description

The first thing to do is to list all the parameters characterizing a specific case. These parameters can be of simple scalar types (e.g. string, integers, floats), however no vectorial quantity can be considered as an input parameter. It describes the Job pattern of the study. This must be defined in a section in the the bd.yaml file. For instance a three parameter space can be declared as:

job:
  param1: float
  param2: float
  param3: str

By default there is one more entry to every job: its unique id.

Run description

Aside from the jobs, a run will represent a particular realisation (computation) of a job. For instance, the run will contain information of the machine it was run on, the executable version, or the number of processors employed. For instance creating the run pattern can be done with:

run:
  compiler: str

By default there are entries created for the user:

• id: the id of the run
• machine_name: the name of the machine where the run must be executed
• nproc: number of processors used to perform the computation (default: 1)
• run_path: the directory where the run will be created and launched
• job_id (integer): the ID of the running job
• state (str): the current state of the run (CREATED, FINISHED, ERROR)
• run_name (string): the name of the run (usually a name is given to a collection of runs, at creation)
• start_time (datetime): time when the run started
• last_step_time (datetime): last time a quantity was pushed to the database

Create the database

Then you have to request for the creation of the database which can be done with a simple command:

canYouDigIt init --truerun

As mentioned, all BlackDynamite scripts inherit from the parsing system. So that when needing to launch one of these codes, you can always claim for the valid keywords:

canYouDigIt init --help

usage: canYouDigIt [--study STUDY] [--host HOST] [--port PORT] [--user USER] [--password PASSWORD] [--bdconf BDCONF] [--truerun] [--constraints CONSTRAINTS]
                   [--binary_operator BINARY_OPERATOR] [--list_parameters] [--yes] [--logging] [--help]

createDB

General:
  --logging             Activate the file logging system (default: False)
  --help                Activate the file logging system (default: False)

BDParser:
  --study STUDY         Specify the study from the BlackDynamite database. This refers to the schemas in PostgreSQL language (default: None)
  --host HOST           Specify data base server address (default: None)
  --port PORT           Specify data base server port (default: None)
  --user USER           Specify user name to connect to data base server (default: tarantino)
  --password PASSWORD   Provides the password (default: None)
  --bdconf BDCONF       Path to a BlackDynamite file (*.bd) configuring current options (default: None)
  --truerun             Set this flag if you want to truly perform the action on base. If not set all action are mainly dryrun (default: False)
  --constraints CONSTRAINTS
                        This allows to constraint run/job selections by properties (default: None)
  --binary_operator BINARY_OPERATOR
                        Set the default binary operator to make requests to database (default: and)
  --list_parameters     Request to list the possible job/run parameters (default: False)
  --yes                 Answer all questions to yes (default: False)

An important point is that most of the actions are only applied when the truerun flag is set.

Creating the jobs

The goal of the parametric study is to explore a subpart of the parametric space. We need to create jobs that are the points to explore.

We need to describe the desired set of jobs, to be explored. This is done in the YAML file describing the study, under the section job_space. For instance it could be:

job_space:
  param1: 10
  param2: [3.14, 1., 2.]
  param3: 'toto'

The actual insertion of jobs can be done with the command:

canYouDigIt jobs create --truerun

You can control the created jobs with:

canYouDigIt jobs info

In the case of our example, 3 jobs should be created as a range of values for the second parameter was provided.

Creating the runs

At this point the jobs are in the database. You need to create runs that will precise the conditions of the realization of the jobs, by giving the value of the run space.

This is specified in the YAML file under the section run_space. For instance with:

run_space:
  compiler: 'gcc'

The default parameters for runs will then be automatically included in the parameters for the not provided ones (e.g. state).

A run now specify what action to perform to realize the job. Therefore, one script must be provided as an entry point to each run execution. This will be given in the YAML file as the exec_file. For instance from the example a bash script is the entry point and provided as follows:

exec_file: launch.sh

Usually, an end-user has a script(s) and configuration files that he wishes to link to the run. This can be done with:

config_files:
  - config.txt
  - script.py

Finally, we have to create Run objects and attach them to jobs, which is done with the command:

canYouDigIt runs create --truerun

After that, all created runs should be present in the database in the state CREATED, ready to be launched. This can be controled with the command:

canYouDigIt runs info

Instrumenting Text simulation files (e.g. a bash script)

BlackDynamite will replace specific text marks in the registered files with the values from the job and run particular point. A precise syntax is expected for BlackDynamite to recognize a replacement to be performed.

For instance:

echo __BLACKDYNAMITE__param1__

shall be replaced by the value of param1 parameter at the run creation.

As an additional example, the script launch.sh taken from the example has lines such as:

echo 'here is the job'
echo __BLACKDYNAMITE__id__
echo __BLACKDYNAMITE__param1__
echo __BLACKDYNAMITE__param2__
echo __BLACKDYNAMITE__param3__

Instrumenting a Python simulation

In a python program, one can benefit from the possibilities of Python to get a handle object on the current job and run. This will also allow to push produced data to the database. This is done by the simplified commands:

# import BlackDynamite
import BlackDynamite as BD
# get the run from the current scope
myrun, myjob = BD.getRunFromScript()

In order to have time entries for runs, the functions start and finish need to be called:

myrun.start()
...
# your code
...
myrun.finish()

Finally, to push data directly to the database, one can use pushVectorQuantity and/or pushScalarQuantity, attached to meaurable quantities:

# pushing vector types (numpy)
myrun.pushVectorQuantity(vector, step, "quantity_id")
# pushing scalar types 
myrun.pushScalarQuantity(scalar, step, "quantity_id")

Executing the runs

Once the runs are created, they can be launched with a command like

canYouDigIt runs launch --truerun

During and after the run the status can be controlled, once again, with:

canYouDigIt runs info

For detailed information on a specific run:

canYouDigIt runs info --run_id RUN_ID_NUMBER

in order to be placed in the context of a specific run:

canYouDigIt runs info --run_id RUN_ID_NUMBER --enter

to execute a specific command

canYouDigIt runs info --exec COMMAND

and applied for a specific run:

canYouDigIt runs info --run_id RUN_ID_NUMBER --exec COMMAND

Manipulating the database

Selecting jobs and runs

All the previous commands may be applied to a subset of runs/jobs. In order to select them one should place constraints, provided by the option --constraint. For instance, listing the runs constraining parameters labeled param1 and param2 could be made with:

canYouDigIt runs info --constraint 'param1 > 1, param2 = 2'

In the exceptional case where parameters of jobs and runs would bear the same name (you should avoid to do that), one can disambiguate the situation with:

canYouDigIt runs info --constraint 'jobs.id > 1, runs.id = 2'

Cleaning Runs

Sometimes it can be necessary to re-launch a set of runs. Sometimes it can be necessary to delete a run. In order to reset some runs, making them ready to relaunch, one should use the following:

canYouDigIt runs clean --constraint 'jobs.id > 1, runs.id = 2' --truerun

To completely delete them:

canYouDigIt runs clean --constraint 'jobs.id > 1, runs.id = 2' --delete --truerun

Altering runs

Rarely it can be necessary to manually change a set of runs. For instance changing the state of a set of runs can be done with:

canYouDigIt runs update --truerun state = toto

Plotting results

For starting the exploration of the collected data, and thus producing graphs, the plotcommand can be employed. While tunable, it cannot produce any type of graphs. However, for quick exploration of the data, one could do:

canYouDigIt runs plot --quantity ekin --legend "%r.id" --marker o

Exporting the results (to zip file)

Under Construction

Fecthing the results

Under construction...