Skip to main content

Light-weight serverless API for OpenModelica

Project description

PyDelica: Serverless OpenModelica with Python

About

PyDelica is an API providing a quick and easy to use interface to compile, customise and run OpenModelica models with Python. Unlike OMPython it does not require the launch of a server session to use OMShell but rather dynamically adjusts files produced after model compilation in order to update options and parameters. The lack of server requirement means models can be run in tandem without calling multiple OMC sessions which can be very taxing on CPU. Furthermore PyDelica is able to detect errors during model runs by reading the stderr and stdout from OpenModelica and throw appropriate exceptions to terminate the program execution.

Installation

To use PyDelica you will require an installation of OpenModelica on your system, the API has been confirmed to work on both Windows and Linux, but if OM is installable on macOS it should also be possible to still use it.

To install run:

pip install pydelica

Getting Started

For the purposes of demonstration the included test model tests/models/SineCurrent.mo will be used.

PyDelica Session

All uses of PyDelica require an instance of the Session class:

from pydelica import Session

session = Session()

Logging

The Session class has one optional argument which is used to set the log level output within OpenModelica itself. The options are based on the -lv flag within OM. By default the level is set to Normal which means no log level output.

As an example if you wanted to run with statistics logging -lv=LOG_STATS you would setup with the following:

from pydelica import Session
from pydelica.logger import OMLogLevel

session = Session(OMLogLevel.STATS)

See the source for more options here.

Building/Compiling Models

Before you can run a model you must first compile it. This is done using the build_model member function which takes the path to the Modelica source file.

model_path = os.path.join('tests', 'models', 'SineCurrent.mo')
session.build_model(model_path)

If the required model is not top level, that is to say it exists within a module or , we can optionally specify the address within Modelica. This is also required if the required model is not the default. For example say model A existed within module M:

model_path = 'FictionalModelFile.mo'
session.build_model(model_path, 'M.A')

Examining Parameters and Options

We can examine all parameters for a given model using the get_parameters method which will return a Python dictionary:

session.get_parameters('SineCurrentModel')

if the parameter is unique to a single model then the model name argument can be dropped. Returning the value for a single parameter is as simple as:

session.get_parameter(<parameter-name>)

For simulation options the analogous methods are get_simulation_options and get_simulation_option respectively for general case, for more specific see below.

Setting Parameters and Options

Set a parameter to a different value using the set_parameter function:

session.set_parameter(<parameter-name>, <new-value>)

Further Configuration

The output file type can be specified:

from pydelica.options import OutputFormat
session.set_output_format(OutputFormat.CSV) # Other options are MAT and PLT

Set the solver:

from pydelica.options import Solver
session.set_solver(Solver.DASSL)    # Other options are EULER and RUNGE_KUTTA

Set the time range:

# Each argument is optional
session.set_time_range(start_time=0, stop_time=10, model_name='SineCurrentModel')

Set tolerance:

# Model name is optional
session.set_tolerance(tolerance=1E-9, model_name='SineCurrentModel')

Set variable filter for outputs:

# Model name is optional
session.set_variable_filter(filter_str='output*', model_name='SineCurrentModel')

Failing Simulation on Lower Assertion Level

By default PyDelica will look for the expression assert | error as an indication of a Modelica assertion failure and then terminate when this is violated. You can override this behaviour using the fail_on_assert_level method of the Session class:

from pydelica import Session

pd_session = Session()
pd_session.fail_on_assert_level('warning')

Possible values ranked by order (highest at the top):

Value Description
'never' Do not throw an exception on Modelica assertion violation
'error' Default. Throw an exception on an assertion of level AssertionLevel.error
'warning' Throw an exception on assertion of level AssertionLevel.warning
'info' Throw an exception on any `assert

Running the Simulation

To run the simulation use the simulate method. If a model name is specified then that model is run, else this is the first model in the model list. At the simulation step parameter values are written to the XML file read by the binary before the model binary is executed.

# Model name is optional, verbosity is also optional and overwrites that of the session
session.simulate(model_name='SineCurrentModel', verbosity=OMLogLevel.DEBUG)

Retrieving Results

To view the results use the get_solutions method which will return a python dictionary containing the solutions for all models after a model run:

solutions = session.get_solutions()

The variables for each model are stored as a Pandas dataframe.

Troubleshooting

PyDelica cannot find OMC

PyDelica relies on either locating OMC on UNIX using the which command, or in the case of Windows using the OPENMODELICAHOME environment variable. Ensure at least one of these is available after installating OpenModelica.

Docker

PyDelica is available on Docker, images can be found here

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for pydelica, version 0.1.14
Filename, size File type Python version Upload date Hashes
Filename, size pydelica-0.1.14-py3-none-any.whl (16.8 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size pydelica-0.1.14.tar.gz (15.7 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page