%matplotlib inline
import os
import fluiddyn as fld
import fluidlab

Tutorial: working in the laboratory (user perspective)

FluidDyn uses the object-oriented programming concepts. It deals with objects, which is a very natural way to represent and drive experiments since experiments consist in objects interacting with each other.

Regarding the laboratory, each physical object (a pump, a traverse, a probe, an acquisition board, a tank filled with a stratified fluid...) is represented and controlled by an instance of a class. The experimental results can also be represented by other types of objects.

Example of a conductivity probe attached to a moving traverse

Let’s consider a real-life example, how to use a conductivity probe attached to a moving traverse. FluidDyn provides the class fluidlab.objects.probes.MovingConductivityProbe which can be used like this:

# import the class representing the moving conductivity probe
from fluidlab.objects.probes import MovingConductivityProbe

# create an instance of this class
probe = MovingConductivityProbe()

try:
    # set a parameter, the sample rate (in Hz)
    probe.set_sample_rate(2000)

    # just move the probe (in mm and mm/s)
    probe.move(deltaz=-100, speed=50)

    # just measure without moving (in s)
    measurements1 = probe.measure(duration=5)

    # move and measure (in mm and mm/s)
    measurements2 = probe.move_measure(deltaz=100, speed=100)
except AttributeError:
    pass

Of course this is a very simple example and there are more options to create the object probe and for the functions. Look at the documentation, i.e. in this case here: fluidlab.objects.probes.MovingConductivityProbe.

Save and load an object

For some classes of FluidDyn, the objects can be saved in a file and loaded afterwards. This is a very useful feature! To see how it works, we can consider the example of a tank filled with a stratified fluid, which is represented by the class fluidlab.objects.tanks.StratifiedTank. Let’s first see how we create a tank:

from fluidlab.objects.tanks import StratifiedTank

# create a tank with a linear stratification (see the doc of the class)
tank = StratifiedTank(
    H=550, S=100,
    z=[0, 500], rho=[1.1, 1])

The numerical object tank contains some information and can be use to do useful. We can for example fill the physical tank with the wanted profile (which makes use of some pumps also controlled by FluidDyn, see the class fluidlab.objects.pumps.MasterFlexPumps):

tank.fill()
Warning: can not fill without pumps. It will only perform a test of
the filling. To really fill the tank, set argument pumps to True or to
an instance of class MasterFlexPumps.

flowrate_tot: 840.00 ml/min
vol_to_pump: 192.00 ml
time for the filling:  0.23 min
volume pumped / volume to pump = 0.9479
The filling is finished.
../_images/tuto_lab_user_11_1.png ../_images/tuto_lab_user_11_2.png

The numerical object tank can be saved in a file tank.h5 with its function save (the documentation explains how to control where the file is saved):

if os.path.exists('/tmp/tank.h5'):
    os.remove('/tmp/tank.h5')
tank.save('/tmp')

If we come back some days later and we want to use again this particular instance of fluidlab.objects.tanks.StratifiedTank. Let’s assume that the file is in a directory /tmp/exp0. If we really know that this file contains the information for loading an object of fluidlab.objects.tanks.StratifiedTank, we can obtain the numerical representation of the tank by doing:

del(tank)
tank = StratifiedTank(str_path='/tmp')

But most of the case, it is easier and safer to use the function fluiddyn.util.util.create_object_from_file() like this:

path_to_tank_h5 = '/tmp/tank.h5'
tank = fld.create_object_from_file(path_to_tank_h5)

The function create_object_from_file() gets the correct class from information written in the file, calls the constructor of this class and return the object.

tank.profile.plot()
../_images/tuto_lab_user_20_0.png

Representation of an experiment

Physically, an experiment consists in interacting objects. The experimentalist wants to control the actions of the objects with a good control in space and time and in a reproducible way. The results are then measurements produced by the measuring objects. Usually, after the experiment has been set up, it is repeated a number of times in order to vary some parameters.

A experimental set-up is represented in FluidDyn by a class derived from the class fluidlab.exp.base.Experiment. The experiment class has attributes that represent the physical objects interacting in the experimental set-up.

Each realisation of the experimental set-up (with a particular set of parameters) is represented by an instance of the experiment class. Each experiment (each realisation) is associated with a particular directory.

In order to create a class, do for example:

from fluidlab.exp.taylorcouette.linearprofile import ILSTaylorCouetteExp

exp = ILSTaylorCouetteExp(
    rho_max=1.1, N0=1., prop_homog=0.1,
    Omega1=0.4, Omega2=0, R1=150, R2=200,
    description="""This experiment is for the tutorial.""")

[attr for attr in dir(exp) if not attr.startswith('_')]
['board',
 'description',
 'first_creation',
 'name_dir',
 'params',
 'path_save',
 'profiles',
 'save_script',
 'tank',
 'time_start']
print(exp.description)
Experiment in a Taylor-Couette.

This tank is 520 mm high. The radius of the outer cylinder is
approximately   200 mm.


Initially linear stratification (ILS)...

This experiment is for the tutorial.
print(exp.path_save)
/home/users/augier3pi/Exp_data/TaylorCouette/ILS/Exp_Omega1=0.40_N0=1.00_2015-06-24_17-32-49

When this experiment has been created, the description files have been automatically saved in the “right” place. This “right” place being defined in the class of the experiment. Then we can easily reload the experiment.

path_save = exp.path_save
del(exp)
exp = fluidlab.load_exp(path_save[-20:-5])
print(exp.path_save)
print('R2 = {}'.format(exp.params['R2']))
/home/users/augier3pi/Exp_data/TaylorCouette/ILS/Exp_Omega1=0.40_N0=1.00_2015-06-24_17-14-39
R2 = 200

Note that I deliberately only use the string path_save[-20:-5] to show that fld.load_exp() is sufficiently clever to find out an experiment that corresponds to this string. Be careful to provide a sufficiently long string to be sure to load the wanted experiment.