Question

python newbie here. I’m writing the code to control an experiment that has multiple variables. The variables are defined in a csv file something like:

Temp, Voltage, Pressure
23, 4.8, 19
23;28, 4.8, 19;23

I read the file and create a dictionary, that part works ok. for example the first line will be:

{'Temp': ['23'], 'Voltage': ['4.8'], 'Pressure': ['19']}

and that will control a single experiment.
The second line will be:

{'Temp': ['30', '28'], 'Voltage': ['4.8'], 'Pressure': ['19', '23']}

What I would like is a way to generate all the possible dictionaries, in sequence, from that line. For example the second line will be expanded to four experiments:

{'Temp': ['30'], 'Voltage': ['4.8'], 'Pressure': ['19']}
{'Temp': ['30'], 'Voltage': ['4.8'], 'Pressure': ['23']}
{'Temp': ['28'], 'Voltage': ['4.8'], 'Pressure': ['19']}
{'Temp': ['28'], 'Voltage': ['4.8'], 'Pressure': ['23']}

The order is important (the last column is the inner loop). I have a list with the keys in the order I want to iterate. For example:

('Temp', 'Voltage', 'Pressure')

in this case.

This is an oversimplification and there is a big number of variables (columns) in the system.

I’m thinking about a parallel dictionary with counters to keep track of the possible cases and current case for each variable, but maybe I’m missing a more elegant way to express this in python.

Regards,

–ga

Answer 1

Yup, you want itertools.product(). You just need to extract the lists from the dict & re-wrap the results with your dictionary structure.

from itertools import product
def _combinations_for_record(d):
    for tmp, vlt, prs in product(d['Temp'], d['Voltage'], d['Pressure']):
        yield {"Temp": [tmp], "Voltage": [vlt], "Pressure": [prs]}