- Nadine
My main motivator to initiate this project was to be able to do research on formal measures of emergence & complexity that
I found that in order to achieve this, quite a few things would need to be tackled, e. g., making measures accessible in the first place (as some of them have been implemented in programming languages that are published under profit software licenses), having software that complies to FAIR principles, or developing a comparative testing environment which enables the testing of different measures under the same constraints.
My thought was: If achieving my goals will require substantial long-term efforts, then anyone else pursuing a similar endeavour would probably need to put the same efforts into it – why not spare potential future redundancy, and write a software library in Python as part of an open-source project to join forces with potential contributors – aiming for FAIR software in the best possible way, implementing an exhaustive set of measures, as well as providing good guidance on how to conduct comparative applications of the measures?
I wanted to have guidance in developing a stable library that will help in applying measures of emergence and complexity as easily as possible, ensure that outcomes are replicable, and allow to draw informed conclusions about comparisons among measures. I considered code reviews & guidance on general software engineering principles very helpful in that respect. I also wanted to know more about the building blocks of collaborative open-source projects, and be around like-minded people with whom I share the same attitudes in the quest for solid & incremental science, cooperation, and inclusivity.
During my participation in OLS-4, I wanted to come up with a library that applies a set of recent measures of emergence and complexity to either simulated or empirical data – using possibly only one line of code, making comparisons among them easy,
While I couldn’t tackle yet the development of educational resources of my library, I managed to finalize software that allows applying one or a set of different measures to one or multiple datasets using only one line of code – an achievement I am extremely happy about! For now, one measure of emergence is implemented and can be applied. As the current code is scalable, adding further measures will be very easy. Moreover, the output of this one line of code delivers a variable which stores both results and parameters so as to be able to easily retrieve cases of interest and visualize them.
During OLS-4, I also started to finally implement some code testing and think about possible “formats” of usage (e. g., shall the library be executed via a command line interface, Docker image, or do I want the user to work directly in Python?). The feedback I got from my mentors and code reviewers in this regard was particularly helpful.
OLS-4 has also got me started on open science project essentials like code of conduct and contributor guidance specifications.
What I’ve done:
import complex_py as cp
# define what data to use/generate, and what measures of emergence
# to apply to those data via the following five variables:
# model_function: contains functions to generate/load your data
# model_variables: contains parameters for generating/loading your data
# emergence_functions: contains functions to calculate measures of emergence
# measure_variables: contains parameters for calculating emergence
# parameters: contains all parameters (data- and measure-related ones)
# compute measures of emergence for all parameter combination:
emergence_df = cp.compute_emergence(model_functions, model_variables,
emergence_functions, measure_variables, parameters)
Still to do:
At the point where OLS-4 started, I was still struggling to make a measure of emergence work in Python via ‘oct2py()’ - a Python library which allows the calling of Matlab functions in Python. The battle continued for about 3 months after which – with a few hacky interventions – I was able to use it in Python. Clearly, a long-term solution will involve porting those Matlab functions into Python (something everyone my mentor and other code reviewers were heavily stressing)… but as this will be quite some task, it’s good to have this work-around in the meantime to progress on other parts of the project.
The support of the broader OLS community, OLS Founders, my mentor, other OLS Project Leads, as well as other people I could ask for help and guidance have tremendously helped me in progressing in both my project as well as my involvement in the open science community and advocacy for open science practices. I particularly valued the 1-to-1 conversations I had with my mentor as well as with multiple OLS experts who were willing to review my code. I also enjoyed the inputs during various cohort calls where I learned about important aspects of Open Leadership.
… implement another measure! I will also tackle the porting of Matlab functions into Python, and start working on educational resources for the library – both for users and potential future contributors.
My main tasks for the next year will be to accomplish the tasks described above and release a first version for others to use as well as to contribute to. This will also involve spreading the word about the library and doing some advertisements. Overall, I hope that this project will contribute to making research on emergence & complexity more open, reproducible, and inclusive in the long run!
I am thankful to