Contributions are welcome, and they are greatly appreciated! Every little bit helps, and credit will always be given.

Bug reports/Feature Requests/Feedback/Questions

It is incredibly helpful to us when users report bugs, unexpected behaviour, or request features. You can do the following:

When doing any of these, please try to be as succinct, but detailed, as possible, and use a “Minimum Working Example” whenever applicable.

Documentation improvements

21cmFAST could always use more documentation, whether as part of the official 21cmFAST docs, in docstrings, or even on the web in blog posts, articles, and such. If you do the latter, take the time to let us know about it!

High-Level Steps for Development

This is an abbreviated guide to getting started with development of 21cmFAST, focusing on the discrete high-level steps to take. See our notes for developers for more details about how to get around the 21cmFAST codebase and other technical details.

There are two avenues for you to develop 21cmFAST. If you plan on making significant changes, and working with 21cmFAST for a long period of time, please consider becoming a member of the 21cmFAST GitHub organisation (by emailing any of the owners or admins). You may develop as a member or as a non-member.

The difference between members and non-members only applies to the first step of the development process.

Note that it is highly recommended to work in an isolated python environment with all requirements installed from environment_dev.txt. This will also ensure that pre-commit hooks will run that enforce the black coding style. If you do not install these requirements, you must manually run black before committing your changes, otherwise your changes will likely fail continuous integration.

As a member:

  1. Clone the repo:

    git clone

As a non-member:

  1. First fork ``21cmFAST <>``_ (look for the “Fork” button), then clone the fork locally:

    git clone

The following steps are the same for both members and non-members:

  1. Install a fresh new isolated environment:

    conda create -n 21cmfast python=3
    conda activate 21cmfast
  2. Install the development requirements for the project:

    conda env update -f environment_dev.yml
  3. Install 21cmFAST. See Installation for more details.:

    pip install -e .
  1. Install pre-commit hooks:

    pre-commit install
  2. Create a branch for local development:

    git checkout -b name-of-your-bugfix-or-feature

    Now you can make your changes locally. Note: as a member, you _must_ do step 5. If you make changes on master, you will _not_ be able to push them.

  3. When you’re done making changes, run pytest to check that your changes didn’t break things. You can run a single test or subset of tests as well (see pytest docs):

  4. Commit your changes and push your branch to GitHub:

    git add .
    git commit -m "Your detailed description of your changes."
    git push origin name-of-your-bugfix-or-feature

    Note that if the commit step fails due to a pre-commit hook, most likely the act of running the hook itself has already fixed the error. Try doing the add and commit again (up, up, enter). If it’s still complaining, manually fix the errors and do the same again.

  5. Submit a pull request through the GitHub website.

Pull Request Guidelines

If you need some code review or feedback while you’re developing the code just make the pull request. You can mark the PR as a draft until you are happy for it to be merged.

Developer Documentation

If you are new to developing 21cmFAST, please read the Contributing section first, which outlines the general concepts for contributing to development, and provides a step-by-step walkthrough for getting setup. This page lists some more detailed notes which may be helpful through the development process.

Compiling for debugging

When developing, it is usually a good idea to compile the underlying C code in DEBUG mode. This may allow extra print statements in the C, but also will allow running the C under valgrind or gdb. To do this:

$ DEBUG=True pip install -e .

See Installation for more installation options.

Developing the C Code

In this section we outline how one might go about modifying/extending the C code and ensuring that the extension is compatible with the wrapper provided here. It is critical that you run all tests after modifying _anything_ (and see the section below about running with valgrind). When changing C code, before testing, ensure that the new C code is compiled into your environment by running:

$ rm -rf build
$ pip install .

Note that using a developer install (-e) is not recommended as it stores compiled objects in the working directory which don’t get updated as you change code, and can cause problems later.

There are two main purposes you may want to write some C code:

  1. An external plugin/extension which uses the output data from 21cmFAST.

  2. Modifying the internal C code of 21cmFAST.

21cmFAST currently provides no support for external plugins/extensions. It is entirely possible to write your own C code to do whatever you want with the output data, but we don’t provide any wrapping structure for you to do this, you will need to write your own. Internally, 21cmFAST uses the cffi library to aid the wrapping of the C code into Python. You don’t need to do the same, though we recommend it. If your desired “extension” is something that needs to operate in-between steps of 21cmFAST, we also provide no support for this, but it is possible, so long as the next step in the chain maintains its API. You would be required to re-write the low-level wrapping function _preceding_ your inserted step as well. For instance, if you had written a self-contained piece of code that modified the initial conditions box, adding some physical effect which is not already covered, then you would need to write a low-level wrapper _and_ re-write the initial_conditions function to modify the box before returning it. We provide no easy “plugin” system for doing this currently. If your external code is meant to be inserted _within_ a basic step of 21cmFAST, this is currently not possible. You will instead have to modify the source code itself.

Modifying the C-code of 21cmFAST should be relatively simple. If your changes are entirely internal to a given function, then nothing extra needs to be done. A little more work has to be done if the modifications add/remove input parameters or the output structure. If any of the input structures are modified (i.e. an extra parameter added to it), then the corresponding class in py21cmfast.wrapper must be modified, usually simply to add the new parameter to the _defaults_ dict with a default value. For instance, if a new variable some_param was added to the user_params struct in the ComputeInitialConditions C function, then the UserParams class in the wrapper would be modified, adding some_param=<default_value> to its _default_ dict. If the default value of the parameter is dependent on another parameter, its default value in this dict can be set to None, and you can give it a dynamic definition as a Python @property. For example, the DIM parameter of UserParams is defined as:

def DIM(self):
    if self._some_param is None:
        return self._DIM or 4 * self.HII_DIM

Note the underscore in _DIM here: by default, if a dynamic property is defined for a given parameter, the _default_ value is saved with a prefixed underscore. Here we return either the explicitly set DIM, or 4 by the HII_DIM. In addition, if the new parameter is not settable – if it is completely determined by other parameters – then don’t put it in _defaults_ at all, and just give it a dynamic definition.

If you modify an output struct, which usually house a number of array quantities (often float pointers, but not necessarily), then you’ll again need to modify the corresponding class in the wrapper. In particular, you’ll need to add an entry for that particular array in the _init_arrays method for the class. The entry consists of initialising that array (usually to zeros, but not necessarily), and setting its proper dtype. All arrays should be single-pointers, even for multi-dimensional data. The latter can be handled by initalising the array as a 1D numpy array, but then setting its shape attribute (after creation) to the appropriate n-dimensional shape (see the _init_arrays method for the InitialConditions class for examples of this).

Modifying the global_params struct should be relatively straightforward, and no changes in the Python are necessary. However, you may want to consider adding the new parameter to relevant _filter_params lists for the output struct wrapping classes in the wrapper. These lists control which global parameters affect which output structs, and merely provide for more accurate caching mechanisms.

C Function Standards

The C-level functions are split into two groups – low-level “private” functions, and higher-level “public” or “API” functions. All API-level functions are callable from python (but may also be called from other C functions). All API-level functions are currently prototyped in 21cmFAST.h.

To enable consistency of error-checking in Python (and a reasonable standard for any kind of code), we enforce that any API-level function must return an integer status. Any “return” objects must be modified in-place (i.e. passed as pointers). This enables Python to control the memory access of these variables, and also to receive proper error statuses (see below for how we do exception handling). We also adhere to the convention that “output” variables should be passed to the function as its last argument(s). In the case that _only_ the last argument is meant to be “output”, there exists a simple wrapper _call_c_simple in that will neatly handle the calling of the function in an intuitive pythonic way.

Running with Valgrind

If any changes to the C code are made, it is ideal to run tests under valgrind, and check for memory leaks. To do this, install valgrind (we have tested v3.14+), which is probably available via your package manager. We provide a suppression file for valgrind in the devel/ directory of the main repository.

It is ideal if you install a development-version of python especially for running these tests. To do this, download the version of python you want and then configure/install with:

$ ./configure --prefix=<your-home>/<directory> --without-pymalloc --with-pydebug --with-valgrind
$ make; make install

Construct a virtualenv on top of this installation, and create your environment, and install all requirements.

If you do not wish to run with a modified version of python, you may continue with your usual version, but may get some extra cruft in the output. If running with Python version > 3.6, consider running with environment variable PYTHONMALLOC=malloc (see ).

The general pattern for using valgrind with python is:

$ valgrind --tool=memcheck --track-origins=yes --leak-check=full --suppressions=devel/valgrind-suppress-all-but-c.supp <python script>

One useful command is to run valgrind over the test suite (from the top-level repo directory):

$ valgrind --tool=memcheck --track-origins=yes --leak-check=full --suppressions=devel/valgrind-suppress-all-but-c.supp pytest

While we will attempt to keep the suppression file updated to the best of our knowledge so that only relevant leaks and errors are reported, you will likely have to do a bit of digging to find the relevant parts.

Valgrind will likely run very slowly, and sometimes you will know already which exact tests are those which may have problems, or are relevant to your particular changes. To run these:

$ PYTHONMALLOC=malloc valgrind --tool=memcheck --track-origins=yes --leak-check=full --suppressions=devel/valgrind-suppress-all-but-c.supp pytest -v tests/<test_file>::<test_func> > valgrind.out 2>&1

Note that we also routed the stderr output to a file, which is useful because it can be quite voluminous. There is a python script, devel/ which can be run over the output (valgrind.out in the above command) to filter it down to only have stuff from 21cmfast in it.

Producing Integration Test Data

There are bunch of so-called “integration tests”, which rely on previously-produced data. To produce this data, run python tests/

Furthermore, this data should only be produced with good reason – the idea is to keep it static while the code changes, to have something steady to compare to. If a particular PR fixes a bug which affects a certain tests’ data, then that data should be re-run, in the context of the PR, so it can be explained.

Logging in C

The C code has a header file logging.h. The C code should never contain bare print-statements – everything should be formally logged, so that the different levels can be printed to screen correctly. The levels are defined in logging.h, and include levels such as INFO, WARNING and DEBUG. Each level has a corresponding macro that starts with LOG_. Thus to log run-time information to stdout, you would use LOG_INFO("message");. Note that the message does not require a final newline character.

Exception handling in C

There are various places that things can go wrong in the C code, and they need to be handled gracefully so that Python knows what to do with it (rather than just quitting!). We use the simple cexcept.h header file from to enable a simple form of exception handling. That file itself should not be edited. There is another header – exceptions.h – that defines how we use exceptions throughout 21cmFAST. Any time an error arises that can be understood, the developer should add a Throw <ErrorKind>; line. The ErrorKind can be any of the kinds defined in exceptions.h (eg. GSLError or ValueError). These are just integers.

Any C function that has a header in 21cmFAST.h – i.e. any function that is callable directly from Python – must be globally wrapped in a Try {} Catch(error_code) {} block. See GenerateICs.c for an example. Most of the code should be in the Try block. Anything that does a Throw at any level of the call stack within that Try will trigger a jump to the Catch. The error_code is the integer that was thrown. Typically, one will perhaps want to do some cleanup here, and then finally return the error code.

Python knows about the exit codes it can expect to receive, and will raise Python exceptions accordingly. From the python side, two main kinds of exceptions could be raised, depending on the error code returned from C. The lesser exception is called a ParameterError, and is supposed to indicate an error that happened merely because the parameters that were input to the calculation were just too extreme to handle. In the case of something like an automatic Monte Carlo algorithm that’s iterating over random parameters, one would usually want to just keep going at this point, because perhaps it just wandered too far in parameter space. The other kind of error is a FatalCError, and this is where things went truly wrong, and probably will do for any combination of parameters.

If you add a kind of Exception in the C code (to exceptions.h), then be sure to add a handler for it in the _process_exitcode function in

Maintaining Array State

Part of the challenge of maintaining a nice wrapper around the fast C-code is keeping track of initialized memory, and ensuring that the C structures that require that memory are pointing to the right place. Most of the arrays that are computed in 21cmFAST are initialized in Python (using Numpy), then a pointer to their memory is given to the C wrapper object.

To make matters more complicated, since some of the arrays are really big, it is sometimes necessary to write them to disk to relieve memory pressure, and load them back in as required. That means that any time, a given array in a C-based class may have one of several different “states”:

1. Completely Uninitialized 1. Allocated an initialized in memory 1. Computed (i.e. filled with the values defining that array after computation in C) 1. Stored on disk 1. Stored and in memory.

It’s important to keep track of these states, because when passing the struct to the compute() function of another struct (as input), we go and check if the array exists in memory, and initialize it. Of course, we shouldn’t initialize it with zeros if in fact it has been computed already and is sitting on disk ready to be loaded. Thus, the OutputStruct tries to keep track of these states for every array in the structure, using the _array_state dictionary. Every write/read/compute/purge operation self-consistently modifies the status of the array.

However, one needs to be careful – you can modify the actual state without modifying the _array_state (eg. simply by doing a del object.array). In the future, we may be able to protect this to some extent, but for now we rely on the good intent of the user.

Purging/Loading C-arrays to/from Disk

As of v3.1.0, there are more options for granular I/O, allowing large arrays to be purged from memory when they are unnecessary for further computation. As a developer, you should be aware of the _get_required_input_arrays method on all OutputStruct subclasses. This is available to tell the given class what arrays need to be available at compute time in any of the input structs. For example, if doing PERTURB_ON_HIGH_RES, the PerturbedField requires the hi-res density fields in InitialConditions. This gives indications as to what boxes can be purged to disk (all the low-res boxes in the ICs, for example). Currently, this is only used to check that all boxes are available at compute time, and is not used to actually automatically purge anything. Note however that InitialConditions does have two custom methods that will purge unnecessary arrays before computing perturb fields or ionization fields.


If you add a new quantity to a struct, and it is required input for other structs, you need to add it to the relevant _get_required_input_arrays methods.

Further note that as of v3.1.0, partial structs can be written and read from disk (so you can specify keys=['hires_density'] in the .read() method to just read the hi-res density field into the object.

Branching and Releasing

The aim is to make 21cmFAST’s releases as useful, comprehendible, and automatic as possible. This section lays out explicitly how this works (mostly for the benefit of the admin(s)).


The first thing to mention is that we use strict semantic versioning (since v2.0). Thus the versions are MAJOR.MINOR.PATCH, with MAJOR including API-breaking changes, MINOR including new features, and PATCH fixing bugs or documentation etc. If you depend on hmf, you can set your dependency as 21cmFAST >= X.Y < X+1 and not worry that we’ll break your code with an update.

To mechanically handle versioning within the package, we use two methods that we make to work together automatically. The “true” version of the package is set with setuptools-scm. This stores the version in the git tag. There are many benefits to this – one is that the version is unique for every single change in the code, with commits on top of a release changing the version. This means that versions accessed via py21cmfast.__version__ are unique and track the exact code in the package (useful for reproducing results). To get the current version from command line, simply do python --version in the top-level directory.

To actually bump the version, we use bump2version. The reason for this is that the CHANGELOG requires manual intervention – we need to change the “dev-version” section at the top of the file to the current version. Since this has to be manual, it requires a specific commit to make it happen, which thus requires a PR (since commits can’t be pushed to master). To get all this to happen as smoothly as possible, we have a little bash script bump that should be used to bump the version, which wraps bump2version. What it does is:

  1. Runs bump2version and updates the major, minor or patch part (passed like ./bump minor) in the VERSION file.

  2. Updates the changelog with the new version heading (with the date), and adds a new dev-version heading above that.

  3. Makes a commit with the changes.


Using the bump script is currently necessary, but future versions of bump2version may be able to do this automatically, see

The VERSION file might seem a bit redundant, and it is NOT recognized as the “official” version (that is given by the git tag). Notice we didn’t make a git tag in the above script. That’s because the tag should be made directly on the merge commit into master. We do this using a Github Action (tag-release.yaml) which runs on every push to master, reads the VERSION file, and makes a tag based on that version.


For branching, we use a very similar model to git-flow. That is, we have a master branch which acts as the current truth against which to develop, and production essentially as a deployment branch. I.e., the master branch is where all features are merged (and some non-urgent bugfixes). production is always production-ready, and corresponds to a particular version on PyPI. Features should be branched from master, and merged back to production. Hotfixes can be branched directly from production, and merged back there directly, as well as back into master. Breaking changes must only be merged to master when it has been decided that the next version will be a major version. We do not do any long-term support of releases (so can’t make hotfixes to v2.x when the latest version is 2.(x+1), or make a new minor version in 2.x when the latest version is 3.x). We have set the default branch to dev so that by default, branches are merged there. This is deemed best for other developers (not maintainers/admins) to get involved, so the default thing is usually right.


Why not a more simple workflow like Github flow? The simple answer is it just doesn’t really make sense for a library with semantic versioning. You get into trouble straight away if you want to merge a feature but don’t want to update the version number yet (you want to merge multiple features into a nice release). In practice, this happens quite a lot.


OK then, why not just use production to accrue features and fixes until such time we’re ready to release? The problem here is that if you’ve merged a few features into master, but then realize a patch fix is required, there’s no easy way to release that patch without releasing all the merged features, thus updating the minor version of the code (which may not be desirable). You could then just keep all features in their own branches until you’re ready to release, but this is super annoying, and doesn’t give you the chance to see how they interact.


To make a patch release, follow these steps:

  1. Branch off of production.

  2. Write the fix.

  3. Write a test that would have broken without the fix.

  4. Update the changelog with your changes, under the **Bugfixes** heading.

  5. Commit, push, and create a PR.

  6. Locally, run ./bump patch.

  7. Push.

  8. Get a PR review and ensure CI passes.

  9. Merge the PR

Note that in the background, Github Actions should take care of then tagging production with the new version, deploying that to PyPI, creating a new PR from master back into master, and accepting that PR. If it fails for one of these steps, they can all be done manually.

Note that you don’t have to merge fixes in this way. You can instead just branch off master, but then the fix won’t be included until the next minor version. This is easier (the admins do the adminy work) and useful for non-urgent fixes.

Any other fix/feature should be branched from master. Every PR that does anything noteworthy should have an accompanying edit to the changelog. However, you do not have to update the version in the changelog – that is left up to the admin(s). To make a minor release, they should:

  1. Locally, git checkout release

  2. git merge master

  3. No new features should be merged into master after that branching occurs.

  4. Run ./bump minor

  5. Make sure everything looks right.

  6. git push

  7. Ensure all tests pass and get a CI review.

  8. Merge into production

The above also works for MAJOR versions, however getting them in to master is a little different, in that they should wait for merging until we’re sure that the next version will be a major version.