These were producing warnings, and thus prevented us from turning on
-Werror on CI, which is something we'd like to have. I checked with
Claudia, and she said it was safe to remove these.

Signed-off-by: Rodrigo Tobar <[email protected]>

This was issuing a warning, so let's get rid of it.

Signed-off-by: Rodrigo Tobar <[email protected]>

Also make sure our CI jobs check that it doesn't go out of sync.

Signed-off-by: Rodrigo Tobar <[email protected]>

The latter has been deprecated since numpy 1.6, and effectively removed
in the latest versions of numpy. The "density" argument has equivalent
semantics, and the documentation states that it should be used instead.

Signed-off-by: Rodrigo Tobar <[email protected]>

This flag contains the *space-separated* flags that are needed for
compiling OpenMP programs. This is not the format expected by
add_compile_options though, so we need to turn the string into a list
before passing it on.

Signed-off-by: Rodrigo Tobar <[email protected]>

There were a few things that were not correctly done around OpenMP
checks that are being corrected here, namely:

 * Change DECLARED -> DEFINED, since the former doesn't mean anything
   (so this code has been always wrong).
 * Change OPENMP_FOUND -> OpenMP_FOUND, with the latter being the more
   generic way of checking the result of find_package.
 * Use the correct casing for the rest of OpenMP* variable, which were
   previously uppercased in our code.
 * Add a check for the OpenMP_CXX_VERSION variable in particular, which
   might be set while OpenMP_VERSION might not.

Signed-off-by: Rodrigo Tobar <[email protected]>

This new port now tries to *not* rely too much on GitHub Actions
conventions (which are not that many, anyway), and also splits the big
test.sh script we used with Travis into smaller scripts in a new .ci
directory that can be invoked separately, therefore allowing us to model
each individual test as a different step on the GitHub Actions workflow.

Signed-off-by: Rodrigo Tobar <[email protected]>

Signed-off-by: Rodrigo Tobar <[email protected]>

The new file had been added to the tree, but wasn't included anywhere in
the documentation, duh!

Signed-off-by: Rodrigo Tobar <[email protected]>

The implicit build.image is deprecated, so let's use the new build.os
key explicitly instead.

Signed-off-by: Rodrigo Tobar <[email protected]>

Signed-off-by: Rodrigo Tobar <[email protected]>

Signed-off-by: Rodrigo Tobar <[email protected]>

We used to depend on >= 1.54, but in a809629 we accidentally raised the
minimum required Boost version to 1.68 (boost::optional::has_value() was
introduced in that version to align with std::optional). This version of
Boost is 5 years old though, so instead of modifying our code to cope
with older versions, we're probably better off declaring a higher
minimum version.

This was the original cause of #13, and thus it should prevent this
issue from happening again in the near future.

Signed-off-by: Rodrigo Tobar <[email protected]>

Signed-off-by: Rodrigo Tobar <[email protected]>

Signed-off-by: Rodrigo Tobar <[email protected]>

The option defaults to false, and indicates whether the lognormal
distribution used to draw lambda values should use a fix value of 0.03,
or a dynamic value based on the halo spin distribution following Kim et
al. (2015), as its m parameter (the natural logarithm of the
distribution's mean).

The Lagos et al. (2023) paper used this dependency, and therefore we're
also updating the sample configuration file that reflects that paper to
ensure reproducibility.

Signed-off-by: Rodrigo Tobar <[email protected]>

Signed-off-by: Rodrigo Tobar <[email protected]>

…ll for type 2

…ll for type 2

…current properties or at infall for satellites

…ent properties or at infall for satellites. Massive transients: read catalogue with halos affected by this

Signed-off-by: Rodrigo Tobar <[email protected]>

- separated the function to redefine subhalo properties into centrals
and satellites. This is done because the one for centrals needs to be
called before the function that defines ages, while the satellite one
needs to be called afterwards.

- Fixed the calculation for satellite subhalos to use the infall
redshift rather than the current redshift (given that it was using all
the properties at infall).

This leads to many changes to make sure one can run the previous model
without the fixing of mass swapping events (to help with
reproduceability).

Another change is to make sure the angular momentum vector L of subhalos
is redefine in the cases where lambda_random is used.

I also changed added a bool input to evolve_galaxy to indicate whether
we want to apply the fix for the mass swapping cases.