-
Notifications
You must be signed in to change notification settings - Fork 10
Migrating to kima v3
Going from kima version 2 to version 3, some backwards-incompatible changes were introduced.
Here are a few tips on how to convert your previous code.
-
The start of the
kima_setup.cpp
files is simpler
While before we had to#include
a bunch of files, now it's enough to do#include "kima.h"
once. -
Setting priors is less error-prone
In version 3, the syntax to define priors changed completely. All priors are now set inside theRVmodel
constructor. Instead ofCprior = new Uniform(-10, 10);
we now use
Cprior = make_prior<Uniform>(-10, 10);
Even if this is slightly more verbose, it's a much easier solution to default priors. This new syntax allows the default prior for the systemic velocity or the slope of the linear trend to depend on the RV data without you having to write anything.
-
No more
default_priors.h
There are still default priors, but kima will use them if you don't do anything in theRVmodel
constructor -
Loading data is simpler
Before we had to doData::get_instance().load(datafile, "ms", 0);
while now this is enough
load(datafile, "ms", 0);
The same in multi instrument mode:
load_multi(datafiles, "kms", 2)
-
Some variables are already defined
Related to the point above, in v3 you can simply writedatafile = "filename"
or
datafiles = {"filename1", "filename2"}
When including correlations with activity indicators, it's enough to set
indicators = {"fwhm", "bis"}
This documentation was created with ❤️ by @j-faria and @jdavidrcamacho, at IA.
- What is kima
- Installation
- Getting started
- Running jobs
- Examples
- Analysis of results
- Changing the priors
- Changing OPTIONS
- Input data
- Output files
- Roadmap
- Contribute
- Troubleshooting
Additional material
- Are the defaults ok?
- Migrating to kima v3
- Transiting planet
- Multiple instruments
- New prior distributions
- Regression network
API