* use pyfftw if environment variable YAM_FFTW is set and different from "no" or "false" (before pyfftw was used if environment variable was set, regardless of its value)
0.7.0
* consequently use float16 in the code, switching to other dtypes may not be completely supported at this point and the possibility to change the dtype may be removed later * use --njobs 0 for sequential processing (before it was 1) and --njobs 1 for processing of each element in a new process * index for cross-correlation is nested one level deeper now to not have too many datsets in one group for cross-correlations of 10min data pieces * add new plottype: velocity change versus time of different station/component combinations together with joint estimate * allow to set compression and other options when creating a dataset * allow to set libver for new HDF5 files (default is 'earliest' for best backwards compatibility)
0.6.0
* yam can now be used on Windows, tests are passing, but multiproccesing is not yet working (see issue 4)
0.5.0
* major refactoring of stretch module, some config options for stretching are renamed, the resulting dictionaries changed in a backwards incompatible way, please (a) use yam version < 0.5 or (b) adapt config file and run yam stretch again for any existing project
0.4.3
* add demean option (before filtering) * add plugin option for timenorm * refactor timenorm function, some kwargs are deprecated * timenorm and spectral_whitening are now working with Trace objects (instead of numpy arrays) * add dpi option for most plots to set resolution of saved images * more verbose error message if coordinates of station cannot be found * fix: guarantee same length of traces which are correlated (necessary if tolerance_shift is set)
0.4.2
* adapt to GFZ FDSN webservices and allow only 1 thread when downloading example data