GitLab

Updated SH viscoelastic modelling to the new grid.

Now it is possible to specify WOLFE_C1_SL and WOLFE_C2_SL in the JSON input file.
Documentation is updated as well.
Smaller output improvements.

Fixed a bug which prevented to stop at max FC when FC_END was not a multiple of FC_INCR.
Additionally L-BFGS will try step length 1 first every time a new frequency stage is choosen.

Added the option WRITE_FILTERED_DATA to write time filtered measured data to disk.
If WRITE_FILTERED_DATA is not set in the JSON file, it will be set to 0.

Added debugging output for VP and renamed some variables to be consistent within the code.

Removed all checkpointing things in the code.

Removed the option HESSIAN from the code.

Workflow file now includes a header.

Update: Merge test

-Remove src_relicts
-Clean up of the Makefile in src/

adjustment of acoustic toy-example, scaling of the elastic explosive source and acoustic free surface condition.

For the exchange of the wavefields a wrapping layer of FDORDER/2 is needed.
Nevertheless a wrapping layer of FDORDER/2 +1 was implemented because of undetected errors in the exchange functions.

Added taper for inverted source time function (use switch TAPER_STF, see taper.c for definition) and individual time windowing (use switch TW_IND, then three columns are expected in PICK_FILE for each trace with picked time - twl plus - twl minus, in this case TWLENGTH_PLUS and TWLENGTH_MINUS are ignored). Time windowing is now also applied during the inversion of the source time function.

char wm[1] should be wm[2], due to reusage of wm later on.