Here are steps to verify Cybershake outputs:

1- Verifications from a programming standpoint:

1. High-frequency: 
   
   For ASCI mode: The number of acceleration files should be 3 times the number of stations. The size of the files for all stations should be equal because the dt value is the same for all stations.
   for Binary mode: Checks whether the hf.bin exists. The number of acceleration files should be 3 times the number of stations. Checks whether all stations have Vs values.
   
   
2. Low-frequency: Checks whether each Rlog is tagged as finished.
   
   
3. Winbin_aio: The number of velocity files should be 3 times the number of stations. The size of the files for all stations should be equal because the dt value is the same for all stations.
   
   
4. Merge-ts: checks whether xyts file (which is used for creating animation) exists.
   
   
5. Merging HF and LF: 
   
   For ASCI mode: The number of acceleration files should be 3 times the number of stations. The size of the files for all stations should be equal because the dt value is the same for all stations.
   For Binary mode: Checks if the hf.bin exists. The number of acceleration and velocity files should be 3 times the number of stations. Checks whether all stations have Vs values.
   
   
6. simulation Config:  check whether the ingredients of the config files are correct for the given version, namely:
   station file path, 1D velocity model, dt for LF, dt for HF.
   
   
   note: dt for LF and HF can be different, but the dt for LF should be equal to dt and consistent with verification for the input (see input verification page dt < 0.495 * hh / V_max from Graves 1996 )
    

7. IM calc: Check is whether .info file exists. If IM calc is done successfully, there will be .csv (which contains the calculated IM) and a .info file.

   
   

   Items to be added for CS 18.9 version:

   
   

8. Run "install simulations" based on the selection file used to generate SRF and VM rather than the VM list created based on existing VMs in the source folder.
   
   
   
9. As part of the simulation automated workflow, calculate the ratio of simulation over empirical for the first realization of each given fault (once the simulation is finished), then compare the mean bias in all location with a user_specifiedl_threshold. If the mean bias is smaller than a threshold, keep continuing with the rest of realizations for that fault, otherwise, stop the queue for that fault and inform the analyst.

2- Verifications from research standpoint:

1. Intensity measures on a map view: plot IM (e.g., PGA, PGV, SA(0.5, 1.0,3.0,5.0), Ds595, Ds575, CAV, AI) on the map view

   - Simulation IM 

   - Empirical IM 

   - The ratio of Simulation / Empirical 

   - IM plot versus Rrup overlayed with the corresponding empirical