Reports until 17:54, Monday 18 November 2013
H1 SUS
arnaud.pele@LIGO.ORG - posted 17:54, Monday 18 November 2013 (8613)
matlab tfs

[Jeff K. Arnaud P.]

The TF of ITMX ran over the week end from matlab 2012b, had the same issue that I encountered on friday morning. The log file saved gives the following error message :

[Warning: Channel H1:SUS-ITMX_L1_WIT_Y_DQ incorrect time 1068669185 (expected 1068669184).]
[Warning: Error in query for H1:SUS-ITMX_L1_WIT_Y_DQ: unknown error]
[Warning: Response channel H1:SUS-ITMX_L1_WIT_Y_DQ has sampling frequency higher than 256 Hz, decimating to the lower rate ...]
[> In get_sch_TF_MO_v9DQ_sma at 384
  In collect_schroeder_TF_v9DQ_sma at 479
  In Matlab_TFs at 499
  In LHO_Matlab_TFs at 141]
Error using decimate>validateinput (line 158)
The input signal X must be a double-precision vector.

So we investigated and fixed the problem which was basically :

Since we have higher sampling rates for the new witness channels recently implemented, (L1_WIT_$DOF_DQ / L2_WIT_$DOF_DQ), the function "get_sch_TF_M0_v9DQ_sma.m" was trying to downsample those channels to 256Hz right after getting the data out of the frame. If the function get_data failed (randomly the case), it would try to decimate an empty vector, and was returning the above error and quit the script. 

We modified get_sch_TF_M0_v9DQ_sma in order to make sure the data is beeing recovered in a while loop before decimating the signal.

I will run an other set of TF on ITMX tonight and commit the changes tomorrow if it succeeds