Still working on my Cabb data, but now I've got a problem with uvglue.
First, I used uvaver to split a dataset into 2049 different datasets, each containing exactly one channel with 1 MHz of bandwidht. After some work, I decided to put it back together and found that uvglue can do the job.
Then, when I imaged the field, I saw that for some strange reason there were 4 point sources instead of 2. To find out why, I imaged different chunks and found that the two "new" sources came into existence after I imaged more than around 1000 channels. To find out exactly what is going on I imaged each channel separately and found that around channel 1000 the point sources seem to do a jump. My fear was that there is something wrong with the data.
Then I tried to find out if uvglue has a problem.
I imaged the dataset from the single channel form, when it was not put back together with uvglue, and in these images, there seems to be no jump.
After watching carefully, I noticed that the output of uvglue was something like:
processing test_1
processing test_2
.
.
.
processing test_998
processing test_999
processing test_100
processing test_100
processing test_100
processing test_100
processing test_100
processing test_100
processing test_100
processing test_100
processing test_100
processing test_101
First I thought that it may be just an error in printing the number, but since after channel 1000 the sources jump (for whatever reason), I guess that uvglue really uses _100 again instead of _1000? Could this be a possibility?
Thanks,
Arpad