[evlatests] Initialization problem

[Gustaaf Van Moorsel]

As suggested by Walter and Bryan, I did the fast-switching test again,
this time in a) fast switching mode and b) regular mode 'imitating'
fast switching.  In either case the time on source and calibrator was
30 sec.

_The result is that the flagging problem occurs in both cases_,  Below I
give the result of LISTR, opcode 'LIST', for the regular case (no fast
switching).  Note that the last integration in the first scan, on
0217+738, occurs at 07:35:40.  The first two integrations on the next
scan, on 0157+747, occur at 07:35:41, whereas the next integration
occurs 9 seconds later at 07:35:50.  This, and the lower amplitudes
of both integrations at 7:35:41, strongly suggest that these two int-
egrations occur after the antennas had started moving and should have
been flagged.  Exactly the same behavior is seen when doing true fast
switching.

The same problem is seen again in the next scan boundary from 0157+747
back to 0217+738, around 07:36:10.  In fact, it occurs at *every* scan
boundary.

Source=0217+738        : 0000, Stokes=RR  , IF=  1, Chans=   1-   1
Amplitudes, 1000 =   1.000 Jy, averging type = Vector
Baselines      126 226 326 426 526 626 726 826 92610261126122613261426
   0/07:35:38  494 291  20 471 562 469 414 418 403 410 575 402 451 461
   0/07:35:39  487 289  13 483 579 471 421 415 407 409 555 402 446 459
   0/07:35:39  486 279  13 473 566 475 421 413 409 404 563 397 453 463
   0/07:35:39  489 287  23 480 579 478 436 413 410 409 559 403 453 470
   0/07:35:40  492 289  18 480 573 474 435 417 403 410 544 397 458 464
Source=0157+747        : 0000, Stokes=RR  , IF=  1, Chans=   1-   1
Baselines      126 226 326 426 526 626 726 826 92610261126122613261426
   0/07:35:41   96  24  15  32  56  35 120  26  96  23 108 115 104  23
   0/07:35:41    7   3   6   6   5   6  14   4  19  10   7   4   9   2
   0/07:35:50   51          60  93  57  51  46  41          40  59  49
   0/07:35:51   59       2  60  68  48  50  54  37  58  74  55  46  59
   0/07:35:51   59  37  23  59  59  55  34  46  57  48  65  49  59  58
   0/07:35:51   67  32   6  62  73  56  48  39  60  54  79  48  55  66
   0/07:35:52   62  34   5  65  53  55  51  51  51  55  75  49  57  53
   0/07:36:09   53  35   7  58  80  48  52  46  59  45  61  44  62  61
   0/07:36:09   60  28   7  55  76  53  62  49  44  52  67  55  48  37
   0/07:36:09   60  28   5  60  61  54  47  42  45  48  57  36  56  50
   0/07:36:10   58  33   8  43  66  60  54  59  60  52  85  46  58  40
Source=0217+738        : 0000, Stokes=RR  , IF=  1, Chans=   1-   1
Amplitudes, 1000 =   0.100 Jy, averging type = Vector
Baselines      126 226 326 426 526 626 726 826 92610261126122613261426
   0/07:36:11  227  97  37  62  27 104  53 121 135  43 195 177 253  92
   0/07:36:11   36 125     133                          68      78 147
Amplitudes, 1000 =   1.000 Jy, averging type = Vector
   0/07:36:20  480         476     477 433 411 404 405     400 451 458
   0/07:36:21  499      14 478 568 468 430 414 407 411 540 401 444 454


Gustaaf


Walter Brisken wrote:
> Quick question for everyone on the list.  Does anyone have any examples of 
> the initialization problem occuring in non-fast-switching cases?  My 
> immediate guess is that fast switching doesn't get a flag after source 
> change because of the way it shares infrastructure with mosaicing that 
> doesn't want that flag.
>
> For fast switching scans, I can perhaps understand missing the source 
> change flag, but I am really surprised that an off-source flag doesn't 
> catch these since a flag is generated if the antenna is off source for any 
> part of an integration.
>