Change Record

Forwards:

This table notify all changes which have been performed on the AEDFv2 since 2004-04-15. Note that these changes impact on the codes which are automaticaly generated from the AEDF description document as they depend directly from the table presented in this document.

Item	Parent EDF table	Reason	Date of change	Author
NUM_CORRBIN	CONFIG_DESCRIPTION	item removed	Wed Jun 30 08:48:13 CEST 2004	fviallef
SWITCH_CYCLE_ARRAY	CONFIG_DESCRIPTION	size of vector modified	Wed Jun 30 08:48:13 CEST 2004	fviallef
SWITCH_CYCLE_LIST	CONFIG_DESCRIPTION	size of vector modified	Wed Jun 30 08:48:13 CEST 2004	fviallef
DATA_DESCRIPTION_ARRAY	CONFIG_DESCRIPTION	size of vector modified	Wed Jun 30 08:48:13 CEST 2004	fviallef
EXPOSURE	MAIN	size of vector modified	Wed Jun 30 08:48:13 CEST 2004	fviallef
TIME_CENTROID	MAIN	size of vector modified	Wed Jun 30 08:48:13 CEST 2004	fviallef
OFFSET_ARRAY	SWITCH_CYCLE	size of vector modified	Wed Jun 30 08:48:13 CEST 2004	fviallef
SOURCE_FLUX	SOURCE_PARAMETER	matrix elem. inverted	Wed Jun 30 08:48:13 CEST 2004	fviallef
SOURCE_PA	SOURCE_PARAMETER	size of vector modified	Wed Jun 30 08:48:13 CEST 2004	fviallef
FLOAT_DATA	TOTAL_POWER_MONITORING	size of matrix corrected	Wed Jun 30 08:48:13 CEST 2004	fviallef
STEP_DURATION	SWITCH_CYCLE	item added	Sun Jul 4 23:20:54 CEST 2004	fviallef
SUBSCAN_NUMBER	SUBSCAN_NUMBER	item renamed	Mon Aug 9 19:49:48 EST 2004	fviallef
SUBSCAN_INTENT	SUBSCAN_INTENT	item renamed	Mon Aug 9 19:49:48 EST 2004	fviallef
NUM_SUBSCAN	NUM_SUBSCAN	item renamed	Mon Aug 9 19:49:48 EST 2004	fviallef
TIME_EXTRA_PRECISION	MAIN, TOTAL_POWER_MONITORING and SOURCE_PARAMETER	item removed	Fri Oct 15	fviallef

Discussion

EXPOSURE
Although the time centroid is available in binary format per baseline for each baseband in an attachment, the EXPOSURE item in the MAIN table retains as a non-key attribute the values which are on an antenna basis. This provides a mean to have quickly a rough estimate of the exposure for each baselines if approximated by the mean value for the pairs of antennas for the baselines. For accurate values these must be extracted from the attachment.
Note that having this EXPOSURE as an item means that it is possible to use it in queries.

Issues: It has been suggested (ref correlator ICD) that the data cell should not contain useless auto and cross-correlations. This suggestion, if retained, introduces enormous complexity in the method to get these correlations from the data cell.
FrancoisViallefond: As it is anticipated that there should be only a very small fraction of useless correlations, these could be simply written as blanked or NaN values to keep the shape of the matrix in the data cell simple. This is the initial suggestion that I made and agreed with RobertLucas. However it seems that there is not a a consensus on this!
FrancoisViallefond: An alternate proposition, half-way between dropping all the useless data and keeping them all as blanked or NaN would be to drop only the useless data which involve antennas during all the integration. These antennas could be identified by using this EXPOSURE item, setting a value of 0 in these cases. The implication with this is that, to detrmine the size of the data cell require not only to know the CONFIG_DESCRIPTION_ID but also this EXPOSURE. RobertLucas does not support this idea as it is still introducing some complexity.

Questions:

FrancoisViallefond:Is it really useful to have the exposure values for each baseband? A mean value seems to me sufficient keeping in mind that for accurate values they are always available in the attachment.

TIME_CENTROID
Although the time centroid is available in binary form per baseline for each baseband in an attachment, the TIME_CENTROID item in the MAIN table retains as a non-key attribute the values which are on an antenna basis. This provides a mean to have quickly a rough estimate of the time centroid on a baseline basis.
Note that the UVW item has no baseband dependency implying that it uses a mean value for TIME_CENTROID, this being adequate for its purpose as an item. Questions:

FrancoisViallefond: Is it really useful to have the time centroid values for each baseband? A mean value seems to me sufficient keeping in mind that for accurate values they are always available in the attachment. Note that the UVW item has no baseband dependency implying that it uses a mean value which is adequate.
FrancoisViallefond: Would it be not more convenient to replace this TIME_CENTROID item by a TIME_OFFSET. This offset would be relative to the TIME_EXTRA_PREC which is the requested time to perform the integration (or sub-integration).

STEP_DURATION
This item has been added allowing to provide the durations for all the steps in a switch-cycle.
Question: Does the hardware allow to have different durations for the different steps of the cycle.

Comments

NUM_CORRBIN
This item has been removed to avoid a functional dependency. The number of bins is obtained via NUM_STEP in the SWITCH_CYCLE table which is indexed in the SWITCH_CYCLE_ARRAY collection.
SWITCH_CYCLE_ARRAY
The dependency with the antennas has been removed because all the antennas must share a common switch cycle definition. If this condition is not satisfied then the groups of antennas with different switch cycles must be in different subarrays.
Note that sharing this common switch cycle definition does not mean that all the involved antennas are perfectly synchronized for the switch phase. The sequence of steps for a cycle in a tuple in the SWITCH_CYCLE table defines the order of the steps along the bin axis in the data cell.
Note that SWITCH_CYCLE_ARRAY is renamed SWITCH_CYCLE_LIST because a SWITCH_CYCLE_ID value may appear more than once in the collection.
DATA_DESCRIPTION_ARRAY
The size of this item is not the sum of the products N_bin*N_switem but simply the sum of Nsw. The reason is that in frequency-switch observing mode the step in the different steps in the switching cycle share the properties of a common spectral window, only a frequency offset being applied. This frequency offset is set the the FREQ_OFFSET table at the coordinate ANTENNA_ID, SPECTRAL_WINDOW_ID as referenced in the collection ANTENNA_ARRAY and indirectly in DATA_DESCRIPTION_ARRAY.
OFFSET_ARRAY
The size is (2,N_bin) and not (N_bin)!
SOURCE_FLUX<BR> The order in the matrix is (N_sto,N_fp) and not (N_fp,N_sto)!
SOURCE_PA
The size is (N_fp) and not (2,N_fp)!
FLOAT_DATA
The size is (N_a,N_c,N_f),not (N_c,N_f)!
OBSERVATION
OBSERVATION renamed SUBSCAN

gipsy

Last modified: Wed Nov 3 15:40:32 CET 2004