[QUANTIT] Use-cases, role in larger scheme (Was: Re: [QUANTITY] Quantity "arguments")

[Ed Shaya]

DIDELON Pierre wrote:

>
>I don't understand. I am lost.
>Errors exist in Pat's model, it is agregate one level "above" in Measurement. 
>Some measurements (or whaever name will be agreed on) will have errors and some not.
>Why trying to forbidd the concept of number without error
>if it can be introduce (for free) and claim as necessary by some people?
>
>  
>
I do not want to forbid a class of objects with no error, but there is 
something  wrong here.

a) If Measurement is really an aggregate of number+unit and error, then 
number+unit is not a subclasss or  superclass or Measurement, and then  
we are saying they  are not related.  Remember, aggregation says nothing 
about class relationship.  But we know that they share much of the same 
properties and capabilities and functions.  So you are going to end up 
creating some superclass for the two of them for convenience. 
b)If Measurement is a subclass of number+unit, then that is really ugly 
because you have this thing with very few instances having a subclass 
with many more instances. 
c) The last option is number+unit is a subclass of Measurement.  Now you 
are saying that the thing that definitely has an error, since all real 
measurements have some amount error,  has a subclass that definitely 
does not have an error.
The only viable option was a).  So now we need a word for something that 
can be a measurement (with error) or a parameter (without error).  This 
is missing in Pat's UML, but a good choice for such a class is quantity.

Here is an example that might work if we go this route.  A certain 
planetary nebula has FluxProperty.  There have been many Observations of 
it.  Each Observation of Flux consists of a set of Measurements of 
various quantities such as the image, of course, and the filter 
throughput, and the airmass etc.  When we define the class Observation 
we can just say it holds a set of Quantities.  Now we can say find 
FluxProperty/Observations  constrained by the QuantityWithNoError, 
airmass > 1.5, and this is directly compared with Measurement airmass = 
1.2+/-.1.

Ed