Problems with propagation of error for multiple variables

  • Thread starter bobey
  • Start date
  • #1
30
0
please help by telling me whether my approach to solve the problems are right or wrong. please refer to the ATTACHMENT for the questions and my approaches...

your help is highly appreciated!
 

Attachments

  • ERROR ANALYSIS-2.doc
    156 KB · Views: 144

Answers and Replies

  • #2
chiro
Science Advisor
4,797
133
Hey bobey.

Your approach seems like you are using a vector-distance based on finding the length of the sum of all independent basis "errors".

The total differential of a multi-variable function describes the infinitesimal change of that function given the infinitesimal changes of all the variables with respect to its derivatives.

What you are doing looks good, but the only thing I would caution you about is what limits you put on the triangle quantities (i.e. the fixed deltas not the infinitesimal values).

Using a small enough value for the triangle delta's of your independent variables should always be OK, but if the function is changing wildly over some region, your errors will be way off. To look at whether this is the case you see how many first derivatives the function has (if any) that are equal to zero and also how the function behaves around your point (whether its derivative suddenly gets really high or really low just after your point).

The method though (calculating the error vector and finding its length) is a sound approach, but the important thing to be aware of is how local this information is and if it's not local (i.e. the stuff that I talked about above happens) then you need to re-consider looking at more global information like later derivatives at that point.

Looking at later derivatives is exactly what is done in numerical analysis when you get the "crazy functions" that have the potential to go wild and if you are interested in error analysis to take into account this behaviour, get some material on numerical analysis and differential equations.
 

Related Threads on Problems with propagation of error for multiple variables

Replies
1
Views
799
  • Last Post
2
Replies
29
Views
21K
Replies
5
Views
1K
Replies
2
Views
2K
Replies
7
Views
3K
Replies
2
Views
491
Replies
2
Views
498
  • Last Post
Replies
4
Views
676
Replies
4
Views
2K
Top