Can I Plot Log (x) with Error Bars without Changing err_x Values?

  • Context: Undergrad 
  • Thread starter Thread starter Lorna
  • Start date Start date
  • Tags Tags
    Error Log
Click For Summary

Discussion Overview

The discussion revolves around the question of how to appropriately plot the logarithm of 'x' values with associated error bars, specifically whether adjustments to the error values ('err_x') are necessary when applying the logarithmic transformation.

Discussion Character

  • Exploratory, Technical explanation, Conceptual clarification

Main Points Raised

  • Lorna inquires whether changes to 'err_x' are needed when plotting 'log(x)'.
  • One participant suggests plotting log(1 + error) for positive errors and -log(1 - error) for negative errors to maintain scale.
  • Another participant questions the significance of the '1' in the proposed formulas and suggests testing with various error values.
  • A further response explains that '1' is chosen to ensure that error bars approach zero as the error approaches zero, while cautioning against using smaller or larger values.
  • Participants express confusion and seek clarification on the reasoning behind the choice of '1'.

Areas of Agreement / Disagreement

There is no consensus on the necessity of changing 'err_x' values, and participants express differing views on the significance of the '1' in the logarithmic transformation.

Contextual Notes

Participants discuss the implications of using different values in the logarithmic transformation and the potential effects on error representation, but no formal conclusions are reached.

Who May Find This Useful

This discussion may be useful for individuals interested in data visualization, particularly in the context of logarithmic transformations and error analysis in scientific plotting.

Lorna
Messages
44
Reaction score
0
Hello,

I have a column of 'x' values, each with an error 'err_x' serving as error bars.

My question is: If I need to plot 'log(x)', Do I need to make any changes to 'err_x' to reflect the log operation before plotting them as error bars ?

Thank you in advance

Lorna
 
Physics news on Phys.org
I would plot log(1 + error) for positive error and -log(1 - error) for negative error. That way error is on the same scale.
 
What is the significance of the '1' ? and why not just log(error)?
 
Lorna said:
What is the significance of the '1' ? and why not just log(error)?

Why don't you pick some errors in -1 < e < 1 and see for yourself?
 
Thank you, althought I still don't understand why "1" not another number
 
Lorna said:
Thank you, althought I still don't understand why "1" not another number

It's the only number that will make the error bars tend to 0 as the error approaches 0. With a smaller value very small errors won't show up at all (calculation error trying to take a negative log); with a larger value errors won't ever get closer than a certain distance from the point no matter how small the errors are.
 
Now I see. Thank you SO MUCH!
 

Similar threads

  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
Replies
28
Views
4K
  • · Replies 37 ·
2
Replies
37
Views
5K
Replies
2
Views
2K
  • · Replies 18 ·
Replies
18
Views
4K
  • · Replies 13 ·
Replies
13
Views
2K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 1 ·
Replies
1
Views
3K
Replies
8
Views
2K