Alternate expressions for the uncertainty propagation

Click For Summary
SUMMARY

This discussion focuses on the two primary formulas for calculating uncertainty propagation in measurements: the linear approximation method and the standard deviation method. The first formula, represented as $$\Delta X=\left|\frac{\partial f}{\partial A}\right| \cdot \Delta A+\left|\frac{\partial f}{\partial B}\right| \cdot \Delta B+\cdots$$, is used for simpler estimates, while the second formula, $$\sigma_{X}=\sqrt{ \left(\frac{\partial f}{\partial A} \right)^{2}\sigma_{A}^2+\cdots }$$, provides a more accurate representation by considering the standard deviations of the variables involved. The choice between these formulas depends on the desired accuracy and the correlation between the variables, as highlighted in the example using Ohm's law.

PREREQUISITES
  • Understanding of basic calculus, specifically partial derivatives
  • Familiarity with the concepts of uncertainty and standard deviation
  • Knowledge of Ohm's law and electrical measurements
  • Basic proficiency in error analysis techniques
NEXT STEPS
  • Study the application of Taylor series in error propagation
  • Learn about the correlation of variables in uncertainty analysis
  • Explore advanced topics in statistical mechanics related to uncertainty
  • Investigate software tools for uncertainty analysis in experimental data
USEFUL FOR

Researchers, engineers, and students in physics or engineering fields who are involved in experimental measurements and require precise calculations of uncertainty propagation.

AndersF
Messages
27
Reaction score
4
TL;DR
What criteria to use to decide which of the two formulas to use.
I have seen that there are two different formulas that we can use when calculating the propagation of uncertainty in a measurement. If ##X=f(A, B, C, \ldots)## is the quantity whose uncertainty we want to estimate, which depends on the quantities ##A,B,C,...##, then we could calculate the propagation of uncertainty either by (1) or by (2):

$$\Delta X=\left|\frac{\partial f}{\partial A}\right| \cdot \Delta A+\left|\frac{\partial f}{\partial B}\right| \cdot \Delta B+\left|\frac{\partial f}{\partial C}\right| \cdot \Delta C+\cdots \tag{1}$$

$$\sigma_{X}=\sqrt{ \left(\frac{\partial f}{\partial A} \right)^{2}\sigma_{A}^2+\left(\frac{\partial f}{\partial B} \right)^{2}\sigma_{B}^2+\left(\frac{\partial f}{\partial C} \right)^{2}\sigma_{C}^2+\cdots } \tag{2}$$

With ##\Delta X, \Delta A,\Delta B,\Delta C,...## the uncertainties in the values of ##X, A,B,C,...## and ##\sigma_X,\sigma_A,\sigma_B,\sigma_C,...## the standard deviation in these measurements.

What is the difference between these two expressions? When do we choose equation (1) or equation (2) for estimating the propagation of uncertainty in a value?

____________________________________________________________________________​

For example, if we wanted to estimate the uncertainty in the calculation of the value of some resistors from the measured values in current and voltage, ##V = 2.04,\space 2.10,\space 2.19 \space V##, ##I =16.8,\space 28.7 ,\space 63.7\space mA##, whit ##\Delta V = 0.01\space V## and ##\Delta I =0.1\space mA## the uncertainties in these measurements, applying (1) and (2) to Ohm's law would give (3) and (4), next to next:

$$\Delta R=\left|\frac{1}{I}\Delta V\right|+\left|\frac{V}{I^{2}} \Delta I\right| \tag{4}$$
$$\sigma_{R}=\sqrt{ \frac{1}{I^2}\sigma_{I}^2 + \frac{V^2}{I^{4}}\sigma_{B}^2} \tag{2}$$

Then, for the ##R=121.42857..., \space 73.17073..., \space 34.37990... \space \Omega## calculated values, if we take ##\sigma_V= \Delta V## and ##\sigma_I= \Delta I##, we would get these values, which in this case are slightly different:

$$\Delta R=1,\space 0.6,\space 0.2\space \Omega$$
$$\sigma_{R} = 0.9,\space 0.4,\space 0.2\space \Omega$$

Therefore, what would be the criteria for deciding whether to use (1) and (2), for this example and for a general case?
 
Physics news on Phys.org
Picking out formulas willy-nilly is going to cause you trouble. If you need more than an estimate- and either will do that - you need to understand error propegation at the level of, say, Taylor's book. The difference between those two expressions is the degree of correlkation.
 

Similar threads

  • · Replies 8 ·
Replies
8
Views
3K
  • · Replies 8 ·
Replies
8
Views
2K
  • · Replies 6 ·
Replies
6
Views
3K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 16 ·
Replies
16
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 2 ·
Replies
2
Views
3K
  • · Replies 6 ·
Replies
6
Views
4K
  • · Replies 13 ·
Replies
13
Views
3K