Understanding the Proper Uncertainty Formula to Use

AI Thread Summary
The discussion revolves around two formulas for calculating uncertainty in products, specifically for a rectangle's area. The first formula, which adds relative uncertainties, is suitable for dependent errors, while the second formula, which uses the square root of the sum of squares, applies to independent errors. Both methods yield the same result in this case due to the dominance of the larger relative error from one side of the rectangle. The choice of formula depends on whether the uncertainties are considered independent or dependent, with implications for engineering tolerances and statistical analysis. Understanding these nuances is crucial for accurate uncertainty calculations in various applications.
TRB8985
Messages
74
Reaction score
15
Homework Statement
A rectangle is 7.60 +/- 0.01 cm long and 1.90 +/- cm wide. Find the area of the rectangle and the uncertainty in the area.
Relevant Equations
See post for provided formulas.
Good evening,

I'm running into a little confusion on the second part of this problem due to finding two different formulas for calculating the uncertainty in multiplied quantities.

The way that I was taught was something like this.

If ##z = x \cdot y##, then: $$\dfrac{\delta z}{z} = \dfrac{\delta x}{x} + \dfrac{\delta y}{y}$$ Where ##\delta x##, ##\delta y##, and ##\delta z## represent the absolute uncertainties in ##x##, ##y##, and ##z## respectively.

With this approach, the uncertainty in the average area of this rectangle (14.4 cm) would be: $$\delta A = (14.4\: \text{cm}) \Big (\dfrac{0.01\: \text{cm}}{7.90\: \text{cm}} + \dfrac{0.01\:\text{cm}}{1.90\:\text{cm}} \Big) = 0.1\:\text{cm} $$ This matches what's listed in my textbook as the solution.

However, when looking a bit more into uncertainties online, I found this:

Statistics tells us that if the uncertainties are independent of one another, the uncertainty in a product is obtained by: $$\delta z = |z| \sqrt{\Big(\dfrac{\delta x}{x} \Big)^2 + \Big(\dfrac{\delta y}{y} \Big)^2 } $$ It's certainly true that the length and width of this rectangle are independent, and if I use the expression above, I still end up with the same answer. So then.. why the need for two formulas?
 
Physics news on Phys.org
Indeed the formula to use depends on whether you consider the errors independent or dependent. In this case, the side lengths are presumably from two independent length measurements (although there could technically be a systematic part as well such as using the same ruler, which may have a slightly faulty grading etc).

The reason you are getting the same result is that the relative error on the short side is so much larger that it completely dominates the result.
 
TRB8985 said:
Homework Statement: A rectangle is 7.60 +/- 0.01 cm long and 1.90 +/- cm wide. Find the area of the rectangle and the uncertainty in the area.
Relevant Equations: See post for provided formulas.
Please check the units of area in your answer.
 
  • Like
Likes TRB8985 and Orodruin
renormalize said:
Please check the units of area in your answer.
Agreed! Looks like I missed the square there in ##\delta A## at the end.

Thank you both, have a great weekend.
 
It depends what matters to you.

An engineer specifying manufacturing tolerances has to worry about worst case. If the bolt has diameter ##r_b\pm \Delta r_b## and the hole has diameter ##r_h\pm \Delta r_h## then it had better be that ##r_b+ \Delta r_b<r_h- \Delta r_h##.
The first formula you quote takes that view. It assumes that the two errors reinforce.

The statistical version considers that the errors could mitigate each other. It interprets the input uncertainties as proportional to standard deviations and computes (to the same proportion) the standard deviation of the result. This is certainly more appropriate when many uncertainties are combined. If the masses of certain coins each have standard deviation 0.1g, independently, then the s.d. of the mass of 100 of them is 1g, as given by the second formula, not 10g as given by the first.
 
Thread 'Collision of a bullet on a rod-string system: query'
In this question, I have a question. I am NOT trying to solve it, but it is just a conceptual question. Consider the point on the rod, which connects the string and the rod. My question: just before and after the collision, is ANGULAR momentum CONSERVED about this point? Lets call the point which connects the string and rod as P. Why am I asking this? : it is clear from the scenario that the point of concern, which connects the string and the rod, moves in a circular path due to the string...
Back
Top