Propagation of uncertainty with tangent function?

5 replies · 7K views · last reply
Cheesycheese213
Messages
55
Reaction score
8
Homework Statement
Calculate the uncertainty of a value (n) calculated by the equation n=1.0003tanθ
Relevant Equations
(shown below)
For a lab, I needed to calculate the uncertainty of a refractive index that was found using Snell's law. I found an equation online for propagation of error for any general function, which was
Equation for propagation of uncertainty of a function f(x,y): (Δz)^2 = (∂f(x,y)/∂x Δx)^2 + (∂f(x,y)/∂y Δy)^2, with Δz^2 on the left.

I thought that since my equation was
Handwritten equation n2 = 1.0003 tan θB, where θB is the angle variable.

I could just get rid of the variable y, and have
Two equations for propagated uncertainty: Δn2 = 1.0003 |d tan θB/dθB · ΔθB| and Δn2 = 1.0003 |(sec² θB) Δθ B|, the second with d tan θB/dθB written as sec² θB.


After inputting my values θ = 53.61 and Δθ = 0.1, I got
Δn = 0.284193...
Which I was sort of confused by? My value for n itself is only around 1.357, and I wasn't sure why the uncertainty was so large in comparison. I was wondering if I did something wrong?

For more context, I am calculating n for many different angles, and when graphing the values either the error bars look really big, or the slope is almost insignificant from the scale.

Thanks!
 
Attachments
  • Equation: Δn2 = |(d tan θB / dθB) Δθ_B|, the propagation of uncertainty for a refractive index n2 that depends on the angle θB through a tangent function.
    Screen Shot 2020-03-08 at 6.52.40 PM.png
    926 bytes · Views: 339
  • Handwritten derivation of the uncertainty formula Δn2 = |(1.0003 sec^2 θB) ΔθB|, showing the derivative of the tangent-based expression multiplied by the angle uncertainty.
    Screen Shot 2020-03-08 at 6.53.06 PM.png
    1,014 bytes · Views: 323
Try using Radians for your angle measurement, instead of degrees. That's the only way you can take derivatives of trig functions.

Also, depending on the software package you use to graph, it could be defaulting to radians.
 
  • Like
Likes   Reactions: sysprog and Cheesycheese213
scottdave said:
Try using Radians for your angle measurement, instead of degrees. That's the only way you can take derivatives of trig functions.

Also, depending on the software package you use to graph, it could be defaulting to radians.
Ohhh thank you so much!
I just tried converting everything to radians, and I got a much smaller number (0.004961...). Its really interesting how it needs radians over degrees :D Thank you again!
 
  • Like
Likes   Reactions: sysprog, jim mcnamara and scottdave
Cheesycheese213 said:
interesting how it needs radians over degrees
It's not so much that it has to be radians. Rather, it is because the standard derivatives we learn, like d sin(x)/dx = cos (x), assume x is in radians.
 
  • Like
Likes   Reactions: Cheesycheese213, scottdave and sysprog
haruspex said:
It's not so much that it has to be radians. Rather, it is because the standard derivatives we learn, like d sin(x)/dx = cos (x), assume x is in radians.

Ohhhh I see! I didn't connect that! Thank you!
 
Yes, like @haruspex said, you could have a scale factor of (π radians)/180° inside the sin(x), so taking the derivative of sin(x) with x in degrees becomes (π/180)*cos(x).

This means the sine would change approx 0.01745 vertically for every Δx = 1° ( when you are near x = 0 ).