Proof of the Frenet-Serret formulae

  • Thread starter Thread starter CAF123
  • Start date Start date
  • Tags Tags
    Formulae Proof
Join the discussion
Registration is free. Start your own thread to ask a follow-up.
2 replies · 14K views
CAF123
Gold Member
Messages
2,918
Reaction score
87

Homework Statement


Consider the unit tangent vector [itex]T[/itex] unit normal vector [itex]N[/itex] and binormal vector [itex]B[/itex] parametrized in terms of arc length s.
1) Show that [tex]\frac{dT}{ds} = \kappa\,N[/tex]

I think this part is fine for me. What I did was: [tex]N(t) = \frac{T'(t)}{|T'(t)|}[/tex] and said, by the chain rule, [itex]\frac{dT}{ds} \frac{ds}{dt}= T'(t)[/itex] which simplified to [tex]N(s) = \frac{|r'(t)|}{|T'(t)|} \frac{dT}{ds} => \frac{dT}{ds} = \kappa N(s)[/tex]
Can somebody confirm this is correct?

2) Use a) to show that there exists a scalar [itex]-\tau[/itex] such that [tex]\frac{dB}{ds} = -\tau\,N[/tex]

I was given a hint to try to show that [itex]\frac{dB}{ds} . B = 0[/itex]
I took the derivative [tex]\frac{d}{ds} B = \frac{d}{ds}(T ×N) = T ×\frac{dN}{ds}[/tex]
Therefore, [tex](T × \frac{dN}{ds}) . B = (B ×T) . \frac{dN}{ds} = N . \frac{dN}{ds}.[/tex] Am I correct in assuming the above is equal to 0?

Many thanks.
 
Physics news on Phys.org
CAF123 said:

Homework Statement


Consider the unit tangent vector [itex]T[/itex] unit normal vector [itex]N[/itex] and binormal vector [itex]B[/itex] parametrized in terms of arc length s.
1) Show that [tex]\frac{dT}{ds} = \kappa\,N[/tex]

I think this part is fine for me. What I did was: [tex]N(t) = \frac{T'(t)}{|T'(t)|}[/tex] and said, by the chain rule, [itex]\frac{dT}{ds} \frac{ds}{dt}= T'(t)[/itex] which simplified to [tex]N(s) = \frac{|r'(t)|}{|T'(t)|} \frac{dT}{ds} => \frac{dT}{ds} = \kappa N(s)[/tex]
Can somebody confirm this is correct?

It's weird that you need to use the chain rule. But first, what is the definition of [itex]\kappa[/itex]??

2) Use a) to show that there exists a scalar [itex]-\tau[/itex] such that [tex]\frac{dB}{ds} = -\tau\,N[/tex]

I was given a hint to try to show that [itex]\frac{dB}{ds} . B = 0[/itex]
I took the derivative [tex]\frac{d}{ds} B = \frac{d}{ds}(T ×N) = T ×\frac{dN}{ds}[/tex]
Therefore, [tex](T × \frac{dN}{ds}) . B = (B ×T) . \frac{dN}{ds} = N . \frac{dN}{ds}.[/tex] Am I correct in assuming the above is equal to 0?

Yes, it is true that [itex]N\cdot \frac{dN}{ds}=0[/itex], but it needs to be proven. To prove this, consider the function [itex]N\cdot N[/itex] and take derivatives.
 
micromass said:
It's weird that you need to use the chain rule. But first, what is the definition of [itex]\kappa[/itex]??
[itex]\kappa[/itex] is the rate of change of the tangent vector of a curve with respect to arc length, ie a measure of how much the tangent vector changes in magnitude and direction as a point moves along a curve. Why is it weird using chain rule - did you have another method in mind - is mine ok?


Yes, it is true that [itex]N\cdot \frac{dN}{ds}=0[/itex], but it needs to be proven. To prove this, consider the function [itex]N\cdot N[/itex] and take derivatives.
Thanks, so an argument could go;[tex]\frac{d}{ds} (N\cdot N) = 2(N\cdot \frac{dN}{ds}).[/tex] We know the dot product is a scalar quantity, so the left hand side is zero (derivative of a constant) and so the right hand side is only true provided this term in brackets is zero.