Calculus of variations MOC

First variation

Let 𝑉 be a function space over 𝕂1 and 𝐹 :𝑉 →𝕂 be a functional, e.g. an Action functional. Loosely speaking, the first variation 𝛿𝐹 tells you how the functional 𝐹 varies in response to an infinitesimal variation 𝑓 →𝑓′ in its input, var i.e.

𝛿𝐹=𝐹[𝑓′]βˆ’πΉ[𝑓].

This is made rigorous by considering 𝛿𝐹 to be a functional in its own right.

A variation of 𝑓 is a map 𝛼? :( βˆ’πœ–0,πœ–0) →𝐹 such that 𝛼0 =𝑓. Let 𝒱𝑓 denote the function space of all variations of 𝑓.2 We define the functional 𝛿𝐹[𝑓] :𝒱𝑓 →ℝ so that

𝛿𝐹[𝑓;𝛼]=𝛿𝐹[𝑓][𝛼]=limπœ–β†’0𝐹[π›Όπœ–]βˆ’πΉ[𝑓]πœ–=𝐹[𝛼?]β€²(0)

This generalizes easily to the 𝑛th variation

𝛿𝑛𝐹[𝑓;𝛼]=𝐹[𝛼?](𝑛)(0).

Extrema of functionals

The main utility of 𝑛th variation is for identifying extrema of functionals as a necessary (but in general insufficient) condition under certain hypotheses. Suppose 𝑉 is topological and 𝐹 :𝑉 →𝕂 is continuous. Further let 𝑓0 βˆˆπ‘‰ be a local extremum of 𝐹.

  • If 𝛿𝐹[𝑓0] exists, then it is zero.
    • If 𝑓0 is a local minimum and 𝛿2𝐹[𝑓0] exists, then it is strictly positive.
    • If 𝑓0 is a local maximum and 𝛿2𝐹[𝑓0] exists, then it is strictly negative.

See also


develop | en | SemBr

Footnotes

  1. Where 𝕂 ∈{ℝ,β„‚}. ↩

  2. cf. homotopy. ↩