Chain rule with functional derivative

Click For Summary

Discussion Overview

The discussion revolves around the application of the chain rule in the context of functional derivatives, specifically examining the differentiation of the natural logarithm of a functional defined as an integral. Participants explore whether the chain rule for ordinary derivatives can be analogously applied to functional derivatives.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested

Main Points Raised

  • Some participants propose that the differentiation of the logarithm of a functional can be expressed as \(\frac{\partial}{\partial h}\ln(F)=\frac{1}{F}\frac{\delta F}{\delta h}=\frac{1}{F}\frac{\partial f}{\partial h}\).
  • Others suggest that the differentiation should consider a fixed parameter \(s'\), leading to the expression \(\frac{\delta}{\delta h(s')} \ln(F)\) and question whether the chain rule applies similarly to functional derivatives as it does to regular derivatives.
  • A participant expresses uncertainty about the validity of applying the chain rule to functional derivatives and notes the lack of examples in existing literature.
  • Another participant mentions the possibility of using the mathematical definition of functional derivatives to prove the validity of the chain rule.

Areas of Agreement / Disagreement

Participants do not reach a consensus on whether the chain rule for ordinary derivatives can be directly applied to functional derivatives. Multiple competing views and uncertainties remain regarding the appropriate approach to differentiation in this context.

Contextual Notes

Participants highlight limitations in available resources on functional differentiation and express challenges in understanding the parameters involved in the calculations.

jfitz
Messages
11
Reaction score
0
Given that

F = \int{f[h(s),s]ds}

does

\frac{\partial}{\partial h}ln(F)=\frac{1}{F}\frac{\delta F}{\delta h}=\frac{1}{F}\frac{\partial f}{\partial h}

?
 
Physics news on Phys.org
jfitz said:
Given that

F = \int{f[h(s),s]ds}

does

\frac{\partial}{\partial h}ln(F)=\frac{1}{F}\frac{\delta F}{\delta h}=\frac{1}{F}\frac{\partial f}{\partial h}

?

I don't think that means anything now. Shouldn't you be calculating

<br /> \frac{\delta}{\delta h(s&#039;)} \textrm{ln}(F)<br />

with some fixed s&#039;?
 
jostpuur said:
I don't think that means anything now. Shouldn't you be calculating

<br /> \frac{\delta}{\delta h(s&#039;)} \textrm{ln}(F)<br />

with some fixed s&#039;?
If so, then is

<br /> \frac{\delta}{\delta h(s&#039;)} \textrm{ln}(F) = \frac{1}{F[h(s&#039;)]}\frac{\delta F}{\delta h(s&#039;)}<br />

?Basically I'm not sure if the same kind of chain rule applies to functional derivatives as to regular ones. I haven't been able to find an example of a derivative of a function containing a functional like this natural log one.
 
jfitz said:
If so, then is

<br /> \frac{\delta}{\delta h(s&#039;)} \textrm{ln}(F) = \frac{1}{F[h(s&#039;)]}\frac{\delta F}{\delta h(s&#039;)}<br />

?

<br /> \frac{\delta}{\delta h(s&#039;)} \textrm{ln}(F[h]) = \frac{1}{F[h]}\frac{\delta F[h]}{\delta h(s&#039;)}<br />

F depends only on the mapping h, not on variable s. It seems to be okey to leave the parameter h (which is a mapping itself) out, and write F=F[h], but F[h(s&#039;)] would not make sense.

It's like here. If you have f:\mathbb{R}^3\to\mathbb{R}, and x\in\mathbb{R}^3, then you can write f(x), but f(x_3) (from x=(x_1,x_2,x_3)) would not make sense.

Basically I'm not sure if the same kind of chain rule applies to functional derivatives as to regular ones. I haven't been able to find an example of a derivative of a function containing a functional like this natural log one.

I have the same problem. I have never encountered a good source on functional differentiation. I can merely calculate, which is often enough, though. If you keep your head clear about simple things like what are parameters for each function, there are not many alternatives left for calculation rules.
 
Thanks
 
Couldn't one take the mathematical definition on the Wikipedia page I linked, and then prove whether the chain rule is valid or not? I think it would be analogous to proving the chain rule for ordinary derivatives via the limit definition of the derivative:

\frac{df}{dx} = \lim_{\Delta x \rightarrow 0} \frac{f(x + \Delta x) - f(x)}{\Delta x}

I would have done this yesterday for you, but I have other stuff I need to do right now.
 

Similar threads

  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 16 ·
Replies
16
Views
3K
  • · Replies 8 ·
Replies
8
Views
2K
  • · Replies 3 ·
Replies
3
Views
4K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 3 ·
Replies
3
Views
4K
  • · Replies 6 ·
Replies
6
Views
2K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 5 ·
Replies
5
Views
2K