求助PRML中的一个变分推导

难道此题无人能解吗，希望有深入研究过PRML的同志来帮帮忙啊。

chensiyuan0510 发表于 2018-10-21 15:06
I still don\'t know how you removed the integral symbol on x and dx.

The arbitrary function phi(x) is chosen as an indicator function.  

soojinfan 发表于 2018-10-23 01:37
The arbitrary function phi(x) is chosen as an indicator function.

y(x) is a function of x, so is p(x,t). Setting phi(x) as an indicater function only replaced phi(x) with 1. The integral symbol on x is still there. There is only one integral symbol in 1.88.

chensiyuan0510 发表于 2018-10-23 08:54
y(x) is a function of x, so is p(x,t). Setting phi(x) as an indicater function only replaced phi(x ...

What about Dirac delta function,  if you more familiar with this term.  What I mean here is if we choose phi(x) in that way, then the integral of x will degenerate. But this is just my own opinion how we can reach (1.88) if all the previous derivations are correct.  

zhongdian_sjtu 发表于 2018-10-18 21:34
不是很确定，仅供参考吧，如果对了就请采纳一下。谢谢

这段计算应该是参考本书的Appendix D. Calculus o ...

您的解答应该是对的，我还找不到直接的证据，PRML后面的章节多次用到1.89，这样的大部头著作应该不可能大篇幅的出错。等我再找找办法，看看对函数求导那个环节是属于微积分的哪个部分。