t*******t
发帖数: 1067 | 1 Thank you all in advance.
I have a function f(x)(>0 for any x) which is continues on [a,b],and wantto
find the the maximum of f(x) on [a,b]. I constructed an approximation like
this:t*Log(int_a^b(e^{f(x)/t}))/(b-a)), now I can prove that when t->0, the
approximation will will goes to max f(x) for x in [a,b], howerver, I could
not find the order of error. My feeling is that the error is in linear order
, but can not prove it. Could anybody help me to figure out this? thanks |
|
