some limits solved by integration



* [2006.12.10] Miguel Lopez wrote:
               ^^^^^^^^^^^^
> On Monday,  4 Dec 2006, Daniel Lakeland wrote:
>
>> On Sun, Dec 03, 2006 at 08:49:50PM +0000, Leo wrote:
>>> Hi all,
>>> 
>>> I am learning maxima. I want to use it to get the value of the
>>
>>> following when n goes to infinity.
>>> 
>>> 	n
>>>        ====
>>>        \	  1
>>> 	   >    -----------
>>>        /     sqrt(n + i)
>>>        ====
>>>        i = 1
>>>        -----------------
>>> 	    sqrt(n)
>
>      
>    
>    This type of problems can be solved easily but one needs:
>
>   1-) A maxima function to test if a function is decreasing.
>
>   2-) We have a limit of the type 0 x infinity, infinity with
>   integration
>  
>   3-) Then if the function beying integrated is decreasing it can be
> proved that we can replace the sum with a integration moreover we
> can replace the function with another such that limit f/g = 1
> (equivalent).  A program for doing this is more or less strait
> forward to do.
>
>    In this case one has the limit = 1/sqrt(n) *
>    integrate(1/sqrt,t,a,inf) = 2
>
>  Observe that replacing f by g such that lim f/g = 1 by tailor, this
> makes the integration much easier.
>
>    1/n * integrate(1/sqrt(n^3+a n^2 + b n + a)) = 1/n *
>    integrate(1/(an^3).
>
>   FWIW
>
>    Miguel.

I don't quite understand this. Could you try to use it to solve the
original limit problem? I have another approach leading to 2*sqrt(2)-2
but I am not 100% sure.

-- 
Leo