Re: finite difference scheme


In article <1133183123.817972.322930@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
"sakhan" <bsm_f02_035@xxxxxxxxxxxxxx> writes:
>hi, iam thankful for the attention that u have paid.
>
>i have used forward scheme at leftend a and backwardschme at right
>i-e b.
>
>the problem now is taht BVP fits well but the resulting system is not
>tridiagonal.
>
>for that i need three pt approxiamation of y```. which i could not
>derive.
>if u have that one plz do send taht to me .
>ihave not implemented the algorithim till now ,that is why iam not sure
>about the results.
>the required tridiagonal system is important as it would be efficient .
>do comment.............
>have a nice day...
>bye.
>

I have none because there exists none. In order to estimate a third derivative
you need at least four points
(interpolating polynomial of degree three with a constant nonzero third
derivative)
and this contradicts the tridiagonal structure.
it is not important to have the tridiagonal structure for efficiency, a (dense)
pentadiagonal is as well, since the solution effort will be O(n) again.
hth
peter
.

Quantcast