Math Forum / Mathematics / Numerical Analysis / July 2009

Hey Folks,
Could someone with a strong background in floating-point accuracy
explain this to me? I posted this in the matlab group but I think I'll
have better luck here. Consider the matrix:

I have a simple 1D diffusion equation du/dt= d/dx[D(x)du/dx]
where diffusion coefficient becomes negative in some
range of r, i.e. where diffusion turns into concentration.
I used an implicit sweep (Thomas) method to solve it

Recently I have been porting a Fortran application to C. It is related
to molecular dynamics, and at the very core of the program we have an
integration engine using trivial Verlet integration and also the
leapfrog scheme. The original code in Fortran was written using

Newton method is widely used (with modifications) to solve nonlinear
equation(s).
Does anyone heard of combining Galerkin and Newton methods together?
What is

All I know are the intials 'RLH' for this site:
http://mymathlib.webtrellis.net/index.html
Fabulous collection covering a very wide spectrum of NA ideas.

◆→Brand Pants,Jeans,and Underwears A&F Pants,Bape Pants,ED Hardy
Pants,Coogi Pants,etc.&Evisu Jean,Iceberg Jean,RMC Jean,LV Jean,Ecko
Unltd Jean,etc.&Versace Underwear,D&G ,LV ,CK ,Gucci ,etc.=== All On
Site www.guomeitrade.com

We have got two vectors x[1], x[2], ..., x[n] and y[1], y[2], ..., y[n] .
There are two values a and b, that x[1] < a < b < x[n] . We assume that y[n]
= f(x[n]), there f is an C^n class function (i.e. is a continous function
and some of it derivatives are continuous too). The ...

It seems to me that the current approach that people are taking
towards CUDA is all wrong.  We should build the fundamental building
blocks first (meaning BLAS) and then all the other projects will fall
out of it naturally with little or no effort.

I'm computing the variance of a sequence and on a constant sequence, I
sometimes get something very small but negative, as in the example below.
I compute the variance as follows:
Var[x]=E[x^2]-E[x]^2

i have simple nonlinear system of equation with unknown x1, x2, ... x8
equations are like linear but there are multiplication terms like :
a1.x1+a2.x2+....+a5.x3.x5+a6.x4.x6+..=b1
and so on.

I was recently surprised to discover that I can beat the performance of the
Intel MKL on the dense QR decomposition of tall thin matrices with my own
small solution written in a high-level garbage collected language (F#).
I suspect the reason is that Intel extract the Q and R ...

hey I m new to ARPACK as well
After Installing the ARPACK it suggested to run dssimp.f
But i m getting error
it is showing like this

*******************************************************************
*******************************************************************
                           ENTER HERE:
  >>> http://your-guide-online.com/about/inkblot-test <<<

I met this problem recently.
Say S is an alternating series,
S = a1 - a2 + a3 - a4 + ....
It could be finite or infinite. For simplicity, let's assume S is finite.

Unfortunately there seems to be very little information on the net about
the Lehmer-Schur algorithm. Almost everything Google finds is either
dead-tree, locked behind some academic journal's paywall, or a mirror of
the Wikipedia article for it or subset of the same information, ...