K*****N
发帖数: 117 | 1 作 numerical computation very intensive 的program, 在windows下面用vidio C++ 编译
译出来，再在dos下面run. 和用unix系统 编译并且run出来的 结果在iteration
到一定时间后出现很大的分歧。 那到底是那个 系统下面的结果更加可靠点呢？ | m******o
发帖数: 2 | 2 It depends on the problem you are solving. If the partial differential
equation itself is unstable (say turbulence, molecular dynamics), then even
the very tiny difference of machine truncation errors will eventually give you
very different answers. However, the statistical quantities(such as
correlation coefficients) should remain the same.
编译
【在 K*****N 的大作中提到】
: 作 numerical computation very intensive 的program, 在windows下面用vidio C++ 编译
: 译出来，再在dos下面run. 和用unix系统 编译并且run出来的 结果在iteration
: 到一定时间后出现很大的分歧。 那到底是那个 系统下面的结果更加可靠点呢？
| K*****N
发帖数: 117 | 3 hi, mojojojo:
yes, that's exactly what i meant and i was working on solving differential
equations. any paper or reference that could prove that statistical
quantities for such system will remain same, even for some unstable
system?
if you could give me some reference, it will be highly appreicated.
best wishes
【在 m******o 的大作中提到】
: It depends on the problem you are solving. If the partial differential
: equation itself is unstable (say turbulence, molecular dynamics), then even
: the very tiny difference of machine truncation errors will eventually give you
: very different answers. However, the statistical quantities(such as
: correlation coefficients) should remain the same.
:
: 编译
| m******o
发帖数: 2 | 4 Unfortunately, that statement is still a belief -- there're evidences but no
rigorous proof.
Some useful comments can be found on the book "understanding molecular
simulation -- from algorithms to applications" by Daan Frenkel & Berend Smit,
section 4.3. You can also find some references there.
even
you
C++
【在 K*****N 的大作中提到】
: hi, mojojojo:
: yes, that's exactly what i meant and i was working on solving differential
: equations. any paper or reference that could prove that statistical
: quantities for such system will remain same, even for some unstable
: system?
: if you could give me some reference, it will be highly appreicated.
: best wishes
| K*****N
发帖数: 117 | 5 woo, i got this book. section 4.3 is very very helpful for my research
project. thank you soo much.
best wishes
kowloon.
【在 m******o 的大作中提到】
: Unfortunately, that statement is still a belief -- there're evidences but no
: rigorous proof.
: Some useful comments can be found on the book "understanding molecular
: simulation -- from algorithms to applications" by Daan Frenkel & Berend Smit,
: section 4.3. You can also find some references there.
:
: even
: you
: C++
|
|
