w*d
5 楼
sometimes not so easy to get a good high order scheme.
As I remember, the exist highest A-stable(?) RK scheme is
6th order.
If you have some special requirement, e.g. the speed,
the robustness, it is even harder to get
a good one. I once used the 4th order A-stable RK scheme
with only 3 stages. It worked very well sometimes. But for some
problems, the round-off error just killed the solution.
some useful references:
general RK:
Solving ordinary differential equations :
Hairer, E. (Er
【在 h***o 的大作中提到】
: 我觉得可以自己推呀, hoho...
As I remember, the exist highest A-stable(?) RK scheme is
6th order.
If you have some special requirement, e.g. the speed,
the robustness, it is even harder to get
a good one. I once used the 4th order A-stable RK scheme
with only 3 stages. It worked very well sometimes. But for some
problems, the round-off error just killed the solution.
some useful references:
general RK:
Solving ordinary differential equations :
Hairer, E. (Er
【在 h***o 的大作中提到】
: 我觉得可以自己推呀, hoho...
r*r
6 楼
^^^^^^^^^
it is SSP instead of A-stable.
As for A-stable, RADAU, developed
by Hairer (2002), uses A-stable
implicit RK method of variable
orders (between 5, 9 and 13).
Numerical experiments show that,
for stiff problems
【在 w**d 的大作中提到】
: sometimes not so easy to get a good high order scheme.
: As I remember, the exist highest A-stable(?) RK scheme is
: 6th order.
: If you have some special requirement, e.g. the speed,
: the robustness, it is even harder to get
: a good one. I once used the 4th order A-stable RK scheme
: with only 3 stages. It worked very well sometimes. But for some
: problems, the round-off error just killed the solution.
: some useful references:
: general RK:
t*o
8 楼
又多学一点
偶最多只用4阶R-K, 一直认为R-K公式是固定的,
u* = u^0 + dt*...
u** = u* + dt*...
想要多精度就是多写几次, u******, 系数不同而已
HOHO
【在 r*r 的大作中提到】
:
: ^^^^^^^^^
: it is SSP instead of A-stable.
: As for A-stable, RADAU, developed
: by Hairer (2002), uses A-stable
: implicit RK method of variable
: orders (between 5, 9 and 13).
: Numerical experiments show that,
: for stiff problems
偶最多只用4阶R-K, 一直认为R-K公式是固定的,
u* = u^0 + dt*...
u** = u* + dt*...
想要多精度就是多写几次, u******, 系数不同而已
HOHO
【在 r*r 的大作中提到】
:
: ^^^^^^^^^
: it is SSP instead of A-stable.
: As for A-stable, RADAU, developed
: by Hairer (2002), uses A-stable
: implicit RK method of variable
: orders (between 5, 9 and 13).
: Numerical experiments show that,
: for stiff problems
d*q
9 楼
taylor的
一种降阶形式
化为代数计算。。
不过用8阶也。。。【 在 twoxiao (老大,一点通) 的大作中提到: 】
一种降阶形式
化为代数计算。。
不过用8阶也。。。【 在 twoxiao (老大,一点通) 的大作中提到: 】
相关阅读
请教各位高手有限元的问题?提问:matlab效率问题(不同机器上)求教,哪有算int(exp(-t)/t, t)的fortran或者c++的程How to save variables in file in MATLABRe: f90 and mpiRe: How to save variables in file in MAT有什么算法可以确定一个点在不在多边形内?菜鸟问题does anyone ever used rlft3 code in NR?hillwind 可好?[转载] Normal mode analysis请问normal disctribution的cdf函数在FORTRAN或C里有求一个特殊矩阵的特征值问个matlab二重数值积分的问题[转载] Matlab编译Stand-alone application问题请教.概率统计问题[转载] 谁做过binary Lennard-Jones 的GCMC?我想知道assistant professor是否有启动资金?就是说,我有无可能Help, Question about Normalize a Matrix.[转载] 问几个名词