Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Question on conditional statement
Posted 4 oct. 2010, 02:51 UTC−4 2 Replies
Please login with a confirmed email address before reporting spam
Hi All,
Can any one tell me the difference between these conditions:
1. if((x<1.5),k*c,0)
2. flc1hs(1.5-x,1e-8)*k*c
3. flc1hs(1.5-x,1e-20)*k*c
Because,
if i use 1 & 3 conditions then my model is stop converging in the middle of simulation. it is solving some time steps but not all.
if i use 2nd condition then my model is converging and solving complete time steps..
So, i am just confused and unable to think/imagine what went wrong with 1& 3.
I hope some one will come up with comments/ideas/suggestions..
Thanks,
Manohar
Can any one tell me the difference between these conditions:
1. if((x<1.5),k*c,0)
2. flc1hs(1.5-x,1e-8)*k*c
3. flc1hs(1.5-x,1e-20)*k*c
Because,
if i use 1 & 3 conditions then my model is stop converging in the middle of simulation. it is solving some time steps but not all.
if i use 2nd condition then my model is converging and solving complete time steps..
So, i am just confused and unable to think/imagine what went wrong with 1& 3.
I hope some one will come up with comments/ideas/suggestions..
Thanks,
Manohar
2 Replies Last Post 5 oct. 2010, 01:38 UTC−4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
1. if((x<1.5),k*c,0)
This is a step function. meens exactly at x=1.5 you will have a very sharp change. Solvers hate it.
2. flc1hs(1.5-x,1e-8)*k*c
Means at x=1.5 you have a smoothing of 1e-8 around it. so your value increases from 0 to 1 in a 1e-8 window. It is still very small (unless you work with lasers etc)
3. flc1hs(1.5-x,1e-20)*k*c
This is bad, nearly the same as 1, since 1e-20 is effectively 0. So your increase window is 0, so it is practically a on-off swicth, like 1. Solvers hate it :)
This is a step function. meens exactly at x=1.5 you will have a very sharp change. Solvers hate it.
2. flc1hs(1.5-x,1e-8)*k*c
Means at x=1.5 you have a smoothing of 1e-8 around it. so your value increases from 0 to 1 in a 1e-8 window. It is still very small (unless you work with lasers etc)
3. flc1hs(1.5-x,1e-20)*k*c
This is bad, nearly the same as 1, since 1e-20 is effectively 0. So your increase window is 0, so it is practically a on-off swicth, like 1. Solvers hate it :)
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
Knowing that "eps" (the smallest number step distinguisable from 1) is 2.22e-16 on my PC, it is really not worth playing with ratios that compare to "eps" even to within a few tousand time.
The binary number system is far from "perfect" and rather different from our decimal representation
--
Good luck
Ivar
Knowing that "eps" (the smallest number step distinguisable from 1) is 2.22e-16 on my PC, it is really not worth playing with ratios that compare to "eps" even to within a few tousand time.
The binary number system is far from "perfect" and rather different from our decimal representation
--
Good luck
Ivar