Hello everyone!
I want to analyze the results of my cascade simulation, but one problem confused me.
In some publications, the relationship of number of defects vs. simulation time is shown in pictures. I also want to plot a picture like the one in article Journal of Nuclear Materials 249 (1997) 77-86, which is shown as the first attachment.
After the analysis of the defect numbers, I will get a file like the second attachment and I can plot Nd vs. time by using the second and the third columns. In order to reduce the statistic error, usually I should simulate, like saying, 10 cases with the same pka condition.
My problem is that, because the variable timestep is applied in my simulation, the second column of my results which represents the simulation time would be different in the 10 cases. Then it seems it is not possible to get the mean value of the number of defects at the exactly same time and which means I cannot get a picture like the first attachment.
Could anybody please give me some advice to help me solve this problem?
Any help would be much appreciated!
Best Regards
Liu
Ok, here are more details about Origin
Frist, use Origin to plot number vs. time curve.
Then, go to tool bar on the top and find "Analysis"
Once click "Analysis", you see a long list of possible actions. Select "Interpolate/Extrapolate"
A window will pop out. You need to click on three things "number of points", "x minimum" and "X maximum". X minimum is the staring time and X maximum is the ending time. "number of points" determine the time interval. For example, from 0 to 1ps, if you ask one thousand points, your time interval now is 0.001ps.
click Ok, you get your curve.
Usually we don't average curves. The defect number of a single cascades is high enough to avoid large statistics fluctuation. Some times fluctuation comes from channeling effect in which the projectile happens to bombard along crystal axial. The channelling effect is a true effect and is not statistics. Back to your questions about averaging of different time steps, you can use software called Origin. There is a function called extrapolation under data analysis. This function can insert points with forced distance. Hence all curves will change to same time steps.
Hi Liu.
I agree with Dr. Shao. I would add just a little tip. You should displace all your curves to t0 ; the time at first defects detection. Then you will able to correlate all your curves.
Best,
Diego
Dear Dr. Shao and Dr. Tramontina
I am sorry for replying this late and I am afraid that I do not quite understand your answers. I am trying to make my question clearer. Could you please give me more help?
After my simulation, I can get the defect numbers and a file like this:
Step time defects
100 0.007195188 2
1000 0.047129853 10
10000 3.426915898 901
Then I think I need to simulate 10 cases with the same PKA condition. Because the variable time step is applied, maybe the next file is like:
Step time defects
100 0.008156612 3
1000 0.054647746 13
10000 3.532166466 920
The other cases are similar, the time is different at the same steps. If I want to plot a picture of defect numbers vs. time, I cannot get the mean value of defect numbers at the exactly same time. And I am wondering what should I do?
Thank you very much!
Dear Dr. Shao, I know Origin. But I do not know how to apply extrapolation at my situation. Could you please give me more advice?
And Dear Dr. Tramontina, the (tj-ti) is not a constant. So even they have the same t0, it seems the subsequent time is still not same.
Do I think some thing wrong?
Thank you very much!
Ok, here are more details about Origin
Frist, use Origin to plot number vs. time curve.
Then, go to tool bar on the top and find "Analysis"
Once click "Analysis", you see a long list of possible actions. Select "Interpolate/Extrapolate"
A window will pop out. You need to click on three things "number of points", "x minimum" and "X maximum". X minimum is the staring time and X maximum is the ending time. "number of points" determine the time interval. For example, from 0 to 1ps, if you ask one thousand points, your time interval now is 0.001ps.
click Ok, you get your curve.
I have actually dealt with this problem in the past. Since you are using a variable time step, you have to do what other researchers are suggesting. Just interpolate between your variable time steps to get a consistent time value. Does this make sense?
I can give more details if you need.
Dear Benjamin Cowen,
Thanks very much for your information!
If I have further questions, I will write messages to you.
With My Best Regards,
Liu
- Badges
- Science topic