- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Santana, Mario <msantana_at_slac.stanford.edu>

Date: Thu, 17 Jul 2014 21:20:46 -0700

If you have a statistical error of 0 this does not necessarily (and in practical terms almost never) mean that your simulation has converged, on the contrary, it means that there were no events there (which probably means that you should run longer). Your 1-D plot shows an error larger than 0 (and smaller than 99), meaning that it did register some events. This is not contradictory because depending how you sent your 1D histogram, each bin can embrace many of your 3D voxels, and therefore statistics can be better.

A simple yet quit effective way to judge whether your simulation is converging is to visualize the histograms and observe if fluences, doses, etc. have smooth isocurves where they are supposed to.

From: 闫渊林 <yanyuanlin_at_impcas.ac.cn<mailto:yanyuanlin_at_impcas.ac.cn>>

Date: Thursday, July 17, 2014 8:29 PM

To: Mario Santana <msantana_at_slac.stanford.edu<mailto:msantana_at_slac.stanford.edu>>

Cc: fluka-discuss <fluka-discuss_at_fluka.org<mailto:fluka-discuss_at_fluka.org>>

Subject: Re: Re: [fluka-discuss]: Percentage error in usrbin

Dear Santana, Mario

Thank you for your response!

Can I judge if the number of events is enough by the statistical error?But sometimes the statistical error in dose_50.bnn.lis is 0 while in z.dat is 45,which one should I refer to?

Best Wish!

yanyuanlin

-----原始邮件-----

发件人: "Santana, Mario" <msantana_at_slac.stanford.edu<mailto:msantana_at_slac.stanford.edu>>

发送时间: 2014年7月16日 星期三

收件人: "闫渊林" <yanyuanlin_at_impcas.ac.cn<mailto:yanyuanlin_at_impcas.ac.cn>>, "fluka-discuss_at_fluka.org<mailto:fluka-discuss_at_fluka.org>" <fluka-discuss_at_fluka.org<mailto:fluka-discuss_at_fluka.org>>

抄送:

主题: Re: [fluka-discuss]: Percentage error in usrbin

1. Yes, it is the statistical error

2. Again, the statistical error (third column is the average, first and second the min and max of your bin)

From: 闫渊林 <yanyuanlin_at_impcas.ac.cn<mailto:yanyuanlin_at_impcas.ac.cn>>

Date: Monday, July 14, 2014 8:09 PM

To: fluka-discuss <fluka-discuss_at_fluka.org<mailto:fluka-discuss_at_fluka.org>>

Subject: [fluka-discuss]: Percentage error in usrbin

Dear exports'

I use usrbin of 2*2*5 bins to scoring dose. I ascii the usrbin result dose_50.bnn to dose_50.bnn.lis and plot the result dose_50.bnn with gnuplot in flair :z.dat(I choose z,and the limit of x,y,z is default). two questions:

1.In dose_50.bnn.lis, what dose the mean of the Percentage errors follow in a matrix A(ix,iy,iz), format (1(5x,1p,10(1x,e11.4)))

4.6420E+00 3.5649E+00 2.8894E+00 3.1111E+00 4.7960E+00 3.3774E+00 2.5223E+00 3.1620E+00 4.6191E+00 3.1215E+00

2.1012E+00 3.2304E+00 4.5331E+00 3.2791E+00 2.5898E+00 3.3436E+00 4.3849E+00 3.5383E+00 3.3751E+00 3.5247E+00

Is it the statistic error of 5 runs? what king of error?

2.In z.dat dose the mean of the 4th column data as follows?

1.7919530&n! bsp;

1.7600398

1.6749210

1.7426424

1.8636131

1.8636131

Thank you !

Yanyuanlin

Received on Fri Jul 18 2014 - 08:15:22 CEST

Date: Thu, 17 Jul 2014 21:20:46 -0700

If you have a statistical error of 0 this does not necessarily (and in practical terms almost never) mean that your simulation has converged, on the contrary, it means that there were no events there (which probably means that you should run longer). Your 1-D plot shows an error larger than 0 (and smaller than 99), meaning that it did register some events. This is not contradictory because depending how you sent your 1D histogram, each bin can embrace many of your 3D voxels, and therefore statistics can be better.

A simple yet quit effective way to judge whether your simulation is converging is to visualize the histograms and observe if fluences, doses, etc. have smooth isocurves where they are supposed to.

From: 闫渊林 <yanyuanlin_at_impcas.ac.cn<mailto:yanyuanlin_at_impcas.ac.cn>>

Date: Thursday, July 17, 2014 8:29 PM

To: Mario Santana <msantana_at_slac.stanford.edu<mailto:msantana_at_slac.stanford.edu>>

Cc: fluka-discuss <fluka-discuss_at_fluka.org<mailto:fluka-discuss_at_fluka.org>>

Subject: Re: Re: [fluka-discuss]: Percentage error in usrbin

Dear Santana, Mario

Thank you for your response!

Can I judge if the number of events is enough by the statistical error?But sometimes the statistical error in dose_50.bnn.lis is 0 while in z.dat is 45,which one should I refer to?

Best Wish!

yanyuanlin

-----原始邮件-----

发件人: "Santana, Mario" <msantana_at_slac.stanford.edu<mailto:msantana_at_slac.stanford.edu>>

发送时间: 2014年7月16日 星期三

收件人: "闫渊林" <yanyuanlin_at_impcas.ac.cn<mailto:yanyuanlin_at_impcas.ac.cn>>, "fluka-discuss_at_fluka.org<mailto:fluka-discuss_at_fluka.org>" <fluka-discuss_at_fluka.org<mailto:fluka-discuss_at_fluka.org>>

抄送:

主题: Re: [fluka-discuss]: Percentage error in usrbin

1. Yes, it is the statistical error

2. Again, the statistical error (third column is the average, first and second the min and max of your bin)

From: 闫渊林 <yanyuanlin_at_impcas.ac.cn<mailto:yanyuanlin_at_impcas.ac.cn>>

Date: Monday, July 14, 2014 8:09 PM

To: fluka-discuss <fluka-discuss_at_fluka.org<mailto:fluka-discuss_at_fluka.org>>

Subject: [fluka-discuss]: Percentage error in usrbin

Dear exports'

I use usrbin of 2*2*5 bins to scoring dose. I ascii the usrbin result dose_50.bnn to dose_50.bnn.lis and plot the result dose_50.bnn with gnuplot in flair :z.dat(I choose z,and the limit of x,y,z is default). two questions:

1.In dose_50.bnn.lis, what dose the mean of the Percentage errors follow in a matrix A(ix,iy,iz), format (1(5x,1p,10(1x,e11.4)))

4.6420E+00 3.5649E+00 2.8894E+00 3.1111E+00 4.7960E+00 3.3774E+00 2.5223E+00 3.1620E+00 4.6191E+00 3.1215E+00

2.1012E+00 3.2304E+00 4.5331E+00 3.2791E+00 2.5898E+00 3.3436E+00 4.3849E+00 3.5383E+00 3.3751E+00 3.5247E+00

Is it the statistic error of 5 runs? what king of error?

2.In z.dat dose the mean of the 4th column data as follows?

1.7919530&n! bsp;

1.7600398

1.6749210

1.7426424

1.8636131

1.8636131

Thank you !

Yanyuanlin

Received on Fri Jul 18 2014 - 08:15:22 CEST

*
This archive was generated by hypermail 2.3.0
: Fri Jul 18 2014 - 08:15:27 CEST
*