freshs:documentation

This shows you the differences between two versions of the page.

Both sides previous revision Previous revision | |||

freshs:documentation [2016/03/03 13:20] josh Broke out a new section on postprocessing |
freshs:documentation [2016/03/03 13:21] (current) josh |
||
---|---|---|---|

Line 383: | Line 383: | ||

The FFS and flatPERM-FFS algorithms continue sampling at each interface until a fixed number of shots progress to the next interface. This type of trial produces successes according to a negative binomial distribution: for a single trial, the success probability is estimated simply as the proportion of successful shots; however for multiple trials (multiple independent FFS/FFS-flatPERM calculations) the correct averaging to determine the success probability is N_success / (<N_fails> + N_success). The number of failures must be saved and averaged in order to then find the expected success rate. | The FFS and flatPERM-FFS algorithms continue sampling at each interface until a fixed number of shots progress to the next interface. This type of trial produces successes according to a negative binomial distribution: for a single trial, the success probability is estimated simply as the proportion of successful shots; however for multiple trials (multiple independent FFS/FFS-flatPERM calculations) the correct averaging to determine the success probability is N_success / (<N_fails> + N_success). The number of failures must be saved and averaged in order to then find the expected success rate. | ||

- | When combining multiple independent simulations, the structure of the interfaces must be the same place for each calculation, and the shot databases post-processed to get the correct averages. Automatic interface placement is therefore contraindicated in this case. Only the probability of success (the total statistical weight) needs to be corrected for this effect; the proportional statistical weights of individual trajectories relative to each other remain correct. | + | When combining multiple independent simulations, the structure of the interfaces must be the same for each calculation, and the shot databases post-processed to get the correct averages. Automatic interface placement is therefore contraindicated in this case. Only the probability of success (the total statistical weight) needs to be corrected for this effect; the proportional statistical weights of individual trajectories relative to each other remain correct. |

==== Using the SQLITE Database of Trajectory Fragments ==== | ==== Using the SQLITE Database of Trajectory Fragments ==== |

freshs/documentation.1457007640.txt.gz · Last modified: 2016/03/03 13:20 by josh