Message: Re: Cross-section Biasing Weights Not Logged In (login)

## Re: Cross-section Biasing Weights

Forum: Biasing and Scoring
Re: Cross-section Biasing Weights
Date: 21 Aug, 2017
From: Marc Verderi <Marc Verderi>

 Hello, You may refer to our last general paper for some more details, but, to summarize, when you modify the interaction law (and changing the cross-section falls in this case) you have two types of weight to consider: ``` - the weight for non-interaction, W_NI, over the step of length l - this the ratio of non-interaction probabilities over the step in the non-biased and biased schemes - in the case of cross-section change this is W_NI = exp[-(s_a-s_b)*l], where s_a and s_b are the analog and biased cross-sections respectively. In your case you have s_b > s_a hence -(s_a-s_b)*l > 0 and W_NI > 1 .``` ``` - the weight for interaction at distance l, W_I - this the ratio of interaction probabilities at distance l into the non-biased and biased schemes - in the case of cross-section change this is W_I = s_a / s_b and is 0.5 in your case.``` Then you have two possible cases : ``` - a step 0 -> l ending with no interaction (eg : geometry boundary) at distance l : in this case the particle weight is multiplied by W_NI ; always > 1 in your case - a step 0 -> l ending with an interaction at distance l : in this case the particle weight is multiplied by W_NI * W_I ; which can take any positive value, including values above 1.``` So having weights > 1 is indeed expected. ``` Cheers, Marc```

 Inline Depth: Outline Depth: Add message:

1 Re: Cross-section Biasing Weights   (ljb - 22 Aug, 2017)
Re: Cross-section Biasing Weights   (Marc Verderi - 23 Aug, 2017)
1 Re: Cross-section Biasing Weights   (ljb - 28 Aug, 2017)
1 Re: Cross-section Biasing Weights   (Marc Verderi - 28 Aug, 2017)
2 Re: Cross-section Biasing Weights   (ljb - 28 Aug, 2017)
3 Re: Cross-section Biasing Weights   (ljb - 28 Aug, 2017)
 to: "Re: Cross-section Biasing Weights"
This site runs SLAC HyperNews version 1.11-slac-98, derived from the original HyperNews