So I set the correction factor to 0.01.
However, after running the program, I checked the output file (Inelastic Scattering Length for PROTON at Beam energy）and found that the interaction length λ hasn’t changed. It is still 15 cm.
I want to know where the factor in the LAM-BIAS card takes effect.
Also, how does this LAM-BIAS affect the calculation of the reaction cross-section of a thin target?
" RESNUCLE results " is φ_int / φ_o ， " Number of stars generated per beam particle " is x/λ (if x<<λ) ， " sigma " is 1/nλ .
we can see that the interaction length λ can be cancelled out and doesn’t play a role in the calculation formula. So how exactly does adding the LAM-BIAS card affect the calculation of the cross-section?
The biasing of the inelastic scattering length λ (through LAM-BIAS) is not meant to alter physics results at will, rather to reduce the associated statistical errors that measure their uncertainty. Note that the calculation of statistical errors implies to run multiple cycles and make use of the RANDOMIZe card, which your input misses (so yielding a non-sense zero error).
Moreover, you mix up reaction cross section with nuclide production cross section.
The first is already known at the beginning of the simulation as 1/nλ (with no statistical error associated) and determines the probability to have a nuclear reaction. Of course LAM-BIAS does not change the physical (true) value of λ appearing in the material table, but reduces it in the simulation according to the input factor, in order to artificially increase the reaction rate and get more statistics as far as the reaction products are concerned (that’s why the running time becomes longer). Nonetheless, the resulting nuclide yields (RESNUCLE results) - relevant to the nuclide production cross section calculation - are already properly corrected back by the same factor.
Note that both the reaction cross section and the nuclide production cross section (finally obtained from the three factors you recalled above, where λ is cancelled out only in the last two) do depend on λ, since RESNUCLE results (namely φ_int) actually depend on λ. In fact, RESNUCLE results depend on x too, that’s why one needs to divide them by the Number of stars (i.e., nuclear reactions), so as to remove the dependence on x (while preserving the dependence on λ).
hello! I’m here to consult you again.
Based on your previous guidance, I tried running my simulation again (with an initial number of particles of 5*10^6 ).
This time, I modified the LAM - BIAS to 0.00083. The cross - section result I finally calculated is (σ{225Ac}=0.589 mb). Without using the LAM - BIAS, (σ_{225Ac}=0.516 mb).
I’m wondering if this change in σ is normal. Have I used the LAM - BIAS correctly?
If I have used the LAM - BIAS card correctly, is it because the number of initial particles I set is too small? Or are there other errors that I haven’t taken into account?
If I want to learn in detail how to calculate the nuclide production cross - section, which FLUKA courses can I take to acquire relevant knowledge?
I’d appreciate it if you could give me some recommendations.
In addition, I have another question regarding “Number of stars generated by prompt PROTON” when calculating the nuclide production cross - section.
During the program run, several .out files will be generated, and the value of “Number of stars generated by prompt PROTON” in each .out file is different. So, when I calculate the nuclide production cross - section, how should I select the appropriate value of “Number of stars generated by prompt PROTON”?
You should look at the associated statistical errors (in the RESNUCLEi results).
On the contrary, it’s because the number of particles was already enough without LAM-BIAS, as to be confirmed by the respective statistical error.
As reference, for the natTh(p,x)225Ac excitation function, I obtain [from FLUKA 4-4.1] a local maximum slightly exceeding 1 mb around 45 MeV proton energy, which underestimates the experimental value you report.
FLUKA is not designed to extract microscopic cross sections, rather to calculate resulting macroscopic quantities.
Here are the results of my simulation: with an initial particle number of (10^8) and without using the LAM - BIAS card. It can be seen that the simulated values are all lower than the experimental data and the results from PHITS and MCNP6, especially when the incident energy is less than 90 MeV.
Since a thin target is set, I’m considering whether adding the LAM - BIAS can make the simulation results match the experimental data. However, you said…
On the contrary, it’s because the number of particles was already enough without LAM-BIAS, as to be confirmed by the respective statistical error.
So，apart from using the LAM - BIAS card, what other methods can be employed to improve the agreement between the simulation results and the experimental data?
I notice that above 90 MeV the FLUKA agreement with the data you report is quite good and much better than the other curves. On the other hand, below 90 MeV there is indeed an apparent underestimation.
That said, I reiterate that LAM-BIAS is not meant to alter the results, rather to reduce their statistical error so as to converge on the ‘true’ simulation value (that can be reached with appropriate statistics also without LAM-BIAS). Where the latter does not match experimental data, the issue has to be addressed on our development side by a consistent improvement of the reaction model and cannot be adjusted on the user side by an (unsupported) arbitrary modification of the FLUKA result.
