Difference between SSP and SPheno numbers

[ahmedthabt]

Hello Florian,

I am studying a model in which a very large cancellation occurs in the calculation of a decay width, and computations are prone to small number errors that get magnified tremendously.

I used SPheno to compute the decay width, and then used SSP to make a scan and plot the decay width. Unexpectedly, the two results are VERY different, and the SSP result was confirmed using an analytical calculation by a fellow physicist.

My understanding of SSP is that it basically runs SPheno and just collects the numbers to be plotted. So how could they yield different numbers, with SSP appearing to be correct?

[FStaub]

Hi,

that sounds strange. As you say, SSP only reads in the SPheno results. It doesn't manipulate the numbers. You you attach an example (the SPheno.spc file from the scan and a plot done with SSP for the width you are interested in)?

Cheers,
Florian

[ahmedthabt]

Hello,
Find attached the SSP input file, the pdf plot, and the spectrum file. You can find what I am plotting in the spectrum file in a block named BR that I created by manipulating the InputOutput_U1plusSeesaw.f90 file to write the branching ratio of H->Z' Z' and also manipulated the BranchingRatios_U1plusSeesaw.f90 to output the decay width rather than the branching ratio (deleted the denominator)

In the SSP input file (U1plusSeesawgt.m) you can see I'm plotting BRhh which is read directly from the spectrum file.

Actually the direct SPheno and SSP numbers agree when I study the case without the huge cancellation, but in the case when those cancellations are present and small numbers play into the game, the two methods differ largely. I guess SSP might be forcing SPheno to compute with higher precision or something.

[FStaub]

Are you sure the you use the same settings in SphenoInput in the les houches and SSP file? Did you check the spectrum files written by SSP in the output directory?

[ahmedthabt]

Yes I am sure they are the same settings. I checked the spectrum files written by SSP and they are the same as in the plot.

[FStaub]

Could you try to run a single point with SSP and make a diff with the SPheno output. That might help to locate the difference.