Auxiliary variable acceptable range of values and distribution

[AxMeNi]

Hi Julien and co,

I've started working with auxiliary variables for a simulation.
I am using continuous auxiliary variables.
I am simulating a discrete variable.
I would like to know what is the origin of this error when I run Deesse run:
ERROR 00556: invalid data in simulation grid: too far beyond the range covered by all the training image(s) (continuous variable)
Do you know what is the exact condtion that would allow me to remove this error ?
My problem is that I have three auxialiary variables, and each of them has the same max and min value in the TI and in the simulation grid.
But each of them don't have the same distributions in the TI and in the SG (only same extrema).
So I would like to know if there is a specific condition in the deesse code for the range and the distribution of the auxiliary variables.
Any help would be highly appreciated!

Cheers,
Axel

PS: I attached the histograms for each variable with on the left the one for the TI and on the right the one for the SG.

[julien-straubhaar]

Hello,

According to the "error 00556", it seems that a conditioning data value of a continuous variable (in the simulation grid) is not in the interval [vmin, vmax], where vmin (resp. vmax) is the min (resp. max) of the corresponding variable in the TI.

You said that this should not be the case. So I suggest you carefully check your input for deesse geone.deesseinterface.DeesseInput, in particular the parameters varname, the parameters dataImage and/or dataPointSet, and the order of the variables in the training image (parameter TI):

• varname[i] is the name used for the i-th variable in the training image;
• and the names of the variables in the data image and/or data point set are set according to the list varname.

Note that if there is really some conditioning data value out of the range of the TI values, you can

• use the parmeter expMax of geone.deesseinterface.DeesseInput: with a negative value, no error will occur and any situation is accepted, whereas a positive value gives the maximal relative expansion of the interval [min, max] of the TI variable so that it covers all conditioning data values
• use rescaling via the parameter rescalingMode, and rescalingTarget...

See the doc: https://geone.readthedocs.io/en/latest/api.html#deesseinterface.DeesseInput

[AxMeNi]

Thanks for your help,
Problem solved!
Axel

[julien-straubhaar]

If there is still a problem, you can show me the piece of code where the DeesseInput class is instantiated.
Julien