Hi,
The issue is that my calculation takes about 30 seconds, and Galapagos always uses the fitness value from the previous calculation instead of the current one. So, when I try to restore a version at the end, the calculation runs again, and the new result no longer matches the value Galapagos originally assumed.
I am working with the RFEM 6 interface and export the results as a CSV file after each calculation. This file is then read back into Grasshopper, and a specific value is extracted, which serves as the fitness value for Galapagos. I am using only a single parameter as the genome for Galapagos.
based on screenshots (so this is a pretty wild guess) I’d say to make sure the export boolean is always False, and gets set to True only after the OK button is clicked on the galapagos solver window:
I say that because -I guess- if that bool is already True when you re-open the Galapagos window for a new simulation, then the old Galapagos fitness value will still be present
Hi Inno,
As I understand Galapagos so far, I need to set the switch to true before opening the Galapagos window and starting the calculation. If the switch is set to false, the same result will always be returned as the fitness value, while the genome can change arbitrarily since no calculation takes place.
The problem is that, in the end, the values and settings in the Galapagos window do not match. When I try to restore a genome configuration, a different fitness value is calculated than the one displayed in the Galapagos window. I think this happens because Galapagos uses the fitness value from the previous calculation to optimize the parameters.
Unfortunately, I don’t know how to change this.
It would be nice, if anyone has an idea.
usually when you have a definition, everything is computed simultaneously, meaning of course it happens in the order dictated by the spaghetti connections, but unless there’s something that breakes the flow, the solution is computed until the end
for instance, let’s say you have just imported data_zero, you open galapagos window, in 30 seconds get the best result, at that point (not before) you have to save the result of data zero
then you proceed in loading data_one, and galapagos will still give you the result of data zero (because galapagos is manually triggered) so if you have a sort of “saving mechanism” that is always ON, that old value will be written before galapagos was triggered
so you open Galapagos, trigger calculation, close its window, nothing is saved (because it was already saved earlier) and proceed loading the new data_two, which gets instantly saved with the result of the previous galapagos iteration which is still available
that is what I’m thing is going on
could you provide the gh file of that? I don’t have the RFEM package, but I assume we might see what’s going on even without those components
The RFEM switch initiates the entire calculation process with RFEM, regardless of whether I use Galapagos or not. When the switch is active, a geometry is exported to RFEM, then calculated in RFEM, and finally exported as a CSV file to a folder.
At the same time, I use the Read File component to read the CSV file and extract a value from it. This value is then supposed to be used as the fitness value. The Read File component updates every time there is a change in the CSV file.
Unfortunately, I won’t have access to my .gh file until Monday, but I’ll be happy to upload it then. In the meantime, I think I’ve reproduced the issue using a Data Dam in a similar way—at least the outcome is the same. That should give you an idea of what the problem with the result is.
