Saved history cannot be loaded? about pynoddy HOT 15 CLOSED

And @flohorovicic - yes; other than the padding, I can load and run .his files created using the Noddy GUI without any (other) issues

from pynoddy.

Oh, and this is the generated .his file ==> seems fine to me.
model.txt

from pynoddy.

from pynoddy.

Good idea - the output saved by Noddy can be loaded by pynoddy with no issue - though the differences seem rather trivial to me (see below; red is from noddy, green from pynoddy). Perhaps a floating point issue? Anyway - I'm curious to see if anyone else can reproduce this, or if I just have something weird going on locally.

model001.txt

from pynoddy.

Gah! There's deffo something funny going on.... I'm getting outputs that have the wrong extent too (tested using .his files generated with the GUI). I can change the cell size, but somehow the extent (calculated via pynoddy) is fixed to (3400.0, 3400.0, 500.0). Maybe I'm just allergic to Macs 😅

his = NoddyHistory("axes.his")
print(".his extent = ", his.get_extent())
pynoddy.compute_model("axes.his","0_output", sim_type='BLOCK')
model = NoddyOutput("0_output")
print("Output extent = ", (model.xmax - model.xmin, model.ymax - model.ymin, model.zmax - model.zmin))
print("              =  %s %d m blocks" % (model.block.shape,model.delx) )

.his extent = (1000.0, 1000.0, 500.0)
Output extent = (3400.0, 3400.0, 500.0)
= (170, 170, 25) 20 m blocks

axes.txt

from pynoddy.

Sorry for the late answer - how could I miss the notification for a pynoddy error :-D I'll have a look - maybe related to an additional (or a missing) new line?

from pynoddy.

So, I can open both files that you attached above just fine. Can you send the model.his that fails over again?

from pynoddy.

I figured out the problem with the save/load issue - somehow pynoddy was installing an older version; when I instead installed it with python setup_without_SALib.py install then it fixed the issue (I noticed because the version number in the .his files was different to what pynoddy should be writing).

Still haven't figured out why the noddy executable is building models with incorrect extents though. Could be something strange going on with my setup again... If you run the axes.his file do you get the correct extent? (1000,1000,500)?

from pynoddy.

Thanks, that's an important point - I need to update the pip release!

Concerning the extent: I get:

from pynoddy.

Thanks for testing - I suppose this must be some strange thing I have going on locally. Will try installing pynoddy from scratch and see if that helps. I guess we can close this issue now 👍

from pynoddy.

Small update - I managed to get noddy to compute with the correct extents by copying the events and block options (and nothing else) from the .his file (Version = 7.11) exported from the Noddy GUI into a .his file created from scratch using pynoddy (Version = 7.03).

Doing a diff between the two files (see attached) it seems there are quite a few differences in the GeophysicsOptions - could this be causing the padding to occur somehow? (given I presume Noddy does pad the model when calculating gravity / magnetic forward models? ).

axes_pynoddy.his.txt

axes_noddy.his.txt

from pynoddy.

from pynoddy.

Thanks a lot for clarifying, @markjessell! It was not obvious to me that you were trying a geophysical forward model, @samthiele. But it should still be possible to load the .his-file created in Noddy with pynoddy. Does that work now (considering the aspect of extra padding)?

from pynoddy.

Thanks @markjessell - that it what I suspected was happening. And @flohorovicic - I'm not trying to run a geophysical forward model, and am running pynoddy in BLOCK mode. I think this is just the "default" setup that is saved when using the Noddy GUI (which must automatically trim the model somehow before displaying it?).

Maybe we should add something to pynoddy that sets GPSRange to 0 unless 'geophys' in sim_type.lower() or 'all' in sim_type.lower()? Either that or it automatically trims the padding off the block - I find it counterintuitive that a model built using the GUI gives different results when run in pynoddy.

from pynoddy.

Problem resolved

from pynoddy.