search

eepeterson / pleiades

PLasma Experiment Iteratively Advanced DEsign and Simulation
https://pleiades.readthedocs.io/en/latest/
MIT License
5 stars 6 forks source link

UnivariateSpline not happy #8

Open jcwright77 opened 4 years ago

jcwright77 commented 4 years ago

in the wham jupyter example, running write_eqdsk(R,Z,psi,plas_currents,"wham_eqdsk.txt","/content/WHAM_Mirror Equilib") produces this error:

ValueError Traceback (most recent call last)

in () ----> 1 write_eqdsk(R,Z,psi,plas_currents,"wham_eqdsk.txt","/content/WHAM_Mirror Equilib") 1 frames /usr/local/lib/python3.6/dist-packages/scipy/interpolate/fitpack2.py in __init__(self, x, y, w, bbox, k, s, ext, check_finite) 180 else: 181 if not np.all(diff(x) > 0.0): --> 182 raise ValueError("x must be strictly increasing if s = 0") 183 184 # _data == x,y,w,xb,xe,k,s,n,t,c,fp,fpint,nrdata,ier ValueError: x must be strictly increasing if s = 0 The plot shows indeed that z is not monotonic. The problem is lines 50/51 in write_eqdsk() spl = UnivariateSpline(z,r,k=1,s=0) fl_spl = UnivariateSpline(fl_dist,z,k=1)#,s=0) Removing s=0 to use the default gets rid of the error. scipy.__version__ 1.4.1
jcwright77 commented 4 years ago

see https://drive.google.com/file/d/1JcFNsz5Ao0NO7ABCphu8QQle1WVMqdar/view?usp=sharing for a colaboratory notebook demonstrating the error and fix. Error is generated first time through, then monkey patch fixes error

eepeterson commented 4 years ago

@jcwright77 Almost everything in that function was created just to set the "vessel" locations to have the same shape as the flux surface and I would be happy if we got rid of it entirely since it's very finicky. I'm not quite sure what the R, Z pairs representing the vessel are for in an eqdsk, but it's my assumption that they're for vessel currents. We could get rid of this in favor of a user supplied vessel shape and a default behavior of finding the bounding box of the last closed flux surface. If WHAM has a 2D vacuum vessel profile spec'd out already we could implement that instead of relying on this shoddy code I wrote a while ago. Thoughts from @Jon-Pizzo would be appreciated too.

Jon-Pizzo commented 4 years ago

I don't believe there is a 2D vacuum vessel specified in the eqdsk file, however I do have the coordinates for it as it was used in genray input files. I can email you a text file containing the R,Z pairs.