Start by loading some boiler plate: matplotlib, numpy, scipy, json, functools, and a convenience class.
In [1]:
%matplotlib inline
import matplotlib
matplotlib.rcParams['figure.figsize'] = (10.0, 16.0)
import matplotlib.pyplot as plt
import numpy as np
from scipy.interpolate import interp1d, InterpolatedUnivariateSpline
from scipy.optimize import bisect
import json
from functools import partial
class Foo: pass
And some more specialized dependencies:
Slict provides a convenient slice-able dictionary interfaceChest is an out-of-core dictionary that we'll hook directly to a globus remote using...glopen is an open-like context manager for remote globus files
In [2]:
from chest import Chest
from slict import CachedSlict
from glopen import glopen, glopen_many
Configuration for this figure.
In [3]:
config = Foo()
config.name = "Schmidt-Projection/Nu01D01/Nu01D01"
#config.arch_end = "maxhutch#alpha-admin/pub"
config.arch_end = "alcf#dtn_mira/projects/alpha-nek/"
config.frame = 1
config.lower = .25
config.upper = .75
Open a chest located on a remote globus endpoint and load a remote json configuration file.
In [4]:
c = Chest(path = "{:s}-results".format(config.name),
open = partial(glopen, endpoint=config.arch_end),
open_many = partial(glopen_many, endpoint=config.arch_end))
sc = CachedSlict(c)
with glopen(
"{:s}.json".format(config.name), mode='r',
endpoint = config.arch_end,
) as f:
params = json.load(f)
We want to grab all the data for the selected frame.
In [5]:
T = sc[:,'frame'].keys()[config.frame]
frame = sc[T,:]
c.prefetch(frame.full_keys())
Plot the bubble height, the 'H' keys, vs. time. Use a spline to compute the derivative.
In [6]:
fig = plt.figure()
nx = frame['t_yz'].shape[0]
ny = frame['t_yz'].shape[1]
plt.imshow(frame['vorticity_xy'].transpose(), origin='lower')
plt.colorbar();
In [7]:
fig = plt.figure()
nx = frame['t_yz'].shape[0]
ny = frame['t_yz'].shape[1]
plt.imshow(frame['t_xy'].transpose(), origin='lower')
plt.colorbar();
In [8]:
def laplace(grid):
from numpy.fft import fftn, ifftn, fftfreq
kx = fftfreq(grid.shape[0])
ky = fftfreq(grid.shape[1])
kgrid = fftn(grid)
for i in range(kgrid.shape[0]):
for j in range(kgrid.shape[1]):
if kx[i] != 0 or ky[j] != 0:
kgrid[i,j] = kgrid[i,j] / (kx[i]**2 + ky[j]**2)
kgrid[0,0] = 0.
rgrid = np.array(ifftn(kgrid), dtype=float)
return rgrid
In [9]:
fig = plt.figure()
nx = frame['t_yz'].shape[0]
ny = frame['t_yz'].shape[1]
plt.imshow(frame['vorticity_yz'][:,int(ny*config.lower):int(ny*config.upper)].transpose(),
origin='lower')
#axs[1].imshow(frame['fz_yz'].transpose(), origin='lower')
plt.colorbar();
In [10]:
streamfunction = laplace(frame['vorticity_yz'][:,:]);
In [11]:
fig = plt.figure()
nx = frame['t_yz'].shape[0]
ny = frame['t_yz'].shape[1]
print(np.max(streamfunction), np.min(streamfunction))
#background = np.tile(frame['t_yz'][:,int(ny*config.lower):int(ny*config.upper)].transpose(), (1,4))
background = np.tile(streamfunction[:,int(ny*config.lower):int(ny*config.upper)].transpose(), (1,4))
foreground = np.tile(streamfunction[:,int(ny*config.lower):int(ny*config.upper)].transpose(), (1,4))
plt.imshow(background,
origin='lower', alpha=0.1)
plt.colorbar();
plt.contour(foreground,
origin='lower', colors='k')
#axs[1].imshow(frame['fz_yz'].transpose(), origin='lower')
Out[11]:
In [12]:
%install_ext http://raw.github.com/jrjohansson/version_information/master/version_information.py
%load_ext version_information
%version_information numpy, matplotlib, slict, chest, glopen, globussh
Out[12]: