Remove stale Dask usage and update example imports

OceanLab/eof.py

@@ -1,9 +1,7 @@
 import numpy as np
 import scipy.linalg as la
-from dask import delayed
 from scipy.signal import hilbert
 import xarray as xr
-from dask.distributed import Client, LocalCluster
 
 # functions
 #=========================================
@@ -130,23 +128,21 @@ def my_eof_interp(M,nmodes,errmin=1e-15,repmax=None):
 #=========================================
 # PERFORM COMPLEX EOF
 #=========================================
-def ceof(lon, lat, data, nkp = 10, parallel = True):
+def ceof(lon, lat, data, nkp = 10):
     ''' Complex (Hilbert) EOF to detect propagating features: waves, meanders, etc.
     Note: the mean field in each coordinate is subtracted within the function.
     Do not subtract the time-mean field before inputing.
-    NaN values are removed in the algorithm. 
+    NaN values are removed in the algorithm.
     The user can input the data as it is.
-    
-    First written in MATLAB and found in Prof. Daniel J. Vimont webpage 
+
+    First written in MATLAB and found in Prof. Daniel J. Vimont webpage
     (https://www.aos.wisc.edu/~dvimont/matlab/Stat_Tools/complex_eof.html)
     ==============================================================================
     INPUT:
        lon      = longitudes (array)
        lat      = latitude (array)
        data     = original data set [time, lat, lon]
        nkp      = number of modes to return (default = 10)
-       parallel = create a standard client kernel for parallel computing
-                  [switch parallel to False, in case you created your own client]
 
     OUTPUT:
        The variables below return inside a DataArray.
@@ -158,12 +154,7 @@ def ceof(lon, lat, data, nkp = 10, parallel = True):
        TAmp     = temporal amplitude [time, nkp]
        TPhase   = temporal phase [time, nkp]
     ==============================================================================
-    ''' 
-    # Configure client for parallel computing
-    if parallel:
-        cluster = LocalCluster()
-        client  = Client(cluster)
-
+    '''
     # Organizing the data as time vs space
     data_ceof = _org_data_ceof(lon, lat, data)
     # We need to remove the mean field (i.e., the trend) in each coordinate to 
@@ -184,9 +175,9 @@ def ceof(lon, lat, data, nkp = 10, parallel = True):
     data_hilbert = hilbert(data_ceof)
     # Compute the covariance matrix in the Hilbert transform
     print('2: Computing covariance matrix')
-    c = delayed(np.dot)(data_hilbert.conjugate().T, data_hilbert).compute()/ntim
+    c = np.dot(data_hilbert.conjugate().T, data_hilbert)/ntim
     print('3: Solving the eigenvalue problem')
-    lamda, loadings = delayed(la.eig)(c).compute() # lamda: eigenvalue, loadings: eigenvectors
+    lamda, loadings = la.eig(c) # lamda: eigenvalue, loadings: eigenvectors
 
     l = lamda.conjugate().T
     k = np.argsort(l)

OceanLab/utils.py

@@ -2,8 +2,6 @@
 import scipy.signal as sg
 import xarray as xr
 
-from dask.distributed import Client
-
 ##### User functions
 #=============================================================================
 # NEAREST DISTANCE
@@ -133,29 +131,4 @@ def timefilter(prop,filtdays=60):
     return m_prop,p_prop
 #=============================================================================
 
-##### Functions for relative imports
-# =============================================================================
-# KERNEL FOR PARALLEL COMPUTING
-# =============================================================================
-def _parallel_client(cpu_params=dict(tpw=2,nw=4,ml=7.5)):
-    """
-    Create client kernel for parallel computing
-    ====================================================
-    INPUT:
-        -> cpu_params: dict containing floats with keys
-            -> tpw: threads_per_worker
-            -> nw: n_workers
-            -> ml: memory_limit per worker [GB]
-    OUTPUT:
-        -> client: configuration of parallel computing
-    ====================================================
-    """
-
-    client = Client(threads_per_worker=cpu_params['tpw'], 
-                    n_workers=cpu_params['nw'], 
-                    memory_limit=str(cpu_params['ml'])+'GB')
-    return client
-#=============================================================================
-
-
 

examples/DYN_EOF.ipynb

@@ -2,16 +2,10 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
-   "source": [
-    "import numpy as np\n",
-    "import matplotlib.pyplot as plt\n",
-    "from sympy import *\n",
-    "from OceanLab import EOF,DYN\n",
-    "import seawater as sw"
-   ]
+   "source": "import numpy as np\nimport matplotlib.pyplot as plt\nfrom sympy import *\nimport OceanLab.eof as EOF\nimport OceanLab.dyn as DYN\nimport seawater as sw"
   },
   {
    "cell_type": "code",
@@ -447,4 +441,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 4
-}
+}

examples/OA.ipynb

@@ -2,15 +2,10 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 102,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
-   "source": [
-    "import numpy as np\n",
-    "import matplotlib.pyplot as plt\n",
-    "from sympy import *\n",
-    "from OceanLab import OA"
-   ]
+   "source": "import numpy as np\nimport matplotlib.pyplot as plt\nfrom sympy import *\nimport OceanLab.oa as OA"
   },
   {
    "cell_type": "code",
@@ -168,4 +163,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 4
-}
+}

requirements.txt

@@ -1,6 +1,4 @@
-numpy >= 1.8.2
 seawater >= 3.3.1
+numpy >= 1.8.2
 scipy >= 1.6.3
 xarray >= 0.18.2
-dask >= 2021.06.0
-dask[distributed] >= 2021.06.0

setup.py

@@ -28,11 +28,9 @@ def read(file):
         "Operating System :: OS Independent",
     ],
     install_requires = [
-        'seawater ~= 3.3',
-        'numpy ~= 1.18',
-        'scipy ~= 1.6',
-        'xarray ~= 0.18',
-        'dask ~= 2021.06',
-        'dask[distributed] ~= 2021.06'
+        'seawater >= 3.3',
+        'numpy >= 1.8.2',
+        'scipy >= 1.6',
+        'xarray >= 0.18',
     ],
 )