2+3

mylist = ['a', 'b', 'c', 'd'] 
mylist

mylist[0]  # 0-based index

mydict = dict(a=1, b=2) 
mydict

mydict['b']  # access element by key

sum([1, 2, 3, 4])

exp(0)

import numpy

numpy.exp(0)  

for i in range(5):
    print i

if 4 > 2:
    print 'no parenthesis'
else:
    print 'nope'

def myfunc(a, b=2):
    """ my documentation
    """
    return a+b

# functions are a particular type of variables
myfunc

myfunc(3)

myfunc(3, b=5)

class Ocean:  
    T = 25        # ocean temperature
    cp = 4000  # sea water heat capacity
    V = 1e6  # volume

indian = Ocean ()
indian

indian.T

class Ocean2:  
    T = 25        # ocean temperature
    cp = 4000  # sea water heat capacity
    V = 1e6  # volume
    
    def energy(self):  # class method, like a function defined within a class
        return self.cp*self.T*self.V

indian = Ocean2()
indian.energy()

class Ocean3:  
    T = 25        # ocean temperature
    cp = 4000  # sea water heat capacity
    V = 1e6  # volume
    
    def energy(self):  # class method, like a function defined within a class
        return self.cp*self.T*self.V
    
    def __add__(self, other):
        res = Ocean3()
        res.T = 0.5*(self.T + other.T)
        res.cp = 0.5*(self.cp+other.cp) 
        res.V = self.V+other.V
        return res

indian = Ocean3()
big = indian + indian

big.energy()

%pylab

array

a = array([2, 5, 7, 8, 3])  # define an array (same as np.array)
a

a[:3]  # first three elements

a[-3:] # last three (negative indexing)

a[::2] # every second element

array([1,2,3]) * 2

[1, 2, 3] * 2

arange(5)  # 0...4 : five elements numbered from 0, same as r_[:5]

arange(1900, 2000, 5)  # alternative syntax: r_[1900:2000:5]

linspace(-1, 1, 10)  # 10 elements between -1 and 1

zeros(5)

a = ones((2,3))
a

a.ndim  # number of dimensions

a.shape

a.size

a.mean    #  a method, itself an object

a.mean()

mean(a)

a = arange(2*3).reshape(2,3)
a

a[1,0]  # multi-dimensional indexing: first index for line, second for columns

a[1]   # grab the whole line

a[:, -1] # last column (negative integer start from the end)

a.mean(axis=0)  # mean over lines

a.mean(axis=1) # mean over columns

#%pylab
%matplotlib inline 

x = arange(10) 
y = x**2

plot(x, y, label='my line')
plot(x, y*0.5, label='my other line')

legend(loc='upper left')
xlabel('x axis')
ylabel('y axis')
title('My Figure')

import dimarray as da

data = [[1.,2,3], [4,5,6]]
a = da.DimArray(data, axes=[['grl', 'ant'], [1950, 1960, 1970]], dims=['item', 'time']) 
a

a.values

a.axes

ax = a.axes[1]
ax.name , ax.values

a.dims   # grab axis names (the dimensions)

a.labels   # grab axis values

a['grl', 1970]

a.ix[0, -1]

a.mean(axis='time')

a['ant', 1950] = nan
a

a.mean(axis='time', skipna=True)

import os 
datadir = da.get_datadir()
os.chdir(datadir)

ls

datadir = da.get_datadir()
ds = da.read_nc('cmip5.MPI-ESM-MR.nc')
ds

temp = ds['temp']
temp

%matplotlib inline
temp.plot()  

temp2 = da.read_nc('cmip5.IPSL-CM5A-LR.nc', 'temp')
temp2

ref1 = temp[1986:2005].mean(axis='time')
ref1

ano1 = temp - ref1
ano1

ano2 = temp2 - temp2[1986:2005].mean(axis='time') # same as for ano1 

ano1.shape, ano2.shape

difference = ano2 - ano1
difference

diffvalid = difference.dropna(axis='scenario', minvalid=1)
diffvalid

#difference.plot()
diffvalid.plot()

newa = da.stack((temp, temp2), axis='model', align=True, keys=['MPI-ESM-MR', 'IPSL-CM5A-LR'])
newa

newa.name = 'temperature'
newa.date = '22nd May 2014'
newa.description = 'my dimarray test'
newa._metadata

newa.write_nc('/tmp/mydata.nc')

ds = da.read_nc('/tmp/mydata.nc')
ds
ds['temperature']._metadata

ds = da.read_nc('cmip5.*.nc', align=True, axis='model')
ds

a = ds.to_array(axis='variable')
a
#a['ohu',:,2100, 'rcp45']

# rename model axis

def rename_axis(model_ax):
    return [ax.replace('data/cmip5.','') for ax in model_ax]    

a.model[:] = rename_axis(a.model)
a

change = a[:,:, 2081:2099].mean('time') - a[:,:,1986:2005].mean('time')
change

#rounded = change['temp'].apply(lambda x: np.round(x, 2)) # roudn to 2 decimals
#rounded.to_pandas()
change['temp'].to_pandas()

#rounded = change['temp'].apply(lambda x: np.round(x, 2)) # roudn to 2 decimals
#rounded.to_pandas()
grouped = change[:,:,['rcp26','rcp85']].group('model','scenario')
grouped

grouped.to_pandas()

grouped.ungroup()

change.reshape('variable,scenario','model')

lon = linspace(-179.5, 179.5, 360)
lat = linspace(-89.5, 89.5, 180)
LON, LAT = meshgrid(lon, lat)
values = cos(radians(LON)) + cos(radians(LAT))
basic = da.DimArray(values, axes=[lat, lon], dims=['lat', 'lon'])
basic.contourf(); colorbar()

basic.mean()

from dimarray.geo import GeoArray
geo = GeoArray(basic)

geo.mean()

geo.axes['lat'].weights

geo.axes['lat'].weights(geo.lat)

geo.axes['lon'].modulo

geo[:,[280.5]]