matplotlib等高線圖：對數色標的比例色條水平

Question

是否可以在對數刻度中設置顏色條的級別，如下圖所示？

以下是可以實現的示例代碼：

import matplotlib.pyplot as plt
import numpy as np
from matplotlib.colors import LogNorm
delta = 0.025

x = y = np.arange(0, 3.01, delta)
X, Y = np.meshgrid(x, y)
Z1 = plt.mlab.bivariate_normal(X, Y, 1.0, 1.0, 0.0, 0.0)
Z2 = plt.mlab.bivariate_normal(X, Y, 1.5, 0.5, 1, 1)
Z = 1e6 * (Z1* Z2)

fig=plt.figure()
ax1 = fig.add_subplot(111)
lvls = np.logspace(0,4,20)
CF = ax1.contourf(X,Y,Z,
         norm = LogNorm(),
         levels = lvls
        )
CS = ax1.contour(X,Y,Z,
         norm = LogNorm(),
         colors = 'k',
         levels = lvls
        )
cbar = plt.colorbar(CF, ticks=lvls, format='%.4f')
plt.show()

我在Windows 7上使用python 2.7.3和matplotlib 1.1.1。

Answer 1

我建議按如下方式生成偽色條（請參閱注釋說明）：

import matplotlib.pyplot as plt
import numpy as np
from matplotlib.colors import LogNorm
import matplotlib.gridspec as gridspec

delta = 0.025

x = y = np.arange(0, 3.01, delta)
X, Y = np.meshgrid(x, y)
Z1 = plt.mlab.bivariate_normal(X, Y, 1.0, 1.0, 0.0, 0.0)
Z2 = plt.mlab.bivariate_normal(X, Y, 1.5, 0.5, 1, 1)
Z = 1e6 * (Z1 * Z2)

fig=plt.figure()

#
# define 2 subplots, using gridspec to control the 
# width ratios:
#
# note: you have to import matplotlib.gridspec for this
#
gs = gridspec.GridSpec(1, 2,width_ratios=[15,1])

# the 1st subplot
ax1 = plt.subplot(gs[0])

lvls = np.logspace(0,4,20)

CF = ax1.contourf(X,Y,Z,
                  norm = LogNorm(),
                  levels = lvls
                 )
CS = ax1.contour(X,Y,Z,
                 norm = LogNorm(),
                 colors = 'k',
                 levels = lvls
                )

#
# the pseudo-colorbar
#

# the 2nd subplot
ax2 = plt.subplot(gs[1])        

#
# new levels!
#
# np.logspace gives you logarithmically spaced levels - 
# this, however, is not what you want in your colorbar
#
# you want equally spaced labels for each exponential group:
#
levls = np.linspace(1,10,10)
levls = np.concatenate((levls[:-1],np.linspace(10,100,10)))
levls = np.concatenate((levls[:-1],np.linspace(100,1000,10)))
levls = np.concatenate((levls[:-1],np.linspace(1000,10000,10)))

#
# simple x,y setup for a contourf plot to serve as colorbar
#
XC = [np.zeros(len(levls)), np.ones(len(levls))]
YC = [levls, levls]
CM = ax2.contourf(XC,YC,YC, levels=levls, norm = LogNorm())
# log y-scale
ax2.set_yscale('log')  
# y-labels on the right
ax2.yaxis.tick_right()
# no x-ticks
ax2.set_xticks([])

plt.show()

這會給你一個這樣的情節：

編輯

或者，在調用colorbar時使用類似新級別和spacing='proportional'選項的內容：

1. 替換這一行：

    lvls = np.logspace(0,4,20) 

   用這些：

    lvls = np.linspace(1,10,5) lvls = np.concatenate((lvls[:-1],np.linspace(10,100,5))) lvls = np.concatenate((lvls[:-1],np.linspace(100,1000,5))) lvls = np.concatenate((lvls[:-1],np.linspace(1000,10000,5))) 

2. 替換這一行：

    cbar = plt.colorbar(CF, ticks=lvls, format='%.4f') 

   有了這個：

    cbar = plt.colorbar(CF, ticks=lvls, format='%.2f', spacing='proportional') 

你將最終得到這個情節：

（ format只是改變了，因為新的刻度不需要4位小數）

編輯2
如果你想自動生成像我使用的那樣的級別，你可以考慮這段代碼：

levels = []
LAST_EXP = 4
N_LEVELS = 5
for E in range(0,LAST_EXP):
    levels = np.concatenate((levels[:-1],np.linspace(10**E,10**(E+1),N_LEVELS)))