这是我使用的一个庞大代码的MWE
示例。基本上,它对位于某个阈值以下的所有值执行了一个KDE(核密度估计)的Monte Carlo积分(该积分方法建议在此问题中使用:Integrate 2D kernel density estimate)。
import numpy as np
from scipy import stats
import time
# Generate some random two-dimensional data:
def measure(n):
"Measurement model, return two coupled measurements."
m1 = np.random.normal(size=n)
m2 = np.random.normal(scale=0.5, size=n)
return m1+m2, m1-m2
# Get data.
m1, m2 = measure(20000)
# Define limits.
xmin = m1.min()
xmax = m1.max()
ymin = m2.min()
ymax = m2.max()
# Perform a kernel density estimate on the data.
x, y = np.mgrid[xmin:xmax:100j, ymin:ymax:100j]
values = np.vstack([m1, m2])
kernel = stats.gaussian_kde(values)
# Define point below which to integrate the kernel.
x1, y1 = 0.5, 0.5
# Get kernel value for this point.
tik = time.time()
iso = kernel((x1,y1))
print 'iso: ', time.time()-tik
# Sample from KDE distribution (Monte Carlo process).
tik = time.time()
sample = kernel.resample(size=1000)
print 'resample: ', time.time()-tik
# Filter the sample leaving only values for which
# the kernel evaluates to less than what it does for
# the (x1, y1) point defined above.
tik = time.time()
insample = kernel(sample) < iso
print 'filter/sample: ', time.time()-tik
# Integrate for all values below iso.
tik = time.time()
integral = insample.sum() / float(insample.shape[0])
print 'integral: ', time.time()-tik
输出结果看起来像这样:
iso: 0.00259208679199
resample: 0.000817060470581
filter/sample: 2.10829401016
integral: 4.2200088501e-05
这句话的意思是,很明显过滤器/样本调用几乎占用了代码运行所需的所有时间。我必须迭代地运行这个代码块数千次,所以它可能非常耗时。
有没有办法加快过滤/采样过程的速度?
添加
这是一个稍微更加真实的MWE
,其中包含了Ophion的多线程解决方案:
import numpy as np
from scipy import stats
from multiprocessing import Pool
def kde_integration(m_list):
m1, m2 = [], []
for item in m_list:
# Color data.
m1.append(item[0])
# Magnitude data.
m2.append(item[1])
# Define limits.
xmin, xmax = min(m1), max(m1)
ymin, ymax = min(m2), max(m2)
# Perform a kernel density estimate on the data:
x, y = np.mgrid[xmin:xmax:100j, ymin:ymax:100j]
values = np.vstack([m1, m2])
kernel = stats.gaussian_kde(values)
out_list = []
for point in m_list:
# Compute the point below which to integrate.
iso = kernel((point[0], point[1]))
# Sample KDE distribution
sample = kernel.resample(size=1000)
#Create definition.
def calc_kernel(samp):
return kernel(samp)
#Choose number of cores and split input array.
cores = 4
torun = np.array_split(sample, cores, axis=1)
#Calculate
pool = Pool(processes=cores)
results = pool.map(calc_kernel, torun)
#Reintegrate and calculate results
insample_mp = np.concatenate(results) < iso
# Integrate for all values below iso.
integral = insample_mp.sum() / float(insample_mp.shape[0])
out_list.append(integral)
return out_list
# Generate some random two-dimensional data:
def measure(n):
"Measurement model, return two coupled measurements."
m1 = np.random.normal(size=n)
m2 = np.random.normal(scale=0.5, size=n)
return m1+m2, m1-m2
# Create list to pass.
m_list = []
for i in range(60):
m1, m2 = measure(5)
m_list.append(m1.tolist())
m_list.append(m2.tolist())
# Call KDE integration function.
print 'Integral result: ', kde_integration(m_list)
Ophion 提供的解决方案对我提供的原始代码非常有效,但在此版本中失败并显示以下错误:
Integral result: Exception in thread Thread-3:
Traceback (most recent call last):
File "/usr/lib/python2.7/threading.py", line 551, in __bootstrap_inner
self.run()
File "/usr/lib/python2.7/threading.py", line 504, in run
self.__target(*self.__args, **self.__kwargs)
File "/usr/lib/python2.7/multiprocessing/pool.py", line 319, in _handle_tasks
put(task)
PicklingError: Can't pickle <type 'function'>: attribute lookup __builtin__.function failed
我尝试移动
calc_kernel
函数,因为这个问题的一个答案Multiprocessing: How to use Pool.map on a function defined in a class?中指出:"您提供给map()的函数必须通过导入您的模块来访问";但我仍然无法使此代码工作。
非常感谢任何帮助。
加2
实现 Ophion 的建议,移除 calc_kernel
函数并直接使用:
results = pool.map(kernel, torun)
我的工作是为了解决PicklingError
,但现在我发现,如果我创建一个初始的m_list
,其中包含超过62-63个项目,我会遇到这个错误:
Traceback (most recent call last):
File "~/gauss_kde_temp.py", line 67, in <module>
print 'Integral result: ', kde_integration(m_list)
File "~/gauss_kde_temp.py", line 38, in kde_integration
pool = Pool(processes=cores)
File "/usr/lib/python2.7/multiprocessing/__init__.py", line 232, in Pool
return Pool(processes, initializer, initargs, maxtasksperchild)
File "/usr/lib/python2.7/multiprocessing/pool.py", line 161, in __init__
self._result_handler.start()
File "/usr/lib/python2.7/threading.py", line 494, in start
_start_new_thread(self.__bootstrap, ())
thread.error: can't start new thread
由于我实际代码中的列表最多可以有2000个项目,所以这个问题使得代码无法使用。第38
行是这一行:
pool = Pool(processes=cores)
我可以帮助你翻译。这段文字是关于编程的内容。它询问了与使用的核心数量有关的问题。下面是需要翻译的内容:
所以显然这与我使用的核心数量有关?
这个问题 "Python中无法启动新线程错误" 建议使用:
threading.active_count()
当我遇到错误时,我希望你能帮我检查正在运行的线程数。我已经检查过,每当达到 374
个线程时程序就会崩溃。如何在编码时解决这个问题?
这是与最后一个问题有关的新问题: 线程错误:无法启动新线程
kernel(sample) < iso
总共需要 2 秒钟,其中kernel(sample)
占用了 99.99% 的时间。 - Danielscipy/stats/kde.py
中。看起来主要步骤是将逆协方差乘以数据点上的度量。你可以尝试在Cython中重新实现这个方法(evaluate
),但如果np.dot
真的是瓶颈,那可能不会有太大帮助。或者,将数据矩阵分成较小的块,并在这些较小的块上调用kernel(block)
。 - lmjohns3