<p>显然,当你打电话的时候</p>
<pre><code>np.arange(summertime.min(), summertime.max(), 10)
</code></pre>
<p>该值的单位<code>10</code>是<code>datetime</code>对象的最小单位,即<em>微秒</em>。给定开始和结束日期,对<code>arange</code>的调用试图创建一个长度为198707200000的数组。这将需要超过1 TB的内存。你知道吗</p>
<p>你说“我试图用频率为10的datetime之间的x轴”,但你没有说10的单位是什么。10分钟?小时?几秒钟?你知道吗</p>
<p>假设是10个小时。可以将增量指定为微秒数,也可以指定为<code>datetime.timedelta</code>对象。例如</p>
<pre><code>In [43]: mn = datetime.datetime(2015, 7, 24, 0, 10, 40)
In [44]: mx = datetime.datetime(2015, 8, 16, 0, 8, 32)
In [45]: np.arange(mn, mx, 10*3600*1000000) # Increment is 10 hours, expressed in microseconds
Out[45]:
array(['2015-07-24T00:10:40.000000', '2015-07-24T10:10:40.000000',
'2015-07-24T20:10:40.000000', '2015-07-25T06:10:40.000000',
'2015-07-25T16:10:40.000000', '2015-07-26T02:10:40.000000',
'2015-07-26T12:10:40.000000', '2015-07-26T22:10:40.000000',
'2015-07-27T08:10:40.000000', '2015-07-27T18:10:40.000000',
'2015-07-28T04:10:40.000000', '2015-07-28T14:10:40.000000',
'2015-07-29T00:10:40.000000', '2015-07-29T10:10:40.000000',
'2015-07-29T20:10:40.000000', '2015-07-30T06:10:40.000000',
'2015-07-30T16:10:40.000000', '2015-07-31T02:10:40.000000',
'2015-07-31T12:10:40.000000', '2015-07-31T22:10:40.000000',
'2015-08-01T08:10:40.000000', '2015-08-01T18:10:40.000000',
'2015-08-02T04:10:40.000000', '2015-08-02T14:10:40.000000',
'2015-08-03T00:10:40.000000', '2015-08-03T10:10:40.000000',
'2015-08-03T20:10:40.000000', '2015-08-04T06:10:40.000000',
'2015-08-04T16:10:40.000000', '2015-08-05T02:10:40.000000',
'2015-08-05T12:10:40.000000', '2015-08-05T22:10:40.000000',
'2015-08-06T08:10:40.000000', '2015-08-06T18:10:40.000000',
'2015-08-07T04:10:40.000000', '2015-08-07T14:10:40.000000',
'2015-08-08T00:10:40.000000', '2015-08-08T10:10:40.000000',
'2015-08-08T20:10:40.000000', '2015-08-09T06:10:40.000000',
'2015-08-09T16:10:40.000000', '2015-08-10T02:10:40.000000',
'2015-08-10T12:10:40.000000', '2015-08-10T22:10:40.000000',
'2015-08-11T08:10:40.000000', '2015-08-11T18:10:40.000000',
'2015-08-12T04:10:40.000000', '2015-08-12T14:10:40.000000',
'2015-08-13T00:10:40.000000', '2015-08-13T10:10:40.000000',
'2015-08-13T20:10:40.000000', '2015-08-14T06:10:40.000000',
'2015-08-14T16:10:40.000000', '2015-08-15T02:10:40.000000',
'2015-08-15T12:10:40.000000', '2015-08-15T22:10:40.000000'], dtype='datetime64[us]')
In [46]: np.arange(mn, mx, datetime.timedelta(hours=10)) # Increment is 10 hours, expressed using a datetime.timedelta
Out[46]:
array(['2015-07-24T00:10:40.000000', '2015-07-24T10:10:40.000000',
'2015-07-24T20:10:40.000000', '2015-07-25T06:10:40.000000',
'2015-07-25T16:10:40.000000', '2015-07-26T02:10:40.000000',
'2015-07-26T12:10:40.000000', '2015-07-26T22:10:40.000000',
'2015-07-27T08:10:40.000000', '2015-07-27T18:10:40.000000',
'2015-07-28T04:10:40.000000', '2015-07-28T14:10:40.000000',
'2015-07-29T00:10:40.000000', '2015-07-29T10:10:40.000000',
'2015-07-29T20:10:40.000000', '2015-07-30T06:10:40.000000',
'2015-07-30T16:10:40.000000', '2015-07-31T02:10:40.000000',
'2015-07-31T12:10:40.000000', '2015-07-31T22:10:40.000000',
'2015-08-01T08:10:40.000000', '2015-08-01T18:10:40.000000',
'2015-08-02T04:10:40.000000', '2015-08-02T14:10:40.000000',
'2015-08-03T00:10:40.000000', '2015-08-03T10:10:40.000000',
'2015-08-03T20:10:40.000000', '2015-08-04T06:10:40.000000',
'2015-08-04T16:10:40.000000', '2015-08-05T02:10:40.000000',
'2015-08-05T12:10:40.000000', '2015-08-05T22:10:40.000000',
'2015-08-06T08:10:40.000000', '2015-08-06T18:10:40.000000',
'2015-08-07T04:10:40.000000', '2015-08-07T14:10:40.000000',
'2015-08-08T00:10:40.000000', '2015-08-08T10:10:40.000000',
'2015-08-08T20:10:40.000000', '2015-08-09T06:10:40.000000',
'2015-08-09T16:10:40.000000', '2015-08-10T02:10:40.000000',
'2015-08-10T12:10:40.000000', '2015-08-10T22:10:40.000000',
'2015-08-11T08:10:40.000000', '2015-08-11T18:10:40.000000',
'2015-08-12T04:10:40.000000', '2015-08-12T14:10:40.000000',
'2015-08-13T00:10:40.000000', '2015-08-13T10:10:40.000000',
'2015-08-13T20:10:40.000000', '2015-08-14T06:10:40.000000',
'2015-08-14T16:10:40.000000', '2015-08-15T02:10:40.000000',
'2015-08-15T12:10:40.000000', '2015-08-15T22:10:40.000000'], dtype='datetime64[us]')
</code></pre>