<div dir="ltr">I think it is unlikely to be possible (or even desirable?) to completely emulate the behaviour of numpy arrays in flex arrays. The underlying data structure and behaviour of numpy arrays is significantly different to flex arrays. For example, when you do a slice in numpy you don't get a copy, but a 'view' - i.e. you get a new object, but that object points to the same underlying data in memory as the original array:<div>
<br></div><div><div><div><a href="http://www.scipy.org/Tentative_NumPy_Tutorial#head-4306be2ef34ad134f9fa0b87d426fa711119c167">http://www.scipy.org/Tentative_NumPy_Tutorial#head-4306be2ef34ad134f9fa0b87d426fa711119c167</a><br>
</div></div></div><div><br></div><div style>When I implemented the multidimensional slicing for flex arrays, I only did so because I required that functionality in code I was developing, not for the fun or challenge of emulating numpy arrays. I am unlikely to have time to add numpy-style assignment to slices in the near future, but if James, or indeed anyone else, requires and is willing to implement such behaviour, I'm sure Nat can grant you svn commit access in order to make the changes you require (providing that the changes play well with the rest of the cctbx code and Phenix, etc.).</div>
<div style><br></div><div style>Cheers,</div><div style><br></div><div style>Richard</div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On 11 April 2013 10:08, James Stroud <span dir="ltr"><<a href="mailto:xtald00d@gmail.com" target="_blank">xtald00d@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div style="word-wrap:break-word"><div class="im"><br><div><div>On Apr 11, 2013, at 9:18 AM, Luc Bourhis wrote:</div><blockquote type="cite">
<div>On 11 Apr 2013, at 09:49, James Stroud wrote:<br><br><blockquote type="cite">It seems that flex arrays do not support slice assignment:<br></blockquote><br>it would not be too difficult to code 1-dimensional slice assignment �but ...<br>
<br>On 11 Apr 2013, at 10:24, <a href="mailto:Graeme.Winter@Diamond.ac.uk" target="_blank">Graeme.Winter@Diamond.ac.uk</a> wrote:<br><br><blockquote type="cite">[...] <br></blockquote><blockquote type="cite"> �����������array[k,:,:] = image.as_numpy_array()[y0:y1, x0:x1]<br>
</blockquote><br>that's way more difficult!<br></div></blockquote></div><div><br></div><br></div><div>In python this type of magic is fairly straightforward because the slice is just a tuple:</div><div><br></div><blockquote style="margin:0 0 0 40px;border:none;padding:0px">
<div><div><font face="Monaco">py> class Example(object):</font></div></div><div><div><font face="Monaco">... � def __setitem__(self, i, v):</font></div></div><div><div><font face="Monaco">... � � print "setting %s to %s" % (i, v)</font></div>
</div><div><div><font face="Monaco">...</font></div></div><div><div><font face="Monaco">py> e = Example()</font></div></div><div><div><font face="Monaco">py> e[5:6, 4, 8:9, "bob"] = 2</font></div></div><div>
<div><font face="Monaco">setting (slice(5, 6, None), 4, slice(8, 9, None), 'bob') to 2</font></div></div></blockquote><div><br></div><div><br></div><div>This should actually be the case on the C side, too, not that it wouldn't take a few lines of code to build an iterator based on the slice.</div>
<div><br></div><div>But as a way to circumvent all of this magic with slices (which would be very cool, by the way), it might be easiest for the programmer (and somewhat intuitive for those familiar with flex) just to add a signature to set_selected() that uses a flex.grid to represent a slice because flex.grid already encapsulates the most useful aspects of slicing semantics.</div>
<div><br></div><div><br></div><div>Here is a prototype for the suggested behavior in python using ndarray as a back-end:</div><div><br></div><blockquote style="margin:0 0 0 40px;border:none;padding:0px"><div><div><font face="Monaco">class Prototype(object):</font></div>
</div><div><div><font face="Monaco">� def __init__(self, data):</font></div></div><div><div><font face="Monaco">� � self.data = numpy.array(data)</font></div></div><div><div><font face="Monaco">� def __getitem__(self, i):</font></div>
</div><div><div><font face="Monaco">� � if isinstance (i, flex.grid):</font></div></div><div><div><font face="Monaco">� � � slices = [slice(*t) for t in zip(i.origin(), i.last())]</font></div></div><div><div><font face="Monaco">� � � return self.__class__(self.data[slices])</font></div>
</div><div><div><font face="Monaco">� � else:</font></div></div><div><div><font face="Monaco">� � � return self.__class__(self.data[i])</font></div></div><div><div><font face="Monaco">� def __setitem__(self, i, v):</font></div>
</div><div><div><font face="Monaco">� � if isinstance (i, flex.grid):</font></div></div><div><div><font face="Monaco">� � � slices = [slice(*t) for t in zip(i.origin(), i.last())]</font></div></div><div><div><font face="Monaco">� � � self.data[slices] = v</font></div>
</div><div><div><font face="Monaco">� � else:</font></div></div><div><div><font face="Monaco">� � � self.data[i] = v</font></div></div><div><div><font face="Monaco">� def __repr__(self):</font></div></div><div><div><font face="Monaco">� � return repr(self.data)</font></div>
</div></blockquote><div><br></div><div><br></div><div>And here it is in action:</div><div><br></div><blockquote style="margin:0 0 0 40px;border:none;padding:0px"><div><div><font face="Monaco">py> grd = flex.grid((1, 1), (3, 3)) �# <== selects [1:3, 1:3]</font></div>
</div><div><div><font face="Monaco">py> ary = Prototype(numpy.arange(25).reshape((5,5)))</font></div></div><div><div><font face="Monaco">py> ary</font></div></div><div><div><font face="Monaco">array([[ 0, �1, �2, �3, �4],</font></div>
</div><div><div><font face="Monaco">� � � �[ 5, �6, �7, �8, �9],</font></div></div><div><div><font face="Monaco">� � � �[10, 11, 12, 13, 14],</font></div></div><div><div><font face="Monaco">� � � �[15, 16, 17, 18, 19],</font></div>
</div><div><div><font face="Monaco">� � � �[20, 21, 22, 23, 24]])</font></div></div><div><div><font face="Monaco">py> ary[grd]</font></div></div><div><div><font face="Monaco">array([[ 6, �7],</font></div></div><div><div>
<font face="Monaco">� � � �[11, 12]])</font></div></div><div><div><font face="Monaco">py> ary[grd] = [[41, 42], [43, 44]]</font></div></div><div><div><font face="Monaco">py> ary[grd]</font></div></div><div><div><font face="Monaco">array([[41, 42],</font></div>
</div><div><div><font face="Monaco">� � � �[43, 44]])</font></div></div><div><div><font face="Monaco">py> ary</font></div></div><div><div><font face="Monaco">array([[ 0, �1, �2, �3, �4],</font></div></div><div><div><font face="Monaco">� � � �[ 5, 41, 42, �8, �9],</font></div>
</div><div><div><font face="Monaco">� � � �[10, 43, 44, 13, 14],</font></div></div><div><div><font face="Monaco">� � � �[15, 16, 17, 18, 19],</font></div></div><div><div><font face="Monaco">� � � �[20, 21, 22, 23, 24]])</font></div>
</div></blockquote><span class="HOEnZb"><font color="#888888"><div><font face="Monaco"><br></font></div><div>James</div></font></span></div><br>_______________________________________________<br>
cctbxbb mailing list<br>
<a href="mailto:cctbxbb@phenix-online.org">cctbxbb@phenix-online.org</a><br>
<a href="http://phenix-online.org/mailman/listinfo/cctbxbb" target="_blank">http://phenix-online.org/mailman/listinfo/cctbxbb</a><br>
<br></blockquote></div><br></div>