Actually it's https://github.com/JuliaLang/julia/pull/14798

On Wednesday, February 24, 2016 at 6:23:14 AM UTC-8, Kristoffer Carlsson 
wrote:
>
> https://github.com/JuliaLang/julia/pull/12608
>
> On Wednesday, February 24, 2016 at 3:21:59 PM UTC+1, Christoph Ortner 
> wrote:
>>
>> Can you post a link to that pull request?
>> Thanks, Christoph
>>
>> On Wednesday, 24 February 2016 04:26:35 UTC, Tony Kelman wrote:
>>>
>>> It is possible to store explicit zero values in sparse matrices, but not 
>>> via the sparse() function at the moment. You currently have to construct a 
>>> sparse matrix with the desired structure of stored elements, then modify 
>>> the field values of Mat.nzval to explicitly set some of them to zero. 
>>> There's an open pull request that will change this, I think the only thing 
>>> we were waiting for before merging was some performance benchmarking of 
>>> packages that heavily utilize the sparse() function.
>>>
>>>
>>> On Tuesday, February 23, 2016 at 2:58:44 PM UTC-8, Yankai Cao wrote:
>>>>
>>>> Hi, Stefan, 
>>>>
>>>> Thanks for your reply.  For my application, the structure of the matrix 
>>>> is fixed, but the values of some elements might become zero from iteration 
>>>> to iteration. Is it possible to keep the position of elements whose values 
>>>> is zero?
>>>> For example: 
>>>> I=[1;1;2]
>>>> J=[1;1;2]
>>>> V=[1.0;0.0;0.0]
>>>> Mat=sparse(I,J,V,2,2)
>>>>
>>>> what I get is 
>>>> [1,1]=1
>>>>
>>>> is it possible to get the following?
>>>> [ 1, 1] = 1.0
>>>>  [2, 2] = 0.0
>>>>
>>>>
>>>> Thanks. 
>>>>
>>>>
>>>>
>>>> Yankai
>>>>
>>>>
>>>>
>>>> On Tuesday, February 23, 2016 at 4:50:07 PM UTC-6, Stefan Karpinski 
>>>> wrote:
>>>>>
>>>>> You can't have two values at the same row and column, which is what 
>>>>> the first two lines of this output would indicate:
>>>>>
>>>>> [1, 1] = 1.0
>>>>> [1, 1] = 1.0
>>>>> [2, 2] = 1.0
>>>>>
>>>>>
>>>>> Since the indices (1,1) appear twice the associated values are added, 
>>>>> giving 2.0.
>>>>>
>>>>> On Tue, Feb 23, 2016 at 5:03 PM, Yankai Cao <[email protected]> 
>>>>> wrote:
>>>>>
>>>>>> I am new to Julia. I have 2 questions about constructting sparse 
>>>>>> matrix from arrays. 
>>>>>>
>>>>>> 1. 
>>>>>> I=[1;1;2]
>>>>>> J=[1;1;2]
>>>>>> V=[1.0;1.0;1.0]
>>>>>> Mat=sparse(I,J,V,2,2)
>>>>>>
>>>>>> According to the Julia guide "If the combine function is not 
>>>>>> supplied, duplicates are added by default." I expect Mat to be
>>>>>> [ 1, 1] =1.0
>>>>>> [ 1, 1] = 1.0
>>>>>>  [2, 2] = 1.0
>>>>>>
>>>>>> however, what I get is 
>>>>>> [1,1]=2
>>>>>> [2,2]=1
>>>>>>
>>>>>> So what is the default behavior without combine function? How do I 
>>>>>> specify the combine function? There is no instructions about it right 
>>>>>> now.
>>>>>>
>>>>>> 2. when i change V=[1.0;0.0;0.0]. I expected 
>>>>>> [ 1, 1] = 1.0
>>>>>>  [2, 2] = 0
>>>>>>
>>>>>> what I get is 
>>>>>> [1,1]=1
>>>>>>  
>>>>>> How can I let julia to combine elements only according to the indexes 
>>>>>> I, J ? How to keep those elements with value 0 ?
>>>>>>
>>>>>>
>>>>>> Thanks in advance. 
>>>>>>
>>>>>>
>>>>>> Yankai
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>

Reply via email to