Forgot to mention that in the first GR.contour attempt I had x and z
swapped...
On Thursday, July 14, 2016 at 12:11:57 PM UTC+3, Kaj Wiik wrote:
>
> I dug this problem bit deeper, it seems that my unique is working but
> GR.contour seems to be complaining even of close coordinates. I tried
> converting everything to Float32 but that didn't help either:
>
> julia> contour(cf[3,:],cf[2,:],cf[1,:])
> *** IDENTICAL DATA POINTS.
> NDP =12629 IP1 = 23 IP2 = 3286 XD=56192.9 YD=0.55634
> ERROR DETECTED IN ROUTINE IDTANG.
>
> julia> cf[:,[23,3286]]
> 3x2 Array{Float32,2}:
> 0.0 6.64646e-20
> 0.55634 0.555965
> 56192.9 56192.9
>
>
> Strange...
>
>
> On Wednesday, July 13, 2016 at 10:58:16 PM UTC+3, Kaj Wiik wrote:
>>
>> Thanks Josef.
>>
>> I tried GR contour but got identical datapoint errors. It seems that it's
>> not trivial to use unique() other than vectors...could not get it working.
>> In fact it seems that the reported datapoints are not in fact identical:
>>
>> julia> contour(c[1,:],c[2,:],c[3,:])
>> *** IDENTICAL DATA POINTS.
>> NDP =12629 IP1 = 1 IP2 = 346 XD=1.02114e-99 YD=0.478687
>> ERROR DETECTED IN ROUTINE IDTANG.
>>
>> julia> c[:,1]
>> 3-element Array{Float64,1}:
>> 1.02114e-99
>> 0.478687
>> 56193.8
>>
>> julia> c[:,346]
>> 3-element Array{Float64,1}:
>> 1.38697e-84
>> 0.477974
>> 56196.2
>>
>>
>> I tried also to call natgrid directly but it seems to be a can of worms.
>>
>> Kaj
>>
>>
>> On Saturday, July 9, 2016 at 11:02:23 AM UTC+3, Josef Heinen wrote:
>>>
>>> You can use the GR framework, e.g.
>>>
>>> srand(0)
>>> xd = -2 + 4 * rand(100)
>>> yd = -2 + 4 * rand(100)
>>> zd = [Float64(xd[i] * exp(-xd[i]^2 - yd[i]^2)) for i = 1:100]
>>>
>>> using GR
>>> contour(xd, yd, zd, colormap=44)
>>>
>>>
>>> <https://lh3.googleusercontent.com/-2Jv_GQhoJLE/V4CvWwlK5oI/AAAAAAAAADY/dS2YfieTyIEXfUjXoDCNZdnSg0dcevKDwCLcB/s1600/Screen%2BShot%2B2016-07-09%2Bat%2B10.00.51.png>
>>>
>>> If you only need the gridded data, use the GR gridit function, e.g. (to
>>> obtain a 200 x 200 grid):
>>>
>>> x, y, z = GR.gridit(xd, yd, zd, 200, 200)
>>>
>>>
>>> On Saturday, July 9, 2016 at 12:57:51 AM UTC+2, Kaj Wiik wrote:
>>>>
>>>> Is there a Julia version of irregularly spaced data gridding that does
>>>> zi = griddata(x,y,z,xi,yi), i.e. all arguments are 1d vectors? It seems
>>>> that Julia interp and contour packages require x, y, z[x,y].
>>>>
>>>>
>>>> https://scipy.github.io/old-wiki/pages/Cookbook/Matplotlib/Gridding_irregularly_spaced_data.html
>>>>
>>>> Thanks,
>>>> Kaj
>>>>
>>>>
