| Re: Problem with REFORM and HISTOGRAM [message #72810] |
Tue, 12 October 2010 05:33 |
rogass
Messages: 200
Registered: April 2008
|
Senior Member |
|
|
On 12 Okt., 14:05, Mrunmayee <gaur...@gmail.com> wrote:
> Sorry, I didn't realize that posts were on again.
>
> I didn't know about this function, so thanks for pointing out. So for
> say 220 points on each line, there will be 220*220 distances, this
> might be memory-prohibitive. With rebinning I mean, won't it?
>
> On Oct 9, 3:33 am, chris <rog...@googlemail.com> wrote:
>
>
>
>> On 8 Okt., 07:03, Mrunmayee <gaur...@gmail.com> wrote:
>
>>> My post did not get posted about 12 hours ago. I live in India and was
>>> hoping the post would be up for gurus in western timezones to possibly
>>> offer help. Now I will wait for yet another 24 hours! I always check
>>> the box to "Send copy to self" which I DID get. And I also notice a
>>> lot of spam has gone. I don't think my post contained anything
>>> 'hotter' than rebin/reform/histogram. So, dear moderator, may I ask
>>> how I offended this group? Anyway. Onto the problem:
>
>>> 1. I have 2 arrays of coordinates, x1arr of N1 size, x2arr of N2 size
>>> and same for y1arr, y2arr, z1arr, z2arr. These are coordinates along 2
>>> different lines and I need all the distances between points on one
>>> line and points on another. I do it by:
>>> x1arrReb = Rebin(x1arr, N1, N2)
>>> x2arrReb = Rebin(Transpose(x2arr), N1, N2)
>>> dx = x1arrReb - x2arrReb
>>> Similarly obtain dy, dz. Then,
>>> d = Sqrt(dx*dx + dy*dy + dz*dz)
>
>>> 2. I need to bin these distances for further computation.
>>> dhist = Histogram(d, nbins=8, locations=bins, reverse_indices=ri)
>
>>> 3. Now I want to check, in each bin, which distance corresponds to
>>> which x1 and which x2 coordinate. From
>>> above, I can do following:
>
>>> x1arrRef = Reform(x1arrReb, N1*N2) & x2arrRef = Reform(x2arrReb,
>>> N1*N2)
>
>>> Then use the Reverse_Indices and get coordinates. Right?
>
>>> 4. Alternatively, I was trying this:
>>> for i = 0,nbins-1
>>> indices = ri[ ri[i]: ri[i+1]-1 ]
>>> rowID = indices/N1 ; This is to get i,j values from the 1D indices.
>>> colID = indices - rowID * N1
>>> x1coord = x1arr[rowID] ; This is the original 1-d N1-element
>>> x1arr.
>>> x2coord = x2arr[colID] ; ------- ditto ------ x2arr.
>>> endfor
>
>>> Problem is, if this way, I get all the coordinates and re-calculate
>>> distances to see if I get the distances in the bin 'i' back, I don't!
>>> And I don't know where my understanding is wrong here. So can someone
>>> please point out? Am I book-keeping indices wrong?
>
>> Why don't you use the distance_measure function? You can also just
>> rebin the dist function into third dimension times its transpose and
>> then ask for the values according to its indices ?
>
>> Cheers
>
>> Cr
If you don' use the /MATRIX flag, the output will be pairwise.
Cheers
CR
|
|
|
|
| Re: Problem with REFORM and HISTOGRAM [message #72811 is a reply to message #72810] |
Tue, 12 October 2010 05:05  |
Mrunmayee
Messages: 24
Registered: May 2009
|
Junior Member |
|
|
Sorry, I didn't realize that posts were on again.
I didn't know about this function, so thanks for pointing out. So for
say 220 points on each line, there will be 220*220 distances, this
might be memory-prohibitive. With rebinning I mean, won't it?
On Oct 9, 3:33 am, chris <rog...@googlemail.com> wrote:
> On 8 Okt., 07:03, Mrunmayee <gaur...@gmail.com> wrote:
>
>
>
>> My post did not get posted about 12 hours ago. I live in India and was
>> hoping the post would be up for gurus in western timezones to possibly
>> offer help. Now I will wait for yet another 24 hours! I always check
>> the box to "Send copy to self" which I DID get. And I also notice a
>> lot of spam has gone. I don't think my post contained anything
>> 'hotter' than rebin/reform/histogram. So, dear moderator, may I ask
>> how I offended this group? Anyway. Onto the problem:
>
>> 1. I have 2 arrays of coordinates, x1arr of N1 size, x2arr of N2 size
>> and same for y1arr, y2arr, z1arr, z2arr. These are coordinates along 2
>> different lines and I need all the distances between points on one
>> line and points on another. I do it by:
>> x1arrReb = Rebin(x1arr, N1, N2)
>> x2arrReb = Rebin(Transpose(x2arr), N1, N2)
>> dx = x1arrReb - x2arrReb
>> Similarly obtain dy, dz. Then,
>> d = Sqrt(dx*dx + dy*dy + dz*dz)
>
>> 2. I need to bin these distances for further computation.
>> dhist = Histogram(d, nbins=8, locations=bins, reverse_indices=ri)
>
>> 3. Now I want to check, in each bin, which distance corresponds to
>> which x1 and which x2 coordinate. From
>> above, I can do following:
>
>> x1arrRef = Reform(x1arrReb, N1*N2) & x2arrRef = Reform(x2arrReb,
>> N1*N2)
>
>> Then use the Reverse_Indices and get coordinates. Right?
>
>> 4. Alternatively, I was trying this:
>> for i = 0,nbins-1
>> indices = ri[ ri[i]: ri[i+1]-1 ]
>> rowID = indices/N1 ; This is to get i,j values from the 1D indices.
>> colID = indices - rowID * N1
>> x1coord = x1arr[rowID] ; This is the original 1-d N1-element
>> x1arr.
>> x2coord = x2arr[colID] ; ------- ditto ------ x2arr.
>> endfor
>
>> Problem is, if this way, I get all the coordinates and re-calculate
>> distances to see if I get the distances in the bin 'i' back, I don't!
>> And I don't know where my understanding is wrong here. So can someone
>> please point out? Am I book-keeping indices wrong?
>
> Why don't you use the distance_measure function? You can also just
> rebin the dist function into third dimension times its transpose and
> then ask for the values according to its indices ?
>
> Cheers
>
> Cr
|
|
|
|
| Re: Problem with REFORM and HISTOGRAM [message #72848 is a reply to message #72811] |
Fri, 08 October 2010 15:33 |
rogass
Messages: 200
Registered: April 2008
|
Senior Member |
|
|
On 8 Okt., 07:03, Mrunmayee <gaur...@gmail.com> wrote:
> My post did not get posted about 12 hours ago. I live in India and was
> hoping the post would be up for gurus in western timezones to possibly
> offer help. Now I will wait for yet another 24 hours! I always check
> the box to "Send copy to self" which I DID get. And I also notice a
> lot of spam has gone. I don't think my post contained anything
> 'hotter' than rebin/reform/histogram. So, dear moderator, may I ask
> how I offended this group? Anyway. Onto the problem:
>
> 1. I have 2 arrays of coordinates, x1arr of N1 size, x2arr of N2 size
> and same for y1arr, y2arr, z1arr, z2arr. These are coordinates along 2
> different lines and I need all the distances between points on one
> line and points on another. I do it by:
> x1arrReb = Rebin(x1arr, N1, N2)
> x2arrReb = Rebin(Transpose(x2arr), N1, N2)
> dx = x1arrReb - x2arrReb
> Similarly obtain dy, dz. Then,
> d = Sqrt(dx*dx + dy*dy + dz*dz)
>
> 2. I need to bin these distances for further computation.
> dhist = Histogram(d, nbins=8, locations=bins, reverse_indices=ri)
>
> 3. Now I want to check, in each bin, which distance corresponds to
> which x1 and which x2 coordinate. From
> above, I can do following:
>
> x1arrRef = Reform(x1arrReb, N1*N2) & x2arrRef = Reform(x2arrReb,
> N1*N2)
>
> Then use the Reverse_Indices and get coordinates. Right?
>
> 4. Alternatively, I was trying this:
> for i = 0,nbins-1
> indices = ri[ ri[i]: ri[i+1]-1 ]
> rowID = indices/N1 ; This is to get i,j values from the 1D indices.
> colID = indices - rowID * N1
> x1coord = x1arr[rowID] ; This is the original 1-d N1-element
> x1arr.
> x2coord = x2arr[colID] ; ------- ditto ------ x2arr.
> endfor
>
> Problem is, if this way, I get all the coordinates and re-calculate
> distances to see if I get the distances in the bin 'i' back, I don't!
> And I don't know where my understanding is wrong here. So can someone
> please point out? Am I book-keeping indices wrong?
Why don't you use the distance_measure function? You can also just
rebin the dist function into third dimension times its transpose and
then ask for the values according to its indices ?
Cheers
Cr
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
