That idea works.  But if the positions are regular as implied in this case,
doing Mark-Compute-Unmark will irregularize the positions component (i.e.,
product of compact arrays to full array -- that is about 160K floats vs.
6).  Some modules will also operate more slowly.




"Jeremy Zoss" <[EMAIL PROTECTED]>@opendx.watson.ibm.com on 06/20/2000 09:49:57
AM

Please respond to [email protected]

Sent by:  [EMAIL PROTECTED]


To:   <[email protected]>
cc:
Subject:  RE: [opendx-users] right-handed vs left-handed




> I have a 2-D grid with x varying fastest and
> z varying slowest.  So I define positions like so
>
> object 1 class gridpositions counts 401 201
> origin  0  0
> delta   0  1
> delta   1  0
>
> This displays nicely and all, but DX thinks that the first index is x and
> the second is y.  So my original grid which is (x,y,z) gets mapped to
> (x,z,y).

Hi!

I may be misreading your question here, but wouldn't it be possible to do
something like this:
  Import
  Mark("positions")
  Compute([a.x, b, a.y])
  UnMark("positions")

And use the compute module to rearrange your positions so that what DX
imports as "y" is actually moved to "z" and you insert a y-value
appropriate
to where your data plane is located?  You might have to use an enumerate to
pass in a vector-valued "b" or simply "b*(a.x/a.x)" with scalar "b" to get
an array for your new positions.  And, of course, if I'm way off base, feel
free to ignore this.  =)

Good luck!
  Jeremy Zoss
  Southwest Research Institute



Reply via email to