numpy-discussion — Discussion list for all users of Numerical Python

Re: [Numpy-discussion] ANN: gracePlot.py

[Rob <eu...@ho...>]

Can it do any 3d volume rendering?  I've heard of grace, but know
nothing about it.  Rob.


Nathaniel Gray wrote:
> 
>      __________________________________________________________________
> 
>                      Announcing:  gracePlot.py v0.5
> 
>              An interactive, user-friendly python interface to the
>                          Grace plotting package.
> 
>      __________________________________________________________________
> 
> * WHAT IS IT?
> 
> gracePlot.py is a high-level interface to the Grace plotting package available
> at: http://plasma-gate.weizmann.ac.il/Grace/  The goal of gracePlot is to
> offer the user an interactive plotting capability similar to that found in
> commercial packages such as Matlab and Mathematica, including GUI support for
> modifying plots and a user-friendly, pythonic interactive command-line
> interface.
> 
> * WHAT FEATURES DOES IT OFFER?
> 
> Since this package is in the early stages of development it does not yet
> provide high-level command-line access to all of Grace's plotting
> functionality.  It does, however, offer:
>     *   Line Plots (with or without errorbars)
>     *   Histograms (with or without errorbars)
>     *   Multiple graphs (sets of axes) per plot
>     *   Multiple simultaneous plots (grace sessions)
>     *   Overlaid graphs, using a 'hold' command similar to Matlab's
>     *   Legends, titles, axis labels, and axis limits
>     *   Integration with Numerical Python and Scientific Python's Histogram
>         object
> 
> Note that all advanced features and customizations are available through the
> Grace UI, so you can compose rough plots in Python and then polish them up in
> Grace.
> 
> * HOW DO I USE IT?
> 
> Here is an example session that creates a plot with two sets of axes, putting
> a line plot in one and a histogram in the other:
>     Python 2.1.1 (#2, Jul 31 2001, 14:10:42)
>     [GCC 2.96 20000731 (Linux-Mandrake 8.0 2.96-0.48mdk)] on linux2
>     Type "copyright", "credits" or "license" for more information.
>     >>> from gracePlot import gracePlot
>     >>> p = gracePlot() # A grace session opens
>     >>> p.plot( [1,2,3,4,5], [10, 4, 2, 4, 10], [1, 0.7, 0.5, 1, 2],
>     ...         symbols=1 )  # A plot with errorbars & symbols
>     >>> p.title('Funding: Ministry of Silly Walks')
>     >>> p.ylabel('Funding (Pounds\S10\N)')
>     >>> p.multi(2,1)  # Multiple plots: 2 rows, 1 column
>     >>> p.xlimit(0, 6)  # Set limits of x-axis
>     >>> p.focus(1,0)  # Set current graph to row 1, column 0
>     >>> p.histoPlot( [7, 15, 18, 20, 21], x_min=1,
>     ...              dy=[2, 3.5, 4.6, 7.2, 8.8]) # A histogram w/errorbars
>     >>> p.xlabel('Silliness Index')
>     >>> p.ylabel('Applications/yr')
>     >>> p.xlimit(0, 6)  # Set limits of x-axis
> 
> The result of this session can be found at:
>     http://www.idyll.org/~n8gray/code/index.html
> 
> * WHERE DO I GET IT?
> 
> gracePlot is available here:
>     http://www.idyll.org/~n8gray/code/index.html
> 
>      ___________________________________________________________
> 
> Cheers,
> -n8
> 
> --
> Nathaniel Gray
> 
> California Institute of Technology
> Computation and Neural Systems
> --
> 
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion

-- 
The Numeric Python EM Project

www.members.home.net/europax

[Numpy-discussion] ANN: gracePlot.py

[Nathaniel G. <n8...@ca...>]

     __________________________________________________________________

                     Announcing:  gracePlot.py v0.5
 
             An interactive, user-friendly python interface to the
                         Grace plotting package.

     __________________________________________________________________

* WHAT IS IT?

gracePlot.py is a high-level interface to the Grace plotting package available
at: http://plasma-gate.weizmann.ac.il/Grace/  The goal of gracePlot is to 
offer the user an interactive plotting capability similar to that found in 
commercial packages such as Matlab and Mathematica, including GUI support for 
modifying plots and a user-friendly, pythonic interactive command-line 
interface.


* WHAT FEATURES DOES IT OFFER?

Since this package is in the early stages of development it does not yet 
provide high-level command-line access to all of Grace's plotting 
functionality.  It does, however, offer:
    *   Line Plots (with or without errorbars)
    *   Histograms (with or without errorbars)
    *   Multiple graphs (sets of axes) per plot
    *   Multiple simultaneous plots (grace sessions)
    *   Overlaid graphs, using a 'hold' command similar to Matlab's
    *   Legends, titles, axis labels, and axis limits
    *   Integration with Numerical Python and Scientific Python's Histogram
        object

Note that all advanced features and customizations are available through the
Grace UI, so you can compose rough plots in Python and then polish them up in
Grace.


* HOW DO I USE IT?

Here is an example session that creates a plot with two sets of axes, putting
a line plot in one and a histogram in the other:
    Python 2.1.1 (#2, Jul 31 2001, 14:10:42)
    [GCC 2.96 20000731 (Linux-Mandrake 8.0 2.96-0.48mdk)] on linux2
    Type "copyright", "credits" or "license" for more information.
    >>> from gracePlot import gracePlot
    >>> p = gracePlot() # A grace session opens
    >>> p.plot( [1,2,3,4,5], [10, 4, 2, 4, 10], [1, 0.7, 0.5, 1, 2],
    ...         symbols=1 )  # A plot with errorbars & symbols
    >>> p.title('Funding: Ministry of Silly Walks')
    >>> p.ylabel('Funding (Pounds\S10\N)')
    >>> p.multi(2,1)  # Multiple plots: 2 rows, 1 column
    >>> p.xlimit(0, 6)  # Set limits of x-axis
    >>> p.focus(1,0)  # Set current graph to row 1, column 0
    >>> p.histoPlot( [7, 15, 18, 20, 21], x_min=1,
    ...              dy=[2, 3.5, 4.6, 7.2, 8.8]) # A histogram w/errorbars
    >>> p.xlabel('Silliness Index')
    >>> p.ylabel('Applications/yr')
    >>> p.xlimit(0, 6)  # Set limits of x-axis

The result of this session can be found at:
    http://www.idyll.org/~n8gray/code/index.html


* WHERE DO I GET IT?

gracePlot is available here:
    http://www.idyll.org/~n8gray/code/index.html

     ___________________________________________________________

Cheers,
-n8

-- 
Nathaniel Gray

California Institute of Technology
Computation and Neural Systems
--

Re: [Numpy-discussion] searching for a better 3d field file format

[Rob <eu...@ho...>]

Hi Janko, yes I would love to have it.  Thank  you!   I once installed
Viz5D, but didn't quite know what to do with it. One thing though, the
Windows version requires an X11 server, but thats better than no program
at all :)

I also installed OpenDX, the FreeBSD port version, but everytime I try
to do anything it core dumps.

I also want to look into VRML.  Rob.


Janko Hauser wrote:
> 
> I would try vis5d, which is a very nice out of the box viewer of
> 5D-Data. I have a module which can write vis5d-files directly from
> numpy-data. The module is not polished, so if there is interest I
> would send it privatly to you.
> 
> http://vis5d.sourceforge.net/
> 
> http://www.ssec.wisc.edu/~billh/vis5d.html
> 
> HTH,
> 
> __Janko
> 
> Rob writes:
>  > I've been using the "brick of bytes" format for my FDTD outputs, but
>  > unfortunately I can' find any type of 3d viewer for Windows.  I am now
>  > using the X11/OpenGL based Animabob, which creates a movie of the series
>  > of dumped files.
>  >
>  > I am thinking of trying some other type of file format and viewer.  I am
>  > wondering what others use to view 3d fields or data?  Of course if it
>  > used Numpy and Python all the better.
>  >
>  > Thanks,   Rob.
>  >
>  >
>  > --
>  > The Numeric Python EM Project
>  >
>  > www.members.home.net/europax
>  >
>  > _______________________________________________
>  > Numpy-discussion mailing list
>  > Num...@li...
>  > https://lists.sourceforge.net/lists/listinfo/numpy-discussion

-- 
The Numeric Python EM Project

www.members.home.net/europax

[Numpy-discussion] searching for a better 3d field file format

[Rob <eu...@ho...>]

I've been using the "brick of bytes" format for my FDTD outputs, but
unfortunately I can' find any type of 3d viewer for Windows.  I am now
using the X11/OpenGL based Animabob, which creates a movie of the series
of dumped files. 

I am thinking of trying some other type of file format and viewer.  I am
wondering what others use to view 3d fields or data?  Of course if it
used Numpy and Python all the better.

Thanks,   Rob.


-- 
The Numeric Python EM Project

www.members.home.net/europax

Re: [Numpy-discussion] Can I further Numpy-ize this code?

[Rob <eu...@ho...>]

I fixed my code by turning the one dimensional arrays into 3 dimensional
ones.  This is much faster and totally sliced.  But there still must be
a way thats even faster.  Rob.


Rob wrote:
> 
> I've been working on this for so long I may be missing the obvious.
> Here is a snippet of code that I would like to at least get rid of one
> more indexing operation.  The problem is the presence of the one
> dimensional array constants which are functions of x,y, or z depending
> on their location:  (you may recognize this as some FDTD code)  Any
> ideas?
> 
> Thanks,  Rob.
> 
>    ####################################/
>    # Update the interior of the mesh:
>    #    all vector H vector components
>    #
> 
> ##   for az in range(0,nz):
>    for ay in range(0,ny):
>       for ax in range(0,nx):
> 
>          dstore[ax,ay,0:nz]=Bx[ax,ay,0:nz]
> 
>          Bx[ax,ay,0:nz] =  Bx[ax,ay,0:nz] * C1[0:nz] + ( (
>             (Ey[ax,ay,1:(nz+1)]-Ey[ax,ay,0:nz] ) / dz -
>             (Ez[ax,ay+1,0:nz]-Ez[ax,ay,0:nz]) / dy ) * C2[0:nz] )
> 
>          Hx[ax,ay,0:nz]=  Hx[ax,ay,0:nz] * C3[ay]  + ( (
>              Bx[ax,ay,0:nz] * C5[ax]  - dstore[ax,ay,0:nz] * C6[ax] ) *
> C4h[ay] )
> 
>          dstore[ax,ay,0:nz]=By[ax,ay,0:nz]
> 
>          By[ax,ay,0:nz] =  By[ax,ay,0:nz] * C1[ax]  +  ( (
>             (Ez[ax+1,ay,0:nz]-Ez[ax,ay,0:nz]) / dx -
>             (Ex[ax,ay,1:(nz+1)]-Ex[ax,ay,0:nz]) / dz ) * C2[ax] )
> 
>          Hy[ax,ay,0:nz]=  Hy[ax,ay,0:nz] * C3[0:nz] +  ( (
>              By[ax,ay,0:nz] * C5[ay] - dstore[ax,ay,0:nz] * C6[ay] ) *
> C4h[0:nz] )
> 
>          dstore[ax,ay,0:nz]=Bz[ax,ay,0:nz]
> 
>          Bz[ax,ay,0:nz] =  Bz[ax,ay,0:nz] * C1[ay] +  ( (
>             (Ex[ax,ay+1,0:nz]-Ex[ax,ay,0:nz] ) / dy -
>             (Ey[ax+1,ay,0:nz]-Ey[ax,ay,0:nz] ) / dx ) * C2[ay] )
> 
>          Hz[ax,ay,0:nz]=  Hz[ax,ay,0:nz] * C3[ax] +  ( (
>              Bz[ax,ay,0:nz] * C5[0:nz]  - dstore[ax,ay,0:nz] * C6[0:nz]
> ) * C4h[ax] )
> 
> 
> --
> The Numeric Python EM Project
> 
> www.members.home.net/europax
> 
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion

-- 
The Numeric Python EM Project

www.members.home.net/europax

[Numpy-discussion] Can I further Numpy-ize this code?

[Rob <eu...@ho...>]

I've been working on this for so long I may be missing the obvious. 
Here is a snippet of code that I would like to at least get rid of one
more indexing operation.  The problem is the presence of the one
dimensional array constants which are functions of x,y, or z depending
on their location:  (you may recognize this as some FDTD code)  Any
ideas?

Thanks,  Rob.

   ####################################/
   # Update the interior of the mesh:
   #    all vector H vector components  
   #
   
##   for az in range(0,nz):
   for ay in range(0,ny):
      for ax in range(0,nx):
         
         dstore[ax,ay,0:nz]=Bx[ax,ay,0:nz]
         
         Bx[ax,ay,0:nz] =  Bx[ax,ay,0:nz] * C1[0:nz] + ( ( 
            (Ey[ax,ay,1:(nz+1)]-Ey[ax,ay,0:nz] ) / dz -             
            (Ez[ax,ay+1,0:nz]-Ez[ax,ay,0:nz]) / dy ) * C2[0:nz] ) 
            
         Hx[ax,ay,0:nz]=  Hx[ax,ay,0:nz] * C3[ay]  + ( ( 
             Bx[ax,ay,0:nz] * C5[ax]  - dstore[ax,ay,0:nz] * C6[ax] ) *
C4h[ay] )
         
         dstore[ax,ay,0:nz]=By[ax,ay,0:nz]
         
         By[ax,ay,0:nz] =  By[ax,ay,0:nz] * C1[ax]  +  ( (
            (Ez[ax+1,ay,0:nz]-Ez[ax,ay,0:nz]) / dx -
            (Ex[ax,ay,1:(nz+1)]-Ex[ax,ay,0:nz]) / dz ) * C2[ax] ) 
            
         Hy[ax,ay,0:nz]=  Hy[ax,ay,0:nz] * C3[0:nz] +  ( (
             By[ax,ay,0:nz] * C5[ay] - dstore[ax,ay,0:nz] * C6[ay] ) *
C4h[0:nz] )
         
         dstore[ax,ay,0:nz]=Bz[ax,ay,0:nz]
         
         Bz[ax,ay,0:nz] =  Bz[ax,ay,0:nz] * C1[ay] +  ( (
            (Ex[ax,ay+1,0:nz]-Ex[ax,ay,0:nz] ) / dy -
            (Ey[ax+1,ay,0:nz]-Ey[ax,ay,0:nz] ) / dx ) * C2[ay] ) 

         Hz[ax,ay,0:nz]=  Hz[ax,ay,0:nz] * C3[ax] +  ( (
             Bz[ax,ay,0:nz] * C5[0:nz]  - dstore[ax,ay,0:nz] * C6[0:nz]
) * C4h[ax] ) 
         




-- 
The Numeric Python EM Project

www.members.home.net/europax

Re: [Numpy-discussion] round() and abs() Ufuncs???

[John J. L. <jj...@po...>]

On Wed, 26 Sep 2001, Chris Barker wrote:
[...]
> OK, Tim Hochberg was nice enough to point out to me that abs() works on
> NumPy arrays. However, it does not work on other sequences, so maybe we
> need this:
>
> def abs(a):
>     return abs(asarray(a))

Numeric.absolute, as Robert Kern pointed out, maybe that hasn't arrived in
your mailbox...


John

Re: [Numpy-discussion] round() and abs() Ufuncs???

[Robert K. <ke...@ca...>]

On Wed, Sep 26, 2001 at 11:58:21AM -0700, Chris Barker wrote:
> Hi all,
> 
> I was recently surprised to find that there are no round(0 or abs()
> Ufuncs with Numeric. I'm imagining that they might exist under other
> names,

around and absolute.

-- 
Robert Kern
ke...@ca...

"In the fields of hell where the grass grows high
 Are the graves of dreams allowed to die."
  -- Richard Harter

Re: [Numpy-discussion] round() and abs() Ufuncs???

[Chris B. <chr...@ho...>]

Chris Barker wrote:

> I was recently surprised to find that there are no round() or abs()
> Ufuncs with Numeric. I'm imagining that they might exist under other
> names, but if not, I submit my versions for critique (lightly tested)

OK, Tim Hochberg was nice enough to point out to me that abs() works on
NumPy arrays. However, it does not work on other sequences, so maybe we
need this:

def abs(a):
    return abs(asarray(a))

-Chris


-- 
Christopher Barker,
Ph.D.                                                           
Chr...@ho...                 ---           ---           ---
http://members.home.net/barkerlohmann ---@@       -----@@       -----@@
                                   ------@@@     ------@@@     ------@@@
Oil Spill Modeling                ------   @    ------   @   ------   @
Water Resources Engineering       -------      ---------     --------    
Coastal and Fluvial Hydrodynamics --------------------------------------
------------------------------------------------------------------------

[Numpy-discussion] round() and abs() Ufuncs???

[Chris B. <chr...@ho...>]

Hi all,

I was recently surprised to find that there are no round(0 or abs()
Ufuncs with Numeric. I'm imagining that they might exist under other
names, but if not, I submit my versions for critique (lightly tested)

-Chris

from Numeric import *

def Uabs(a):
    """

    A Ufunc version of the Python abs() function

    """
    a = asarray(a)
    if a.typecode() == 'D' or a.typecode() == 'F':# for complex numbers
        return sqrt(a.imag**2 + a.real**2)
    else:
        return where(a < 0, -a, a)

def Uround(a,n=0):
    """

    A Ufunc version of the Python round() function. It should behave in
the same way

    Note: I think this is the right thing to do for negative numbers,
    but not totally sure. (Uround(-0.5) = 0, but Uround(-0.5000001) =
-1)

    It won't work for complex numbers

    """

    a = asarray(a)
    n = asarray(n)
    return floor((a * 10.**n) + 0.5) / 10.**n
    



-- 
Christopher Barker,
Ph.D.                                                           
Chr...@ho...                 ---           ---           ---
http://members.home.net/barkerlohmann ---@@       -----@@       -----@@
                                   ------@@@     ------@@@     ------@@@
Oil Spill Modeling                ------   @    ------   @   ------   @
Water Resources Engineering       -------      ---------     --------    
Coastal and Fluvial Hydrodynamics --------------------------------------
------------------------------------------------------------------------

[Numpy-discussion] FAST and EASY data plotting for Python, NumPy and Qt

[Gerard V. <gve...@la...>]

Announcing PyQwt-0.29.91

FAST and EASY data plotting for Python, NumPy and Qt

PyQwt is a set of Python bindings for the Qwt C++ class library.  The Qwt
library extend the Qt framework with widgets for Scientific and Engineering
applications.  It contains QwtPlot, a 2d plotting widget, and widgets for data
input/output such as and QwtCounter, QwtKnob, QwtThermo and QwtWheel. 

PyQwt requires and extends PyQt, a set of Python bindings for Qt.

PyQwt requires NumPy.  NumPy extends the Python language with new data types
that make Python an ideal language for numerical computing and experimentation
(like MatLab, but better).  

The home page of PyQwt is http://gerard.vermeulen.free.fr


NEW in PyQwt-0.29.91:

1. compatible with PyQt-2.5/sip-2.5 and PyQt-2.4/sip-2.4.

2. compatible with NumPy-20.2.0, and lower.

3. *.exe installer for Windows (requires Qt-2.3.0-NC).

4. build instructions for Windows and other versions of Qt.

5. HTML documentation with installation instructions and a reference listing
   the Python calls to PyQwt that are different from the corresponding C++
   calls to Qwt.

6. fixed reference counting bug in the methods with NumPy arrays as arguments.

7. new methods:
        QwtPlot.axisMargins()
        QwtPlot.closestCurve()
        QwtPlot.curveKeys()
        QwtPlot.markerKeys()
        QwtPlot.title()
        QwtPlot.titleFont()
        QwtScale.labelFormat()
        QwtScale.map()

8. changed methods:
	QwtCurve.verifyRange() -- cleaned up interface
	QwtPlot.adjust()       -- is now fully implemented
	QwtPlot.enableLegend() -- (de)selects all items or a single item

9. removed methods (incompatible with Python, because unsafe, even in C++):
	QwtCurve.setRawData()
	QwtPlot.setCurveRawData()
	QwtSpline.copyValues()

Gerard Vermeulen

[Numpy-discussion] append number to vector

[Henry H. <ha...@xm...>]

I often find myself wanting to append a number to a vector.  After fumbling
experimentation I use 

def comma(ar,inint):
    "comma(array,integer) returns array with integer appended"
    t=array([inint,])
    return(concatenate((ar,t),1))

which is used like the comma in apl, i.e.  ar <- ar, inint.


This seems klutzy to me.  Is there a simpler way to do it?  If ar were a
list, ar.append(inint) works, but no such luck after ar has become an array.

Thanks, Henry Harpending, University of Utah

[Numpy-discussion] expect a new wave of Numpy'ers

[Rob <eu...@ho...>]

I just announced a ham radio type FDTD simulation to the ham
newsgroups.  I couldn't resist trying out a ham type antenna, instead of
all the waveguide stuff I've been doing.  Rob.


-- 
The Numeric Python EM Project

www.members.home.net/europax

Re: [Numpy-discussion] Offset 2D arrays

[Mike R. <ro...@fs...>]

>>>>> " " == Herbert L Roitblat <roi...@ha...> writes:

     > Konrad's solution is MUCH more elegant.

     > Message ----- From: "Konrad Hinsen" <hi...@cn...>
[snip]

    >> How about this:
    >> 
    >> b = 0*a b[:-1, :-1] = a[1:, 1:]
    >> 

  I think it looks cleaner as well.  I've managed to create a function
(with the help of the tips on this list) which can offset a 2d array
in either or both the x and y dimensions.  I strongly suspect that
someone who *gets* python and numeric slicing better than I, can come
up with a cleaner approach.

def getindicies(o, l):
    if o > 0:
        s1 = o; e1 = l; s2 = 0; e2 = l - o
    elif o < 0:
        s1 = 0; e1 = l + o; s2 = -o; e2 = l
    else:
        s1 = 0; e1 = l; s2 = 0; e2 = l
    return s1, e1, s2, e2

# return a 2d array whose dimensions match a with the data offset
# controlled by x and y.
def offset(a, x, y):
    sy1, ey1, sy2, ey2 = getindicies(y, a.shape[0])
    sx1, ex1, sx2, ex2 = getindicies(x, a.shape[1])
    b = zeros(a.shape)
    b[sy1:ey1,sx1:ex1] = a[sy2:ey2,sx2:ex2]
    return b

a = array(((1, 2, 3), (4, 5, 6), (7, 8, 9)))

# no offset
print offset(a, 0, 0)
# offset by 1 column in x
print offset(a, 1, 0)
# offset by 1 column (opposite dir) in x
print offset(a, -1, 0)
# offset by 2 columns in x
print offset(a, 2, 0)
# offset by 2 columns in y
print offset(a, 0, 2)



Thanks,

Mike Romberg (ro...@fs...)

Re: [Numpy-discussion] Offset 2D arrays

[Herbert L. R. <roi...@ha...>]

Konrad's solution is MUCH more elegant.
HLR
----- Original Message -----
From: "Konrad Hinsen" <hi...@cn...>
To: <roi...@ha...>
Cc: <chr...@ho...>; <ro...@fs...>;
<num...@li...>
Sent: Tuesday, September 18, 2001 10:05 PM
Subject: Re: [Numpy-discussion] Offset 2D arrays


> > This will work:
> > b=zeros ((3,3))
> > b[:2,:2] = b[:2,:2] + a[1:,1:]
> >
> > You need to know the size of a to use this scheme.
>
> How about this:
>
>    b = 0*a
>    b[:-1, :-1] = a[1:, 1:]
>
> Works for any shape and type of a.
>
> Konrad.
> --
> --------------------------------------------------------------------------
-----
> Konrad Hinsen                            | E-Mail: hi...@cn...
> Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.56.24
> Rue Charles Sadron                       | Fax:  +33-2.38.63.15.17
> 45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
> France                                   | Nederlands/Francais
> --------------------------------------------------------------------------
-----
>

Re: [Numpy-discussion] Offset 2D arrays

[Konrad H. <hi...@cn...>]

> This will work:
> b=zeros ((3,3))
> b[:2,:2] = b[:2,:2] + a[1:,1:]
> 
> You need to know the size of a to use this scheme.

How about this:

   b = 0*a
   b[:-1, :-1] = a[1:, 1:]

Works for any shape and type of a.

Konrad.
-- 
-------------------------------------------------------------------------------
Konrad Hinsen                            | E-Mail: hi...@cn...
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.56.24
Rue Charles Sadron                       | Fax:  +33-2.38.63.15.17
45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
France                                   | Nederlands/Francais
-------------------------------------------------------------------------------

Re: [Numpy-discussion] Offset 2D arrays

[Herbert L. R. <roi...@ha...>]

This will work:
b=zeros ((3,3))
b[:2,:2] = b[:2,:2] + a[1:,1:]

You need to know the size of a to use this scheme.

----- Original Message ----- 
From: "Chris Barker" <chr...@ho...>
To: "Mike Romberg" <ro...@fs...>
Cc: <num...@li...>
Sent: Monday, September 17, 2001 2:03 PM
Subject: Re: [Numpy-discussion] Offset 2D arrays


> Mike Romberg wrote:
> > 
> >   I am attempting to create 2D arrays which are offset copies of a
> > given starting array.  For example if I have a 2D array like this:
> 
> > have any ideas?
> 
> This is not quite as clean as i would like, but this will work:
> 
> >>> a = array([[1, 2, 3],
> ...        [4, 5, 6],
> ...        [7, 8, 9]])
> >>> m,n = a.shape
> >>> b[:m-1,:n-1] =  a[1:,1:]
> >>> b
> array([[5, 6, 0],
>        [8, 9, 0],
>        [0, 0, 0]])
> >>>
> 
> if b does not have to be the same shape as a, then it is really easy:
> 
> >>> b =  a[1:,1:]
> 
> -Chris
> 
> 
> -- 
> Christopher Barker,
> Ph.D.                                                           
> Chr...@ho...                 ---           ---           ---
> http://members.home.net/barkerlohmann ---@@       -----@@       -----@@
>                                    ------@@@     ------@@@     ------@@@
> Oil Spill Modeling                ------   @    ------   @   ------   @
> Water Resources Engineering       -------      ---------     --------    
> Coastal and Fluvial Hydrodynamics --------------------------------------
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
> 

Re: [Numpy-discussion] Offset 2D arrays

[Chris B. <chr...@ho...>]

Mike Romberg wrote:
> 
>   I am attempting to create 2D arrays which are offset copies of a
> given starting array.  For example if I have a 2D array like this:

> have any ideas?

This is not quite as clean as i would like, but this will work:

>>> a = array([[1, 2, 3],
...        [4, 5, 6],
...        [7, 8, 9]])
>>> m,n = a.shape
>>> b[:m-1,:n-1] =  a[1:,1:]
>>> b
array([[5, 6, 0],
       [8, 9, 0],
       [0, 0, 0]])
>>>

if b does not have to be the same shape as a, then it is really easy:

>>> b =  a[1:,1:]

-Chris


-- 
Christopher Barker,
Ph.D.                                                           
Chr...@ho...                 ---           ---           ---
http://members.home.net/barkerlohmann ---@@       -----@@       -----@@
                                   ------@@@     ------@@@     ------@@@
Oil Spill Modeling                ------   @    ------   @   ------   @
Water Resources Engineering       -------      ---------     --------    
Coastal and Fluvial Hydrodynamics --------------------------------------
------------------------------------------------------------------------

Re: [Numpy-discussion] Offset 2D arrays

[Eric N. <no...@ph...>]

For "refine indexing and slicing" read "re-define indexing and
slicing".  Oops :)

Re: [Numpy-discussion] Offset 2D arrays

[Eric N. <no...@ph...>]

Mike,

As was pointed out to me when I had a similar query, one way to do
this is to define a class which inherits UserArray and refine indexing
and slicing.  I actually shifted by an offset of one in the opposite
direction to what you seem to require.  I had intended to generalize
to arbitrary offsets, but haven't had the time yet.  Anyway, you're
welcome to grab my code at
http://www.physics.ubc.ca/~mbelab/python/arrayone as a starting point
for your class.  There are still some issues and quirkiness with the
code, but they're documented along with work-arounds, and suggestions
for fixes have been made on this mailing list.  Again, it's a matter
of time...

regards,
Eric


On Mon, Sep 17, 2001 at 05:20:06PM -0600, Mike Romberg wrote:
> 
>   I am attempting to create 2D arrays which are offset copies of a
> given starting array.  For example if I have a 2D array like this:
> 
> array([[1, 2, 3],
>        [4, 5, 6],
>        [7, 8, 9]])
> 
>   I would like to offset it by some amount in either or both the x and
> y dimension.  Lets say that both the  x and y offset would be 1.  Then
> I would like to have an array like this:
> 
> 
> 
> array([[5, 6, 0],
>        [8, 9, 0],
>        [0, 0, 0]])
> 
>   Here I don't really care about the values which are now zero.  The
> main point is that now I can compare the data values at any given
> (x,y) point with the values at the adjacent point (over one on each
> axis).  This would be useful for the kinds of calculations we need to
> do.  I just can't come up with a numeric way to do this.  Does anyone
> have any ideas?
> 
> Thanks alot,
> 
> Mike Romberg (ro...@fs...)
> 
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion

-- 
********************************
Eric Nodwell
Ph.D. candidate
Department of Physics
University of British Columbia

tel: 604-822-5425
fax: 604-822-5324 
no...@ph...

[Numpy-discussion] Indexing in MATLAB

[Chris B. <chr...@ho...>]

OOPS! I replied to an arbitray message to get the address,and I forgot
to change the subject, so here is the same message I just posted, but
with an appropriate subject.

Hi all,

The MATLAB Digest just put out a little article about array indexing in
MATLAB. I thought some of yo might find it interesting, and it might
give some ideas for NumPy2. Most of what MATLAB has, NumPy has an
equivalent, but I would love to see what matlab calls vector indexing,
and a more natural way to do masks.

I know vector indexing would be a pretty tricky thing to have work, at
least with slices being references and all, but it would be a very nice
thing!! Perhaps some brilliant person can figure out an elegant and
efficient way to do it. 

Here is where you will find the article:

http://www.mathworks.com/company/digest/sept01/matrix.shtml


-Chris


-- 
Christopher Barker,
Ph.D.                                                           
Chr...@ho...                 ---           ---           ---
http://members.home.net/barkerlohmann ---@@       -----@@       -----@@
                                   ------@@@     ------@@@     ------@@@
Oil Spill Modeling                ------   @    ------   @   ------   @
Water Resources Engineering       -------      ---------     --------    
Coastal and Fluvial Hydrodynamics --------------------------------------
------------------------------------------------------------------------

[Numpy-discussion] Offset 2D arrays

[Mike R. <ro...@fs...>]

  I am attempting to create 2D arrays which are offset copies of a
given starting array.  For example if I have a 2D array like this:

array([[1, 2, 3],
       [4, 5, 6],
       [7, 8, 9]])

  I would like to offset it by some amount in either or both the x and
y dimension.  Lets say that both the  x and y offset would be 1.  Then
I would like to have an array like this:



array([[5, 6, 0],
       [8, 9, 0],
       [0, 0, 0]])

  Here I don't really care about the values which are now zero.  The
main point is that now I can compare the data values at any given
(x,y) point with the values at the adjacent point (over one on each
axis).  This would be useful for the kinds of calculations we need to
do.  I just can't come up with a numeric way to do this.  Does anyone
have any ideas?

Thanks alot,

Mike Romberg (ro...@fs...)

Re: [Numpy-discussion] PyArray_Check problem.

[Chris B. <chr...@ho...>]

Hi all,

The MATLAB Digest just put out a little article about array indexing in
MATLAB. I thought some of yo might find it interesting, and it might
give some ideas for NumPy2. Most of what MATLAB has, NumPy has an
equivalent, but I would love to see what matlab calls vector indexing,
and a more natural way to do masks.

I know vector indexing would be a pretty tricky thing to have work, at
least with slices being references and all, but it would be a very nice
thing!! Perhaps some brilliant person can figure out an elegant and
efficient way to do it. 

Here is where you eill find the article:

http://www.mathworks.com/company/digest/sept01/matrix.shtml


-Chris


-- 
Christopher Barker,
Ph.D.                                                           
Chr...@ho...                 ---           ---           ---
http://members.home.net/barkerlohmann ---@@       -----@@       -----@@
                                   ------@@@     ------@@@     ------@@@
Oil Spill Modeling                ------   @    ------   @   ------   @
Water Resources Engineering       -------      ---------     --------    
Coastal and Fluvial Hydrodynamics --------------------------------------
------------------------------------------------------------------------

[Numpy-discussion] RE: Segfault with latest lapack_litemodule using complex matrices.

[Paul F. D. <pa...@pf...>]

A. Mirone rode to the rescue and the fix is available as
Numeric-20.2.1.tar.gz or in CVS.

-----Original Message-----
From: Travis Oliphant [mailto:oli...@ie...] 
Sent: Saturday, September 15, 2001 6:58 PM
To: Paul F. Dubois
Subject: Segfault with latest lapack_litemodule using complex matrices.



Hi Paul,

With the latest changes to the lapack_lite module, I now get segfaults
when 
I'm trying to load call LinearAlgebra.eigenvectors using a
complex-valued 
matrix.

Could you check this on your system?

Thanks,

-Travis

Re: [Numpy-discussion] re: gcc 3.0 vs. 2.95.2

[Rob <eu...@ho...>]

Peter Soendergaard wrote:
> 
> Hello,
> 
> I just browsed over the archives for Numpy-discussion and saw this, and
> decided to sign up.
> 
> I work on the ATLAS-project http://math-atlas.sourceforge.net/ and we have
> had similar problems with gcc3.0 Gcc 3.0 has a completely new backend
> which produces much slower floating point code on i386 machines. It is
> most visible on the Athlon, but it also also shows on on P4 and PIII
> machines. We havn't yet figured out if there are some optimizations that
> can make this go away, but if you need performance stick with the old 2.95
> release for now.
> 
> By the way, if you would like to use Atlas in NumPy (I don't know if you
> do it already) I might be of some help. There is c-interfaces to the BLAS
> bundled with ATLAS, supporting both row-major and column-major storage.
> 
> Cheers,
> 
> Peter.
> 
> FROM: Rob
> DATE: 09/04/2001 18:02:05
> SUBJECT:  [Numpy-discussion] Python and Numpy compiled on Athlon
> optimized gcc3.01
> 
> Just for kicks last night I installed gcc3.01 which claims to have Athlon
> optimization, ie. -march=athlon.  I recompiled Python and Numpy, and then
> ran a big simulation.  The new compiler ran 200 seconds slower than the
> old gcc2.95 with plain -march=pentium.  I need to go to the gcc website
> and see just what optimization they are claiming.  Maybe I should have
> also used -02.   Rob.
> 
> --
> The Numeric Python EM Project
> 
> http://www.members.home.net/europax
> 
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion


Hi Peter,  great to know that I'm not alone.   Maybe I can build an EM
simulator using all integer math :).  I think I mentioned this on the
list before, but Win2k on my laptop (1Ghz) runs Numpy and Python faster
than my 1.2Ghz Athlon DDR machine using FreeBSD :(   Also, for
reference, I have a 3DNow optimized MP3encoding program Gogo that
encodes mp3's 10x faster on the athlon than Lame does on my FreeBSD
system on the laptop.  Go figure!

-- 
The Numeric Python EM Project

www.members.home.net/europax