Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.sao.ru/precise/Midas_doc/courier/courier-93jul.ps Äàòà èçìåíåíèÿ: Thu Sep 8 11:57:55 1994 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 02:47:43 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ìåòåîðèòèêà

Newsletter of the ESO­MIDAS Users' Community
18 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993
The ESO­MIDAS Courier is published twice per year by the ESO­MIDAS Group of the European Southern
Observatory.
Contributions as well as suggestions and comments are invited and can be sent to the editor. In particular,
authors of ESO­MIDAS application software that would be of general interest for the ESO­MIDAS community,
are invited to make this software available.
If you are not on the mailing list and want to receive future issues of the Courier, contact the editor. More updated
information about ESO­MIDAS can be obtained from the ESO­MIDAS Bulletin Board, accessible via anonymous
ftp.
The ESO­MIDAS TM Courier:
Editor: Rein H. Warmels
Typeset by FrameMaker TM
Published by the European Southern Observatory
Karl­Schwarzschild­Straúe 2,
D 85748 Garching bei MÝnchen,
Federal Republic of Germany
ISSN 1018­3051
The following MIDAS Support services are available
for sending suggestions, comments or to obtain help
when problems arise.
Telephone: +49­89­32006456
Telex: 5282822 eso d,
attn. ESO­MIDAS
Telefax: +49­89­3202362,
attn. ESO­MIDAS
DECnet/Span: ESO::MIDAS
Bitnet/EARN: MIDAS@DGAESO51
UUCP/Eunet: midas@eso.uucp
Internet: midas@eso.org
Anonymous ftp: ftphost.hq.eso.org (134.171.8.4)

Newsletter of the ESO­MIDAS Users' Community
ESO­MIDAS Courier, Vol. 3, No. 1, July 1993 17
ESO­MIDAS Request Form
This is a request form for release 93NOV of the ESO­MIDAS 1 system. To obtain a new
release, please complete this form and return it to us (see reverse). Material will be shipped only
to users with a valid ESO­MIDAS User Agreement. If you still have a tape or tape mailing box
from the last release, you MUST return these before we send the new release.
ESO­MIDAS User Agreement No 2 :
Technical Support Coordinator:
E­Mail:
Tape format:
Documentation requested:
Date: Signature:
1. ESO­MIDAS TM is copyright protected software developed by the European Southern Observatory for the pur­
pose of Image Processing of Astronomical Data.
2. for new user agreements t.b.d.
1600 bpi VAX/VMS backup
6250 bpi VAX/VMS backup
1600 bpi TAR format
6250 bpi TAR format
QIC­24 TAR format
Exabyte 8mm TAR format
DDS/DAT 4mm VAX/VMS backup
ftp (TAR files only)
DDS/DAT 4mm TAR format
ESO­MIDAS User Guide 93NOV
ESO­MIDAS Environment Version 1.1
IDI­routines
AGL Reference Manual Version 3.60

Newsletter of the ESO­MIDAS Users' Community
16 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993

Newsletter of the ESO­MIDAS Users' Community
ESO­MIDAS Courier, Vol. 3, No. 1, July 1993 15
drive your coffee consumption through the roof
while waiting for the results, because the execu­
tion of such a procedure is excruciatingly slow
except for small one­dimensional spectra
A much better alternative is the MIDAS command
REPLACE/IMAGE which offers you the possibil­
ity to execute those IF statements inside a com­
piled module. As a simple example let us compute
an image result.bdf which is set to 0.0 for all those
pixels of galaxya.bdf with a value not greater than
0.0 and equal to galaxya.bdf for all other pixels
REPLACE/IMAGE galaxya result
<,0.0=0.0 will do exactly that. This means,
that for all pixels of the input frame galaxya.bdf
with a value less than or equal to 0.0 we set the
corresponding pixel in result.bdf to 0.0; for all
other pixels the output pixel will be set to the
value of the corresponding pixel of galaxya.bdf.
The number of pixels falling into the given
replacement interval is reported and also stored
into keyword OUTPUTI(15).
Instead of setting those pixels of result.bdf to the
constant 0.0 you might want to set them to the cor­
responding pixels of image galaxyb.bdf. Then,
change the command to REPLACE/IMAGE
galaxya result <,0.0=galaxyb.
In order to replace only those pixels which are
equal to 0.0 by the pixels of galaxyb.bdf we use
REPLACE/IMAGE galaxya result
0.0,0.0=galaxyb.
If we want to test the pixel values not in the input
frame but in a different frame, e.g. a 0/1 mask
badpix.bdf indicating bad pixels via 0, we use
REPLACE/IMAGE galaxya result bad­
pix/<,0.0=galaxyb.
As another example, let us set all the `bad' pixels
to the pixels of galaxyb.bdf and all the good ones
to the corresponding pixels of galaxya.bdf multi­
plied by 5.3. The command to achieve that is
REPLACE/IMAGE galaxya result bad­
pix/<,0.0=galaxyb,galaxya*5.3.
And, finally, let us treat the case where we want to
3. see MIDAS Users Guide, Vol. A, Chapter 3 for
a detailed discussion of immediate commands
set the output frame equal to the input frame gal­
axya.bdf if all its pixels are less than the pixels of
a second input frame galaxyb.bdf, otherwise the
output frame should be equal to galaxyb.bdf. This
cannot be done in a single REPLACE command,
but a short sequence of MIDAS commands will
produce the required output frame:
COMPUTE/IMAGE &d = galaxya­gal­
axyb
REPLACE/IMAGE &d &a 0.0,>=1.0,0.0
COMPUTE/IMAGE &a = {outputi(15)}
REPLACE/IMAGE &a result <,0.0=
galaxya,galaxyb 1
First, we create an intermediate frame middumm­
d.bdf containing the difference of the two input
frames.Then, we create another dummy frame
middumma.bdf with the same size as galaxya.bdf.
The keyword OUTPUTI(15) is set to a positive
integer if the value of at least one pixel of mid­
dummd.bdf is greater than or equal to 0.0, i.e. if at
least one pixel of galaxya.bdf is not less than the
corresponding pixel of galaxyb.bdf; otherwise
OUTPUTI(15) is set to 0. In the next command
all pixels of that dummy frame are set to OUT­
PUTI(15), i.e. all pixels are set to 0 if all pixels
of galaxya.bdf are less than the corresponding pix­
els of galaxyb.bdf, otherwise all pixels are set to a
positive integer. Thus, the last command will
either set all pixels of result.bdf to the values of
galaxya.bdf or to the ones of galaxyb.bdf as
required.
Also in the REPLACE/IMAGE command all tests
and calculations use single floating point data and
image pixels are converted to real data on the fly if
necessary. But, contrary to the COMPUTE com­
mands all frames involved must have the same
dimensions (size).
The given examples do not illustrate all the differ­
ent options available, consult the HELP text of
REPLACE/IMAGE to obtain a complete list of all
possibilities and more examples.
1. this option of REPLACE/IMA will only work from
release 93NOV on

Newsletter of the ESO­MIDAS Users' Community
14 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993
works as above but additionally takes the square
of the data.
The basic mathematical functions, like square
root, exponent, log, as well as the main trigono­
metric functions like sine, cosine, etc. are sup­
ported; for a detailed list consult the HELP info of
COMPUTE/IMA 1 .
So, another example would be COMPUTE/IMAGE
result = galaxya ­ 12*(LOG10(gal­
axyb)­1.1)
What happens, if we evaluate a function with a
pixel value for which that function is not defined,
like taking the log of a negative pixel, or if we
divide by an image where some pixels are zero? In
such a case the pixel in the output image is set to a
predefined NULL value and the total number of
these undefined pixels is reported. The first 2 ele­
ments of the real keyword NULL are involved
here: the NULL value is stored in the 2 nd element
of keyword NULL and may be modified by the
user (it is set to 0.0 by MIDAS, initially) and the
undefined pixel count is stored in NULL(1) by
the COMPUTE/IMA command.
The constants involved in the command may be
fixed, like in the example above, or values stored
in keywords and descriptors or elements of tables
or single pixel values. For example, if you want to
divide an image by its maximum value, you would
type COMPUTE/IMAGE image = image/
image,{lhcuts(4)}. For, the real descriptor
LHCUTS holds the minimum and maximum val­
ues of an image in its 3 rd and 4 th element. Note,
that all `variables' in curly brackets ``{}'' will be
converted to ASCII by the MIDAS monitor before
the character input string is fed to the program
which executes the command. This conversion is
done according to the format indicated via the
SET/FORMAT command, default values are I4.4
for integer and G15.5 for real and double variables
(in FORTRAN notation).
The COMPUTE/IMA command takes world coor­
dinates into account (as specified in the descrip­
tors START and STEP of an image) and works
only on those parts of the input images which
overlap in world coordinate space. Therefore, all
the input images must not necessarily have the
same size. The size of the resulting image is set
1. Note, that the trigonometric functions expect
their arguments in degrees not radians
accordingly to the overlapping area of all input
frames. Also, the stepsize of all the input images
has to match, otherwise an error occurs.
If you just want to do computations with different
images on a pixel by pixel basis and ignore their
start and step values, use the command COM­
PUTE/PIXEL which has the same syntax and
provides the same functionality as COMPUTE/
IMAGE (but needs input images of same size).
In general, you want to copy all descriptors of the
input frame to the result frame. But of which input
frame if there are more than one? The default
strategy in MIDAS is to copy all (non­standard)
descriptors of the first input image in the expres­
sion 2 . Thus, the descriptors of galaxya.bdf would
be copied into result.bdf in our first example. If,
instead, we wanted to copy the descriptors of gal­
axyb.bdf to the result frame, we would have to
indicate that via: COMPUTE/IMAGE
result,galaxyb = galaxya ­
12*(galaxyb­1.1)
A few basic mathematical constants can be set via
the application procedure matconst.prg. The
MIDAS command @a matconst will create the
double precision keyword MATCONST and fill its
five elements by the values of p, 2p, p/180, e and
square root of 2. E.g. if you want to multiply an
image with p, enter the command COMPUTE/
IMAGE result = galaxya*{mat­
const(1)}.
By the way, COMPUTE/IMAGE may also be used
as a pocket calculator if only constants are
involved by just typing the algebraic expression,
e.g. COMPUTE/IMA log10(123.456) will
print out the resulting value of 2.091512 and
also store it into keyword OUTPUTR(1).
The COMPUTE commands apply the given calcu­
lations to every pixel of the input images. There
are many applications, however, where you want
to apply different calculations depending upon the
value of each individual pixel. Suppose, you don't
want to code your application directly in the FOR­
TRAN or C programming language. You could
quite simply do it via a MIDAS procedure by
using the IF statements and the immediate com­
mands involving single pixels 3 . But this would
2. you can switch off the automatic descriptor cop­
ying via the MIDAS command SET/MIDAS_SYS
DSCCOPY=NO

Newsletter of the ESO­MIDAS Users' Community
ESO­MIDAS Courier, Vol. 3, No. 1, July 1993 13
tiresolution transforms of the given image. The
de­noising strategy is implemented at these vari­
ous resolution scales. This contributes in part to
the power and effectiveness of wavelet­based
image restoration.
Using multiresolution to detect very faint
extended objects
The multiresolution views of an image are very
valuable, in their own right, for detecting faint
structures. For this, we have used the commands
TRANS/WAVE to create the wavelet­transformed
image, using a 6­scale option. Then, we have dis­
played the resulting resolution scales with the
command VISU/PLAN. A visual appraisal of
these images may lead to the choice of one of
them.
The types of faint extended structures examined
include cometary comas, and infalling matter in
the central regions of galaxies. Images have
come from HST Planetary Camera, the NTT
SUSI, and the 1.2m at Calar Alto. The wavelet
analysis allowed for insightful perspectives on
very faint structures in the images. No direct clues
were provided in the original images. Examples of
this work can be found in Murtagh, BÆnhardt,
Zeilinger and Starck (``Detection of faint extended
structures by multiresolution wavelet analysis'',
Messenger, August 1993).
A further topic which we hope to check out in the
near future is the use of the wavelet transform for
detection of spectral lines. The burgeoning litera­
ture on the use of the wavelet transform on astro­
nomical data points to its increasingly widespread
use. Some of its associated properties, such as its
multiresolution treatment of data, make it particu­
larly relevant for astronomical data. All in all, the
new wavelet package in MIDAS represents a
powerhouse of ideas, many of which are still wait­
ing to be developed and tested.
References
Gabor, D. 1946, Journal of I.E.E. 93, 429
Grossmann, A., Morlet, J., 1984, SIAM J. Math.
Anal, 15, 723.
Mallat, S., 1989: IEEE Trans. on Pattern Anal. and
Math. intel. Vol. 11, 7
Expeditions in the MIDAS Jungle: Computing
Klaus Banse, European Southern Observatory
Finally, we have fought our way through all these
details of the data structures in MIDAS and can
move on to some of the real stuff now. A data
analysis system may have glamorous graphical
user interfaces, dazzling graphics and sophisti­
cated typesetting facilities, but that's no big help if
there are not efficient and robust computing tools
available to the user. So, this time let's explore in
some depth the commands which support compu­
tations involving the different data objects of
MIDAS.
Let us begin with the workhorse of the MIDAS
commands, COMPUTE/IMA 1 . This command
works on algebraic expressions written in the
1. the corresponding command for tables is (you
guessed it right!) COMPUTE/TABLE but we leave a
detailed exploration of that command for a later issue
standard mathematical way involving MIDAS
images and constants. The usual hierarchy of
arithmetic operations is applied, to override it use
brackets. All calculations are done using single
floating point data representation and computa­
tion, if an input image has a different data format,
the image pixels are converted to floating point
format on the fly when the image is read in. The
pixels of the output image are stored as real data
as well.
Thus, with the command COMPUTE/IMAGE
result = galaxya ­ 12*(gal­
axyb­1.1) we subtract 1.1 from each pixel of
image galaxyb.bdf, multiply that by 12.0 and sub­
tract all that from the image galaxya.bdf to pro­
duce the output image result.bdf. And the
command COMPUTE/IMAGE result =
(galaxya ­ 12*(galaxyb­1.1))**2

Newsletter of the ESO­MIDAS Users' Community
12 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993
PLAN, and VISUAL/PERS. A typical sequence
command would be:
. SET/CONTEXT WAVELET
. RANSF/WAVE NGC2997 WNGC
. VISUAL/WAVE WNGC
Here NGC2997 is the input image, and the result
is written in the file WNGC. Default parameters
are chosen. The step 3 will create three windows
on the screen and load each scale of the wavelet
transform in a window. Three files are created by
the same command (scale_1.bdf,
scale_2.bdf, and scale_3.bdf).
HEADER/WAVE writes on the screen information
about the wavelet file (which wavelet transform
was used, name and size of the image, etc.) and
INFO/WAVE computes statistical informations
for all scales (minimum, maximum, standard devi­
ation, etc.).
EXTRACT/WAVE creates an MIDAS image from
a scale of the wavelet transform, and INSERT/
WAVE replaces a scale by an image. By these two
commands, we can modify the wavelet transform.
An image can be reconstructed from its wavelet
transform by RECONS/WAVE. The following
sequence multiplies the first scale of the wavelet
transform by 10, so the high frequencies in the
reconstructed image (called NGC_REC) will be
enhanced.A typical command sequence would be:
. SET/CONTEXT WAVELET
. RANSF/WAVE NGC2997 WNGC
. EXTRACT/WAVE WNGC s1 1
. COMPUTE/IMA s1 = s1*10
. ENTER/WAVE WNGC s1 1 WNGC_OUT
. RECONS/WAVE WNGC_OUT NGC_REC
Three applications using the wavelet transform
have been implemented: Filtering (FILTER/
WAVE), deconvolution (LUCY/WAVE, GRAD/
WAVE, CIITER/WAVE, DIRECT/WAVE, and
comparison (COMPARISON/WAVE, PLOT/SNR,
PLOT/COR) of two images.
Two commands concerns one dimensional data:
TRAN1D/WAVE which computes 1D wavelet
transform and REC1D /WAVE which reconstructs
the signal at the full resolution from its wavelet
transform.
More information about these commands can be
found in the MIDAS Manual Vol 2, release
93NOV.
A Few Applications
After the package was installed is was extensively
used for a number of objectives. These objectives
were concentrated on image restoration and mul­
ti­resolution detection techniques.
Image restoration with simultaneous
de­noising.
Thresholding the wavelet­transformed image has
been a strategy proposed by a number of authors
for noise removal. In a 1992 wavelets conference,
Donoho (Stanford) stated: ``... shrinking noisy
wavelet coefficients via thresholding offers very
attractive alternatives to existing methods of
recovering signals from noisy data. .... Agreement
of diverse theoretical and empirical work is very
encouraging and suggests that wavelets will soon
have a large impact on how scientists treat noisy
data.''
An enhanced Richardson­Lucy image restoration
or sharpening implementation is available with
the command LUCY/WAVE. If no thresholding is
carried out in wavelet space, then a classical Rich­
ardson­Lucy restoration is obtained. An investiga­
tion of the use of this enhanced RL method for
de­noising is presented in a paper by Starck and
Murtagh (``Image restoration or sharpening with
noise suppression using the wavelet transform'',
submitted). For removing noise, and improving
the visual appearance of restored images, the
results can be most impressive. In the paper
cited, a study is also carried out as to how and
where this de­noising approach improves the
quality of the basic RL method. Simulated
Hubble Space Telescope data is used for this
appraisal, and it is found that the quality of faint
object detections is significantly improved.
From another perspective, this de­noising strategy
additionally solves another problem. This relates
to how the user often wishes to treat the back­
ground of the image differently from the fore­
ground objects of interest. Greater smoothing in
the background may actually be helpful, both in
terms of avoiding the creation of artifacts, and
also in terms of avoiding unnecessary flux redis­
tribution. But why should one simply view an
astronomical image as background plus superim­
posed objects? Why not recognize that there is
often a continuum of superimposed objects? The
wavelet transform naturally allows a set of mul­

Newsletter of the ESO­MIDAS Users' Community
ESO­MIDAS Courier, Vol. 3, No. 1, July 1993 11
Wavelet: A New MIDAS Context
Jean­Luc Starck, European Southern Observatory
Fionn Murtagh, Space Telescope­European Coordinating Facility
Introduction
The Fourier transform is a tool widely used for
many scientific purposes, but it is well suited only
to the study of stationary signals where all fre­
quencies have an infinite coherence time. The
Fourier analysis brings only global information
which is not sufficient to detect compact patterns.
Gabor (1946) introduced a local Fourier analysis,
taking into account a sliding window, leading to a
time frequency­analysis. This method is only
applicable to situations where the coherence time
is independent of the frequency. This is the case
for instance for singing signals which have their
coherence time determined by the geometry of the
oral cavity. Morlet introduced the Wavelet Trans­
form in order to have a coherence time propor­
tional to the period.
The continuous wavelet transform of a function
f(x) is described by Grossmann and Morlet
(1984):
where a (>0) is a dilatation parameter and b is a
translation parameter. The transform is character­
ized by three main properties: i) it is a linear trans­
formation, ii) it is covariant under translations, iii)
it is covariant under dilations. We get in the Fou­
rier space:
The transformation is a set of filterings. When the
scale a varies, the filter is only reduced or dilated.
The function f(x) can be restored from its wavelet
transform. The restoration formula is:
W a b
,
( ) 1
a
­­­­­­­ y k x b
--
a
­­­­­­­­­­­
Õ Ü
Ô Æ f x
( ) dx
¥
--
¥
Ð
=
C “ a n
,
( ) ay “ k an
( ) f “ n
( )
=
f x
( ) 1
C y
­­­­­­­ 1
a
­­­­­­­ C a b
,
( ) y x b
--
a
­­­­­­­­­­­
Õ Ü
Ô Æ dadb
a 2
­­­­­­­­­­­­­
0
¥
Ð
¥
--
¥
Ð
=
with
This admissibility condition is generally verified
if:
Many 2d extensions of the continuous wavelet
transform are possible. Mallat (1989) introduced
the multi­resolution analysis and proposed a quick
algorithm based on the separability of the varia­
bles. This method allows us to have a non­redun­
dant transform, so that the number of pixels in the
transform is the same as in the image. This is par­
ticularly interesting in image compression. Other
algorithms exist which are non redundant (the
number of pixels in the transform is larger than the
number of pixels in the image): ``Þ trous algo­
rithm'', pyramidal algorithms. These algorithms
have the advantage that they use isotropic wave­
lets, which are closer to physical patterns.
Some of the technical applications of the use of
the wavelet transform to astronomical imaging
derived from this scheme are:
. optimal image addition;
. filtering;
. deconvolution;
. structure detection.
Command Overview
Below follows a brief overview of the commands
available in the WAVELET context.
TRANSF/WAVE computes the wavelet transform
of an image. Eight algorithms are implemented:
``a trous'' algorithm, Mallat's algorithm, pyrami­
dal algorithms, etc. A file is created which con­
tains the result, and it can be visualized by
commands VISUAL/WAVE, VISUAL/CUBE,
VISUAL/CONT, VISUAL/SYNT, VISUAL/
C y
y “ n
( ) 2
n
­­­­­­­­­­­­­­­­­­­­ dn ¥
<
0
¥
Ð
=
y “ 0
( ) 0
=

Newsletter of the ESO­MIDAS Users' Community
10 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993
Basically, this means displaying and changing
parameters. Also, commands exist to store the
current parameter settings and to retrieve these
after a session is restarted.
When the context CCD is enabled, a keyword
structure is defined, containing all CCD specific
keywords required to execute the various reduc­
tion steps. The keywords are roughly grouped
according to the reduction step. In the reduction
process the system needs specific information
stored in the frame descriptor. E.g. the standard
MIDAS way to store the exposure time is in the
descriptor O_TIME. However, in case this infor­
mation is stored somewhere else the system
should be able to find it. Therefore, the CCD key­
word structure is equipped with a number of key­
words that contain the names of important frame
descriptors. If these descriptors are absent sensible
defaults will be used. Finally, a number of key­
words are created to keep general information
about the observation and the status of the reduc­
tion.
Handling HST Data in MIDAS
M. Almudena Prieto, Max Planck Institute for Extraterrestrial Physics,
Garching bei MÝnchen
After a little more of three years of HST operation
the HST archive contains at the present time about
28000 ``public'' observations. Having expired their
proprietary period of one year, any of these obser­
vations become available to the general astronom­
ical community and the interested user can access
and retrieve his/her preferred ones via
STARCAT 1 . The availability of such, still
increasing, amounts of data requires MIDAS to
provide their users with adequate procedures to
read and manipulate HST data in the MIDAS
environment.
Specific MIDAS procedures for analysing HST
data, either spectroscopy or imaging, are now
available. These routines are supposed to form
part, in a future, of a much general MIDAS con­
text for handling HST data. Currently, they allow
the user to convert a set of HST data, extracted
from a given instrument and observing mode, in
the appropriate scientific format to be manipulated
with the usual MIDAS tools. In their present
stage, they only operate with the final, science cal­
ibrated data, therefore, no attempt for any re­cali­
bration is made. Presently, they are only
implemented for their most frequent observing
1. STARCAT (Space Telescope ARchive and CATa­
logue) has been jointly developed by the ESO/Archive,
the ST­ECF Archive and the CADC Science Data
Archives.
modes of HST. The instruments considered are
FOC, WF/PC, FOS and GHRS.
The general structure of the procedures is the fol­
lowing one: read FITS format versions of the HST
data into MIDAS, rename the files to their associ­
ate HST observing names, and finally, transform
the files to an adequate scientific format to be dis­
played and manipulated in MIDAS (e.g., creation
of a WF/PC mosaic out of the four corresponding
CCD images). Each instrument has associated its
specific MIDAS procedures to perform those
tasks.
Interested users can get the package of procedures
by anonymous ftp in mc3.hq.eso.org
(134.171.8.4). For UNIX users, the package plus a
README are in a compressed tar file called:
midaspub/92NOV/unix/hst.tar.Z For
VMS users, the procedures are located in:
midaspub/92NOV/vms/hst/*.prg. A
README is also available in the hst subdirec­
tory.
The README file contains a detailed description
of what each procedure does and how they must
be used. Potential users should follow the steps
indicated in order to guarantee a successful appli­
cation. Questions and comments are welcome;
please, send a mail to either the MIDAS group or
to the author at alm@mpe­garching.m­
pg.de.

Newsletter of the ESO­MIDAS Users' Community
ESO­MIDAS Courier, Vol. 3, No. 1, July 1993 9
A New CCD Context in MIDAS
Rein H. Warmels, European Southern Observatory
Introduction
Since the last update of the MIDAS CCD package
in 1986 a number of new instruments have been
installed on the La Silla Telescopes. In addition,
new CCD's became available offering large pixel
areas and higher quantum efficiency. With these
innovations the variety of observing modes has
grown and, as an obvious consequence, the
amount and the diversity of data taken have dra­
matically increased. It is clear that the MIDAS
CCD reduction software should be able to cope
with these improvements and hence requires com­
patibility with the hardware as it exists now.
Requirements
When designing the basic layout of the CCD soft­
ware the following basic requirements were kept
in mind:
. it should be robust;
. it should be user friendly, easy to use;
. it should offer processing possibilities for a
large variety of instruments;
. it should be able to operate both in an auto­
matic mode (to handling large quantities of
data) as well as in single command mode (for
single image analysis);
. it should offer flexible reduction procedures;
. it should be intelligent.
How intelligent the system is depends on the
capabilities of other parts in the data acquisition
(ESO Archive) and reduction (MIDAS) systems.
In this respect the development of the CCD pack­
age took place at the right time: the new MIDAS
Data Organizer package now offers the possibility
to create a MIDAS table containing the science
frames to be reduced and their associated calibra­
tion frames.
General Overview
The CCD package provides a set of basic com­
mands that perform the various calibration steps.
These are combining of calibration frames, sub­
traction of the bias level determined from the
overscan area or from a separate bias frame, cor­
rection for dark current, division by the flat field,
correction for illumination, and correction for the
fringe pattern. Also, tools are provided for trim­
ming the frames of the unwanted overscan strip,
and for correcting the frame for bad pixels intensi­
ties.
All operations steps successfully finished are
recorded in the descriptor of the (partially)
reduced frame. That facility, which includes
updating the HISTORY descriptor, avoids repeti­
tion of reduction sequences, and provides the user
with the information what has been done to the
data.
By combining the basic reduction steps a com­
plete reduction pipeline procedure is built, that
enables the user to do a completely automatic
reduction of all science frames. The pipeline pro­
cedure is controlled by a set of reduction key­
words, that control which of the available
reduction steps will be executed. Obviously, in
order to get the desired result, these keywords
should be set correctly.
Apart from commands that do the actual work, a
number of commands will help the user to manage
keywords and descriptors and their content/value.
Deadline for the next ESO­MIDAS Courier
The next issue of the ESO­MIDAS Courier, Volume 3, Number 2, will be published in January 1994.
This issue will include more detailed information about the new 93NOV release as well as on­going
projects of the MIDAS Group. Again, we hope to include contributions by MIDAS users or software
developers. Please, send your contributions, preferably in computer­readable format, to the MIDAS
account or to the editor before November 30 th (see back page for mail address).

Newsletter of the ESO­MIDAS Users' Community
8 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993
The Observation Summary Table (OST), and the
corresponding information for a given input file is
stored into one of its rows.
Classification of Images
A natural way to get an overview of the data one
has is to try to classify the files into a set of
groups. One may want for instance to group the
frames according to the exposure type or one may
need to put together all the files observed in a
given instrument mode. An instruction for group­
ing Flat Field exposures could be: Select all files
which have a descriptor IDENT matching one of
the substrings 'FF', 'SKYFL' or 'FLAT' and can
be thought as a logical expression understood by
the SELECT/TABLE command. A set of default
rules may be provided for the classification pur­
pose but the possibility should also be given to the
user to write his/her own rules because what is
written in the observation log file usually doesn't
follow any standard conventions, and for instance,
the rule mentioned above would not flag as such
flat field exposures which have been identified
with the string ``f.f''. Furthermore, writing such a
expression can be tedious unless the user is pro­
vided with a utility which facilitates the formula­
tion of the query. An interface using the Table
editor has been designed. It enables the user to
enter interactively constraints on the existing
fields of the OST. Relational operators (e.g. >, <,
!= or =) as well as values, range of values and wild
card characters (*) may be used. The given con­
straints are ANDed and translated into a expres­
sion understandable by the SELECT command.
This expression is stored in the header of the table.
In addition, a command exists to edit it interac­
tively.
The classification process tags each frame with a
set of character strings provided by the user that
identify the image in a unique way and that will be
stored into columns of the OST. For instance a sci­
entific exposure observed in the Red Imaging
Mode of EMMI using the filter number 505 may
be described with the two strings ``SCI'' and
``FR505''.
At this stage of the process, it is already possible
to make a preliminary statistical analysis of the
observing run and this without having to inspect
interactively individual files. One can, for
instance, plot the sequence number of the images
against the date and use different symbols accord­
ing to the exposure type in order to have an over­
view of what has been observed. Or one may want
to plot the r.m.s. noise of the flat field exposures
against their mean pixel value to check the general
consistency of the frames and whether there are
any outliers.
Association of Images
The association of science frames with suitable
calibration frames is achieved by using the same
rule generating interface as referred to above even
though the rules to be applied are different: One
may want for instance to look for all the flat fields
which have been taken within a certain time inter­
val of a given science exposure. It only makes
sense to associate to a given scientific frame cali­
bration frames observed with the same telescope/
instruments configuration, in other words images
might be associated only if they belong to the
same classes that have been defined during the
association process. This list of classes, for
instance optical­path class, detector mode class
has to be provided in the command line.
The Association Process creates a MIDAS table
which be used by any reduction package. It con­
tains one column for each of the type of exposures
(SC, BIAS, DK,WCAL, etc.) and each row con­
tains for the corresponding science image the set
of suitable calibration frames.
The Context DO within MIDAS
The Data Organizer may be used by activating the
context DO within MIDAS (set/context
DO). The following commands are available:
. CREATE/OST: create an Observation Sum­
mary Table (OST) from a set of FITS files or
MIDAS images (.bdf);
. CREATE/RULE: create classification rules for
an OST;
. APPLY/RULE: apply a given rule to a OST;
. ASSOCIATE/IMA: associate suitable calibra­
tions frames to a scientific exposure;
. GROUP/ROW: group the rows of a table by the
value of one of its column.

Newsletter of the ESO­MIDAS Users' Community
ESO­MIDAS Courier, Vol. 3, No. 1, July 1993 7
The first modification is required in order to have
MIDAS ported to the new generation of 64 bit
CPUs (e.g. the Alpha chip from DEC) while the
latter will resolve possible alignment problems.
Technical Background
The major difficulty in porting MIDAS to the
OSF/1 system on an Alpha chip, has been the new
length of ``long int'' variables in the C program­
ming language which changed from 32­bits to
64­bits.
The new length of ``long'' variables although con­
sistent along the MIDAS C code, is incompatible
with our previous assumption that an ``INTEGER''
Fortran variable, when passed as an argument to a
C routine, is received as a pointer to a ``long''.
In other words, the MIDAS Fortran­to­C, and
therefore, the Standard interfaces define their
arguments as ``long'' and pointers to ``long''.
Although this is correct on a 32­bit machine, it
will not work on an Alpha because, still in For­
tran, ``INTEGER'' variables are 32­bits.
Other solutions like a particular Fortran­to­C for
Alpha or an option of the Alpha Fortran compiler
which assigns 64 bits to INTEGER, have been
rejected in favour of the much cleaner solution of
modifying the C implementation of the MIDAS
Standard interfaces where ``long'' arguments will
be re­defined as ``int''.
Nothing will change with respect to the Fortran
interfaces of MIDAS, so, users who have written
MIDAS application programs in Fortran are not
affected and may continue reading just for pleas­
ure (?). Even MIDAS applications written in C
will not be affected as long as they are running on
32­bit machines, which are currently all CPUs
except DEC's Alpha chip under OSF/1.
With this message we want to inform the MIDAS
community and alert any developers of MIDAS
applications in the C language (only they are
really concerned...). For more information please
contact Preben GrosbÜl (pgrosbol@eso.org)
or Klaus Banse (kbanse@eso.org).
The ESO­MIDAS Environment document is
under configuration control in order to ensure a
high degree of stability. The MIDAS Group
judges that this proposal will provide a clean and
reliable implementation of MIDAS on the next
generation of computers with a minimum impact
on the user community. It is therefore important
that the issue is discussed in detail by users in
order to ensure that a change is implemented only
with full consensus in the community.
A message concerning these changes has been
sent to all MIDAS site managers in the start of
July with a deadline for comments of July 31.
Since no objections were received the changes
are considered accepted and the modifications will
be incorporated in the 94NOV release of MIDAS.
A new version of the ESO­MIDAS Environment
document MID­SPE­ESO­11000­0001 will then
be issued in January 1994.
DO: The New Data Organizer Context
MichÕle Peron, European Southern Observatory
Before being able to actually reduce and analyse a
new set of observations, the observer has to pre­
pare, sort out and arrange the data, that is for
instance, make a first quality check, classify data
according to a set of rules and associate with each
science frame a set of relevant calibration frames.
This task can be cumbersome because of the com­
plexity of the instruments and the large number of
data files they produce.
A first version of the Data Organizer has been
developed using the existing capabilities of the
Table File System.
The Observation Summary Table
The Data Organizer uses as input a list of FITS
files or MIDAS images as well as a list of FITS
keywords or MIDAS descriptors which are con­
sidered to be relevant (e.g. exposure time, tele­
scope setting, instrument mode). Each of these
keywords is mapped into one column of a table
that is called the Observation Summary Table.

Newsletter of the ESO­MIDAS Users' Community
6 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993
account.
To ease the installation of MIDAS, the MIDAS
Group is considering the distribution of already
installed copies of MIDAS for PC/Linux. Eventu­
ally, this could be done on CD­ROMs.
Our hardware and software configuration in the
next table is only given for information and is not
intended to be the unique nor the minimum hard­
ware option to run MIDAS. A ``LINUX HARD­
WARE COMPATIBILITY LIST'' is available and
updated in the USENET newsgroup com­
p.os.linux (latest version 0.6, dated 01
August 1993). It is also available in the file
midaspub/linux.hw.list under our anon­
ymous ftp account. MIDAS requests only a 386 or
486 CPU, Linux release SLS 1.01 or higher, a
minimum of 16 Mbytes of memory but some
more is recommended if big images are going to
be processed, and 75 to 150 Mbytes of disk space
for MIDAS depending on which packages will be
installed
PC configuration:
. i486DX/25
. 20 Mbytes RAM
. Adaptec 1542B SCSI board.
. 1 Gbyte SCSI hard disk.
. Western­Digital 8013 Ethernet board.
. Local Bus S3 Video Card.
Linux configuration:
. Linux SLS 1.01 0.99.pl9­1
. cc: GNU compiler 2.3.3 (included)
. f2c: Fortran 77 to C translator 22 (included)
. X11 R5 (included)
. MIDAS beta­release 93NOV (28 Mbytes).
The next table is a comparison between perform­
ance between MIDAS on a PC and on a Sparc sta­
tion2. Here we should remark that on Linux there
is no real Fortran compiler but a Fortran­to­C
translator, and access to disk on the Sparc2 is
about 5 times faster than on our PC.
. MIDAS installation time (only core):
PC/Linux: 44 min.
Sparc2: 30 min.
. Size of binaries (only core of MIDAS) (a):
PC/Linux: 32 Mbytes
Sparc2: 41 Mbytes
. cwhetstones (b):
PC/Linux: 10 MIPS
Sparc2: 10 MIPS
. MIDAS benchmark (c):
PC/Linux: 2435 sec.
Sparc2: 2045 sec.
. Wavelet tutorial (d):
PC/Linux: 313 sec.
Sparc2: 405 sec.
(a) SunOS was linked using a MIDAS shared
library while Linux was not.
(b) Whetstone benchmark in C. ``A Synthetic
Benchmark'' by H.J. Curnow and B.A. Wichman
in Computer Journal, Vol 19, Number 1, February
1976.
(c) filter/median on a square gaussian image of
1000x1000.
(d) MIDAS package 100% C code.
In order to limit our administrational overhead, we
will not distribute the ESO­MIDAS PC/Linux
version to individuals but only to registered sites.
Thus,we will giveMIDAS site managers permis­
sion to distribute PC versions of MIDAS to people
associated with their institute.
MIDAS 94NOV: Modification of Type Definitions
Preben GrosbÜl, European Southern Observatory
Introduction
The MIDAS Group herewith announces a chnage
of type definitions of arguments in the C­imple­
mentation of the Standard interfaces as of the
94NOV release of MIDAS (November 1994).
The modification consists of:
. a change of the arguments of type ``long int'' to
``int'' in all SC and TC routines;
. a change of the type of the parameter `unit'
(e.g. in SCKRDC) from type ``char **'' to ``int
*''.

Newsletter of the ESO­MIDAS Users' Community
ESO­MIDAS Courier, Vol. 3, No. 1, July 1993 5
ties of the MIDAS table file system. More infor­
mation about the package can be found in separate
article elsewhere in this Courier.
Since the software could not be included in the
MIDAS beta­test version, it will not be included
in the official 93NOV release. However, the pack­
age will be available on the MIDAS anonymous
ftp account at soon as the testing has completed.
New Context CCD
The DO context is used in the new CCD context,
already announced in the last MIDAS Users'
Meeting. As the DO package the CCD software
will not be included in the official 93NOV release
of MIDAS but will be made available on the
anonymous ftp account. More details about the
package can be found in this Courier.
Contexts SPEC and LONG
The package LONG has been totally refurbished
since the 92NOV release. It incorporates 1D and
long­slit spectroscopy and presents all functionali­
ties previously spread in three different packages
(SPEC, SPECTRA and LONG). A tutorial (com­
mand TUTORIAL/LONG) describes the com­
mands of the package.
The result of this exercise is that the functions of
the context SPEC for reduction of 1D spectros­
copy data are now covered by the package
LONG.The context SPEC still exists but includes
only some low­level commands used by both
main spectroscopy packages LONG and
ECHELLE.
ESO­MIDAS available on PC/Linux
Carlos Guirao, European Southern Observatory
In the previous issue of the MIDAS Courier of
January 1993 we described the efforts of the
MIDAS Group to port MIDAS to one of the
announced Unix systems for PCs. In particular,
the MIDAS Group decided to use Linux, a public
domain Unix system, as the most suitable Unix for
the MIDAS community. Linux not only satisfied
the requirements to install MIDAS (C and Fortran
compilers plus X11) but it is in compliance with
ANSI­C and POSIX standards and, in addition, it
also includes all the network software necessary to
integrate the PC into a LAN. Linux is supported
by the Free Software Foundation, and collaborates
with Linux porting and providing public domain
software (like GNU).
The popularity of Linux has increased enormously
in the last months, as can be inferred from the
``linux'' newsgroup, one of the most active news­
group currently in the USENET service, and with
that the interest of the astronomical community
for having MIDAS ported to it. This became
apparent during the 5 th ESO/ST­ECF Data Analy­
sis Workshop, where the MIDAS Group showed
the progress with the port of MIDAS to a PC/486
running Linux SLS 1.0.
Now, the situation has again improved substan­
tially and we are pleased to announce that the
complete ``core'' of MIDAS been successfully
implemented and verified on Linux SLS 1.01.
Some other MIDAS packages have also been
tested by their authors (like WAVELET, PEPSYS,
ECHELLE, AND LONG) and a prerelease of our
next release of MIDAS 93NOV has been already
distributed to several test­sites for a complete
check out.
The installation of MIDAS with shared libraries
(see under General News), that will reduce the
needs of disk space for binary files to less than a
half, was recently completed. The Graphic User
Interface (GUI) packages for Linux are, however,
pending of the delivery of Motif 2.1 and the new
release of our X­builder for X11 Release 5
(August­September 1993).
Our next official release of MIDAS 93NOV will
be fully supported for Linux and the distribution
will be done as usual in QIC­24, QIC­150, Exa­
byte, DAT/DDS and reel­magtapes in ``tar'' for­
mat, and through our ``anonymous ftp'' account on
ftphost.hq.eso.org (134.171.8.4) in ``compressed
tar'' format, under the restricted ``midas'' ftp
account (password is required).
Contact Resy de Ruijsscher to request a tape or to
activate a password to login on the ``midas ftp''

Newsletter of the ESO­MIDAS Users' Community
4 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993
MIDAS installed on Solaris 2.1
MIDAS has been installed and verified success­
fully on a SUN SparcClassic running the new Sys­
tem V operating system from SUN Solaris 2.1.
The installation was done using the public domain
GNU C compiler, which is available for Solaris
2.1 on ``anonymous ftp'' accounts, and the SUN
licensed Fortran compiler SC2.0.1. The installa­
tion will be also tested using the SUN licensed C
compiler.
The only problem with the port of MIDAS to
Solaris 2.1 worth mentioning was the misalign­
ment by default of double­precision data types,
which is incompatible with the alignment
expected in the C code. This problem is solved
with the new ``­dalign'' option in the Fortran com­
piler, which is included automatically with the
new Solaris 2.1 preinstallation.
Shared Libraries
New in 93NOV is the installation of MIDAS with
shared libraries, which have the major benefit of
the reduction of about 50% in the size of binary
files. The following is a list of systems where
MIDAS is implemented using shared libraries,
and other systems will follow this list in the com­
ing months
. SUN/SunOS 4.1.n;
. Solaris 2.1;
. HP/HPIX;
. PC/Linux.
Application News
Graphical User Interfaces
The version 93NOV of MIDAS will incorporate a
set of four MOTIF based interfaces: XHelp for the
on­line documentation, XDisplay for display­re­
lated commands, XLong for 1D and long­slit
spectroscopy and XAlice (see below) or spectral
analysis. All these interfaces conform to the ESO
GUI Common Conventions which define the
Look and Feel of all ESO interfaces in the fields
of telescope and instrument control, archives and
data analysis. The version 92NOV incorporated
many Athena based interfaces (XSpectra,
XEchelle, XFIlter, XStella). Some of them
(XEchelle, XFilter, XStella) have not been yet
ported to MOTIF and compiled versions for Sun
and HP will be available under anonymous ftp
account on our host ``ftphost.hq.eso.org''.
New Context XALICE
There is a new interface available for interactive
analysis of spectral data. It is called XAlice and
can be used from MIDAS by typing crea/gui
alice. It was developed by Juan Veliz at La
Silla. The main features are:
. Flux integration: it is possible to integrate the
flux of a spectral feature above or below the
continuum. A gaussian fit is performed to
determine the wavelength of the feature. It
computes parameters like: fwhm, equivalent
width, flux, continuum, a.s.o;
. Rebinning: logarithmic, inverse and red­shift
rebinning are implemented;
. Filtering: smooth and median filtering is
allowed;
. Multi­gaussian fitting: gaussian curves of up to
9 components can be fitted after giving the
fwhm, the central X value and the amplitude
interactively.
New Context WAVELET
The 93NOV release of MIDAS will contain the
new context WAVELET. It is written by Jean­Luc
Starck, during his three months visit of the
MIDAS Group at ESO. The package can be
invoked with the command SET/CONTEXT
WAVELET. It contains about twenty commands,
most of which concern images. The command
TUTORIAL/WAVE gives an overview of the vari­
ous wavelet transform algorithms, what kind of
data they produce, and how to visualize these
data. Several wavelet transform algorithms, and
tools to manipulate the wavelet transform have
been implemented. More information about the
new package is described in a separate article in
this Courier.
New Context DO
A new context DO (for Data Organizer) written
by MichÕle Peron is almost completed and will be
released for internal use soon. The package ena­
bles the user to sort and group his data (e.g. data
from one or more observing runs), based on a set
of selection criteria, and using many of the facili­

Newsletter of the ESO­MIDAS Users' Community
3 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993
access to the new restricted ``midas'' account and
to retrieve a file. In the examples commands typed
% ftp 134.171.8.4 (if ftphost.hq.eso.org doesn't work)
connected to mc3.hq.eso.org.
220 mc3 FTP server (Version 2.1aWU(2) Mon Jul 19 13:43:34 MET DST 1993) ready.
Name (ftphost.hq.eos.org:jdp): anonymous (or ftp)
331 Guest login ok, send your complete e­mail address as password.
Password: jdp@awislpa.goya.es (use your e­mail address)
230 Bla, bla, bla, bla...
230 Please read the file README
230 Guest login ok, access restrictions apply.
ftp> get README (get the information file)
200 PORT command successful.
150 Opening ASCII mode data connection for README (479 bytes).
226 Transfer complete.
local: README remote: README
495 bytes received in 0.0032 seconds (1.5e+02 Kbytes/s)
ftp> quit (exit the ftp session)
221 Goodbye.
% ftp 134.171.8.4 (if ftphost.hq.eso.org doesn't work)
connected to mc3.hq.eso.org.
220 mc3 FTP server (Version 2.1aWU(2) Mon Jul 19 13:43:34 MET DST 1993) ready.
Name (ftphost.hq.eos.org:jdp): midas (restricted ftp account)
331 Password required for midas:??????
230 This is the restricted midas account (password supplied by ESO­IPG)
230 Please read the file README
230 Guest login ok, access restrictions apply.
ftp> get README (get the information file)
200 PORT command successful.
150 Opening ASCII mode data connection for README (779 bytes).
226 Transfer complete.
local: README remote: README
495 bytes received in 0.0032 seconds (1.5e+02 Kbytes/s)
ftp> quit (exit the ftp session)
221 Goodbye.
by the user are underlined; comments are in ital­
ics.
Patches for the 92NOV release
Since the previous issue of the ESO­MIDAS Cou­
rier no new patch file for the 92NOV version was
put on the anonymous ftp account.
(ftphost.hq.eso.org or IP 134.171.8.4)
under the midaspub directory. Hence, the patch
file P92NOV.01 is still the most recent one. Distri­
bution tapes of ESO­MIDAS contains, besides the
latest official release of ESO­MIDAS, those
patches that have been released at the time the
tape was generated. They are included on the dis­
tribution tape as a separate tar­format file. When
patches are included in the distribution tape, the
sources of the command ``patch'' are also included
as the third tar file.
Patches will not be distributed on tape to sites that
have already received the distribution tape, except
for very particular cases (e.g. those sites without
access to the Internet). Sites that have received
and installed the distribution tape are strongly rec­
ommended to check for the existence of new
patches in our ``anonymous ftp'' account under
Important.
System News
Announcement for the 94NOV
release
The MIDAS Group proposes to modify the type
definitions of arguments in the C­implementation
of the Standard interfaces as of the 94NOV
release of MIDAS (November 1994). Elsewhere
in this ESO­MIDAS Courier you will find more
detailed information about this proposal.

Newsletter of the ESO­MIDAS Users' Community
2 ESO­MIDAS Courier, Vol. 3, No. 1, July 1993
General News
Next Release 93NOV
The next release of MIDAS, release 93NOV, is
now in the ``alpha'' state. That means that our cur­
rent development version has been frozen in mid
July and will be made available for internal testing
at ESO with the name 93N08 until mid­Septem­
ber. After this period it will become the ``beta''
release. Tapes containing this release will be dis­
tributed to the beta­test sites.
Reports from beta­test sites are expected to come
until the beginning of November after which the
official release of MIDAS will be distributed to
those sites whose request forms for the 93NOV
have been received.
The 93NOV release of MIDAS will be installed
and verified on the following machines/systems:
. SUN/SunOS 4.1.n and SUN/Solaris 2.1;
. HP/HPIX;
. SG/IRIX;
. IBM/AIX;
. CONVEX;
. VAX/VMS;
. VAX/Open­VMS;
. PC/Linux;
. PC/SCO.
New ``anonymous ftp'' account for
MIDAS
The anonymous ftp (File Transport Protocol) was
first announced in the MIDAS Courier Volume 1,
Number 2, December 1991. Since then, the
ESO­IPG has provided the MIDAS user commu­
nity with software and information related to
MIDAS (e.g. copies in PostScript format of the
MIDAS Courier), using this ftp account.
This service has proved to be very successful, as
indicated by the number of connections received
every day. Because of its speed and comfort, the
ftp service has already been selected by many
MIDAS sites to check and retrieve new releases
and patches for MIDAS.
Our Internet name is still the same
``ftphost.hq.eso.org'' or IP address 134.171.8.4.
You can enter ``anonymous'' or ``ftp'' as username
and give your name or e­mail address as pass­
word. Under the directory ``/midaspub'' you will
find all public domain information about MIDAS,
including patches and sources for most MIDAS
packages. The ``core'' of MIDAS is, however, not
available under this directory. It is restricted to
MIDAS sites with a valid User Agreement.
Taking effect the first of September we will
upgrade our ``ftpd'' (File Transport Protocol dae­
mon) using a new release of the software. How­
ever, together with this upgrade it will be
necessary to change our method of providing
access to the restricted MIDAS. In our previous
modified version of the ``ftpd'' daemon we imple­
mented a restricted ``/midas'' directory that con­
tained the complete compressed­tar sources of
MIDAS. It was necessary to login as ``ftp'' or
``anonymous'' first, and then to sublogin with the
``user midas'' command, which required a pass­
word, to get access to the ``/midas'' directory.
With the new ``ftpd'' daemon the policy remains
the same, but the restricted sublogin to ``/midas''
disappears. A new ftp account called ``midas''
replaces it. With this new approach the generic
``ftp'' or ``anonymous'' account grants you ftp
access to public ESO directories like ``/midaspub''
and others. On the other hand, the ftp ``midas''
account is restricted to only those activated hosts
provided with a valid password. Those sites that
have been already registered with the old method
will be properly informed about the new changes
and they will be provided with new passwords to
gain access to the new restricted ``midas'' account.
New sites that want to obtain the new official ver­
sion of MIDAS 93NOV via ftp, please contact
Resy de Ruijsscher to activate your site and to
provide you with the corresponding password.
Below you will find two examples. The first one
shows how to access through the generic ``ftp'' or
``anonymous'' account to public directories at
ESO. The second example shows how to get
General, System and Application News
The MIDAS Group, European Southern Observatory

European Southern Observatory
Newsletter of the MIDAS Users' Community
MIDAS Courier
Volume 3, Number 1 July 1993
In this ESO­MIDAS Courier:
General, System, and Application News
ESO­MIDAS available on PC/Linux
Modification of Type Definitions
DO: The New Data Organizer Context
Deadline for the next ESO­MIDAS Courier
A New MIDAS CCD Context
Handling HST Data in MIDAS
WAVELET: a New MIDAS Context
Expeditions in the MIDAS Jungle: Computing
ESO­MIDAS Request Form
The number of
MIDAS users will
probably increase
considerably in the
near future, now that
ESO­MIDAS has suc­
cessfully been ported
to Personal Comput­
ers running the Santa
Cruz Operating Sys­
tem (SCO­UNIX), or
the public domain
UNIX Linux (the
most popular UNIX
system available on
the network). A sepa­
rate article describes
the port of MIDAS to
Linux.
Now
over
200
sites Editorial
In this ESO­MIDAS Courier you will find information about the coming
release of ESO­MIDAS, the 93NOV release. Besides descriptions of new
applications, you will find important news about system related subjects like
the use of shared libraries, the port of ESO­MIDAS to PC's and Graphical
User Interfaces (GUI's). Following the ESO Common Conventions, the
XHelp has been rewritten while other new GUI's have been added, one of
which is the GUI for the display. Other interfaces are now being planned and
will follow in the near future. All new GUI's will be based on OSF/MOTIF. In
addition to the new application packages developed in­house, you can also
read about the new HST context written at the Max Planck Institute for Extra­
terrestrial Physics in Garching. The ESO­MIDAS Group gratefully acknowl­
edges this contribution.
As can be inferred from this cover page ESO­MIDAS is now running at more
than 200 sites. This number was reached this month: User Agreement number
93A200 was signed by the Astronomical Institute of the Wroclaw University,
Poland on July 28. By the time this Courier is in print the number of User
Agreements already increased to 203. The distribution of sites is as follows:
154 in Europe, 16 in Asia­Pacific, 22 in North America, 9 in Latin America,
and 2 elsewhere. Clearly, with such a large user community, efficient support
may become difficult. Therefore, we were glad to learn that the vast majority
of our users present at this years MIDAS Users Meeting were still satisfied
with the support they obtain from the ESO­MIDAS Group.