Bienvenue !

A Vue d'Aile est une association loi 1901 crée dans le but de diffuser le patrimoine naturel, architectural, culturel et historique régional, au moyen de photos aériennes. Nos pilotes utilisent des modèles réduits pour prendre de spectaculaires photos aériennes qui, nous l'espérons, vous permettrons de décorer agréablement votre intérieur ou votre bureau.

Moved

I've moved my blog to marckossa.over-blog.com.

Mon blog se trouve maintenant sur marckossa.over-blog.com.

My laptop's speed

Results of LinPack test (used to classify HPC in the top500 list).

Linpack SSE2 Double Precision Unrolled Benchmark n @ 100
Via Microsoft 32-bit C/C++ Optimizing Compiler Version 15.00.30729.207 for 80x86
Tue Sep 08 09:28:26 2009

Speed 1053.34 MFLOPS

Numeric results were as expected

CPUID and RDTSC Assembly Code
CPU GenuineIntel, Features Code BFE9FBFF, Model Code 000006E8
Genuine Intel(R) CPU T2500 @ 2.00GHz Measured 1995 MHz
Has MMX, Has SSE, Has SSE2, Has SSE3, No 3DNow,
Windows GetSystemInfo, GetVersionEx, GlobalMemoryStatus
Intel processor architecture, 2 CPUs
Windows NT Version 5.1, build 2600, Service Pack 3
Memory 3328 MB, Free 2046 MB
User Virtual Space 2048 MB, Free 2042 MB

On a 2 cpu virtual machine hosted on a X4150 Sun server, with eight 3 GHz Xeon processors, the speed is :

Linpack SSE2 Double Precision Unrolled Benchmark n @ 100
Via Microsoft 32-bit C/C++ Optimizing Compiler Version 15.00.30729.207 for 80x86
Tue Sep 08 10:32:19 2009

Speed 1826.67 MFLOPS

Numeric results were as expected

CPUID and RDTSC Assembly Code
CPU GenuineIntel, Features Code 0FEBBBFF, Model Code 00010676
Intel(R) Xeon(R) CPU E5450 @ 3.00GHz Measured 2993 MHz
Has MMX, Has SSE, Has SSE2, Has SSE3, No 3DNow,
Windows GetSystemInfo, GetVersionEx, GlobalMemoryStatus
Intel processor architecture, 2 CPUs
Windows NT Version 5.2, build 3790, Service Pack 2
Memory 3840 MB, Free 3223 MB
User Virtual Space 2048 MB, Free 2042 MB

SNMP Monitoring for GlassFish

Sun is delivering important addons to the GlassFish J2EE server. One of these add-ons is the SNMP Monitoring feature, which I'd like to discuss briefly here. Being part of the technical team that delivered the feature, I have to say that I was impressed by the overall way it was handled. It was released, not when the date set by marketing 8 months ago was rearched, but when all important sub-features were implemented, and when there were no major bugs left (and not because important bugs were closed as "not reproducible", but because code was actually fixed). It happened to correspond to the marketing date, but that's only because of the hard overtime work done of some team members. Great work !!

Here are some examples of what can be found out about a running GlassFish server, using some simple snmp requests. snmpwalk is a standard Unix command widely available.

In SNMP, indexes are used to walk through a hierarchy of information. In the J2EE MIB, they are used this way: ""



Here are some sample commands:

#
# Let's set snmpwalk options once for all
$
$ OPT="-m +J2EE-MIB:ALL -M +/tmp/:/etc/sma/snmp/mibs -c public -v 1 $HOST:10161"
$
# What domain do we have in the deployement ?
$ snmpwalk $OPT J2EE-MIB::j2eeDomMoName
J2EE-MIB::j2eeDomMoName.1 = STRING: "domain1"
#
# We have one domain (index 1). What are the servers in this domain ?
$
$ snmpwalk $OPT J2EE-MIB::j2eeSrvMoName.1
J2EE-MIB::j2eeSrvMoName.1.1 = STRING: "cluster=cl1,name=cl1_ins1"
J2EE-MIB::j2eeSrvMoName.1.2 = STRING: "name=server"
J2EE-MIB::j2eeSrvMoName.1.3 = STRING: "name=sa_ins3"
J2EE-MIB::j2eeSrvMoName.1.4 = STRING: "cluster=cl1,name=cl1_ins2"
#
# We have 4 servers. What are their states ?
$
$ snmpwalk $OPT J2EE-MIB::j2eeSrvSMState
J2EE-MIB::j2eeSrvSMState.1.1 = INTEGER: stopped(3)
J2EE-MIB::j2eeSrvSMState.1.2 = INTEGER: running(4)
J2EE-MIB::j2eeSrvSMState.1.3 = INTEGER: stopped(3)
J2EE-MIB::j2eeSrvSMState.1.4 = INTEGER: stopped(3)
#
# Only one is running (index 2). What are the applications deployed in it ?
$
$ snmpwalk $OPT J2EE-MIB::j2eeAppMoName.2
J2EE-MIB::j2eeAppMoName.2.3 = STRING: "name=conapp1,server=server"
J2EE-MIB::j2eeAppMoName.2.4 = STRING: "name=__ejb_container_timer_app,server=server"
J2EE-MIB::j2eeAppMoName.2.5 = STRING: "name=MEjbApp,server=server"
J2EE-MIB::j2eeAppMoName.2.6 = STRING: "name=__JWSappclients,server=server"
J2EE-MIB::j2eeAppMoName.2.11 = STRING: "name=webapp1,server=server"
#
# There are 5 applications. Let's take the last one (index 11). When was it started ?
$
$ snmpwalk $OPT j2eeAppSMStartTime.2.11
J2EE-MIB::j2eeAppSMStartTime.2.11 = STRING: "Fri Feb 06 10:32:33 MET 2009"
#
# What modules are deployed in it ?
$
$ snmpwalk $OPT J2EE-MIB::j2eeModMoName.2.11
J2EE-MIB::j2eeModMoName.2.11.14 = STRING: "application=webapp1,name=//server/webapp1webmod2,server=server"
J2EE-MIB::j2eeModMoName.2.11.15 = STRING: "application=webapp1,name=//server/webapp1webmod1,server=server"
#
# There are 2 modules in this application. What are they ?
#
snmpwalk $OPT J2EE-MIB::j2eeModType.2.11
J2EE-MIB::j2eeModType.2.11.14 = INTEGER: web(3)
J2EE-MIB::j2eeModType.2.11.15 = INTEGER: web(3)
#
# They are web modules
# Let's take the first module (index 14). When was it started ?
$
snmpwalk $OPT J2EE-MIB::j2eeModSMStartTime.2.11.14
J2EE-MIB::j2eeModSMStartTime.2.11.14 = STRING: "Fri Feb 06 10:32:33 MET 2009"
#
# How many servlets are in it ?
$
$ snmpwalk $OPT J2EE-MIB::j2eeSletMoName.2.11.14
J2EE-MIB::j2eeSletMoName.2.11.14.36 = STRING: "application=webapp1,name=webapp1webmod2_Servlet2,webmodule=//server/webapp1webmod2,server=server"
J2EE-MIB::j2eeSletMoName.2.11.14.37 = STRING: "application=webapp1,name=jsp,webmodule=//server/webapp1webmod2,server=server"
J2EE-MIB::j2eeSletMoName.2.11.14.38 = STRING: "application=webapp1,name=default,webmodule=//server/webapp1webmod2,server=server"
J2EE-MIB::j2eeSletMoName.2.11.14.39 = STRING: "application=webapp1,name=webapp1webmod2_Servlet1,webmodule=//server/webapp1webmod2,server=server"
#
# There are 4 servlets.
# On this server, how many resources are deployed ?
$
$ snmpwalk $OPT J2EE-MIB::j2eeRsrcMoName.2
J2EE-MIB::j2eeRsrcMoName.2.1 = STRING: "application=MEjbApp,name=bean-pool,ejb-module=mejb.jar,server=server,stateless-session-bean=MEJBBean"
J2EE-MIB::j2eeRsrcMoName.2.2 = STRING: "name=keep-alive,server=server"
J2EE-MIB::j2eeRsrcMoName.2.3 = STRING: "name=MConnectorPool,server=server"
J2EE-MIB::j2eeRsrcMoName.2.4 = STRING: "name=orb.threadpool.thread-pool-1,server=server"
J2EE-MIB::j2eeRsrcMoName.2.5 = STRING: "name=transaction-service,server=server"
J2EE-MIB::j2eeRsrcMoName.2.6 = STRING: "name=eis/ConnectorMonitoring"
J2EE-MIB::j2eeRsrcMoName.2.7 = STRING: "name=connection-queue,server=server"
J2EE-MIB::j2eeRsrcMoName.2.8 = STRING: "name=__TimerPool,server=server"
J2EE-MIB::j2eeRsrcMoName.2.9 = STRING: "name=jdbc/__TimerPool,server=server"
#
# There are 9 resources. Of what type are they ?
#
snmpwalk $OPT J2EE-MIB::j2eeRsrcType.2
J2EE-MIB::j2eeRsrcType.2.1 = INTEGER: other(1)
J2EE-MIB::j2eeRsrcType.2.2 = INTEGER: other(1)
J2EE-MIB::j2eeRsrcType.2.3 = INTEGER: other(1)
J2EE-MIB::j2eeRsrcType.2.4 = INTEGER: other(1)
J2EE-MIB::j2eeRsrcType.2.5 = INTEGER: jta(6)
J2EE-MIB::j2eeRsrcType.2.6 = INTEGER: jca(9)
J2EE-MIB::j2eeRsrcType.2.7 = INTEGER: other(1)
J2EE-MIB::j2eeRsrcType.2.8 = INTEGER: other(1)
J2EE-MIB::j2eeRsrcType.2.9 = INTEGER: jdbc(3)
#

This SNMP MIB will be of great help to administrators who need an easy and efficient way to access J2EE-related information, when monitoring and managing GlassFish deployments.

Sun GlassFish Portfolio Landing Page: http://www.sun.com/glassfishportfolio
GlassFish (Community) Homepage: http://glassfish.org
Sun GlassFish Enterprise Server Homepage: http://www.sun.com/software/products/appsrvr/index.xml

M31

34 exposures of 100s each, with Canon EOS 350D and Canon 85mm f/1.8.














The goal was also to image "R And", my favorite variable star.

Color balance with EOS 350D

I photographed a white piece of paper in full sunshine, at noon, in raw mode.
I cropped the center of the image, and developped it in 48 bits. Here are the stats reported by Iris on the center part:

Red: Moyenne : 667.6 Médiane : 671 Sigma : 14.6 Maximun : 742.0 Minimum : 413.0
Green: Moyenne : 1294.5 Médiane : 1310 Sigma : 35.4 Maximun : 1481.0 Minimum : 508.0
Blue: Moyenne : 1106.0 Médiane : 1114 Sigma : 26.0 Maximun : 1245.0 Minimum : 540.0

Taking the median, the RGB adjustments should be:

R: 1.95 (1310 /671)
G: 1
B: 1.18 (1310/1114)

or (reversed, to avoid overflowing on integers):

R: 1
G: 0.51 (671/1310)
B: 0.60 (671/1114)

To pretrait or not to pretrait, that is the question...

When doing astronomical images, it is sometimes better NOT to do pre-processing. When the number of dark frames is small for example, as it adds more noise to images than it can remove.

Here are some discussions about this.

Here are 2 images, obtained from the same raw pictures: 30 30-seconds images of m31, taken with Canon 85mm f/1.8.





This one has no pre-processing: hot pixels are visible as small red, green and blue strings (click on it to see it larger).














This one had full pre-processing: dark, flats and offset. When viewed full-size, it doesn't have more signal (it probably actually has less), and overall is doesn't look as good. I only had 1 dark frame to use, as well as only 1 offset and 1 flat field.















For comparision, this one has no pre-processing, but only 10 images stacked together (3 times less). The signal/noise ration is clearly less than the first one with 30 images.

XInclude, jaxp and xerces

XInclude defines a powerful way to control inclusions of xml fragments in another xml file:

<xi:include
    xi="http://www.w3.org/2001/XInclude"
    href="source.xml"
    xpointer="xpointer(string-range(chapter/p[1],'Sentence 2')/range-to(string-range(/chapter/p[2]/i,'3.',1,2)))">
</xi:include>

This looks good, however in reality none of the current parser implementations support that ! For example, Xerces (implemented in jaxp 1.3 & 1.4) only supports xpointer="element(...)". This makes it almost useless, unless a whole tree has to be included. It is not possible to easily merge two xml files this way.

Frédéric Laurent already noticed this 3 years ago... It's time for a Java expert to implement it in Xerces !

In the backyard...

Sunday afternoon at home... Roe Deer ("Chevreuil").



Always a magical moment...

Aerials Photo over Meylan, France

Here are some recent Aerial Photos taken in the Meylan area:





More pictures can be seen on my RC Blog.

Reticule eyepiece and resolution

Here are the corresponding arc measures when the 12.5 mm Baader reticule eyepiece is used with the Celestron 8 (no focal reducer). For example, the small circle represents 15 arc seconds of diameter.
15 seconds is also the size of one pixel on the EOS 350D with the 85mm lens.

Virtualization is not.

Creating a "virtual machine" with VirtualBox or VMware is misleading, no "virtual machine" is created, only a virtual view of the real underlying hardware, which must be physically present. Usually, "virtual" refers to something that doesn't exist at all, which is not the case here.

In comparison, platforms emulators are real virtual machines. For example, MAME can emulate hardware that doesn't physically exist anymore. In that sense, it can really create a "Virtual Machine".

Satellite galaxies of M31

It is little known that M31, the Great Andromeda Galaxy has not 2 but 4 satellites:

M110 (NGC 205), the ellongated one north of M31 (mag 8.9)
M32 (NGC 221), the round galaxy south of M31 (mag 9.0)

NGC 185 (mag 10.1)
NGC 147 (mag 10.5)

I was able to capture all 4 of them in this simple picture, a composite of 12 individual images taken with the camera piggybacked to my C8, with no guiding. No preprocessing was done, and the frame when a plane crossed the field was also included... Image still goes to magnitude 16.5 !



Right above the white border, on the top part, the variable star VX And is easily located, due to its vivid red color.

Compact camera and Infinity focus

I usually set my Pentax A20 to infinity focus when doing Aerial Photography. However, I just discovered that it just isn't right:





Here is an extract from a distant scene taken with Autofocus: it is sharp.











Same scene taken with focus set to Infinity: completely blurred !




Hyperfocal mode gives bad results as well !

Barnard's Star

(voir ici pour une traduction automatique de cette page en français)


This is the star with the highest proper motion of the Heavens: more than 10" every year. See the Wikipedia page on this star here.

I took this picture yesterday, with my Canon 350D DSLR piggy-backed on the C8, with the 85mm f/1.8 lens. Single 30 seconds exposure, no processing except contrast enhancement, and double size for clarity. It is possible to see stars of magnitude 14 on this picture...

The bright star on the left is 66 Oph, magnitude 4.8. It is highly overexposed, and appears less bright than it should be. This is due to the limited dynamic range of the Canon EOS 350D captor . Its design range is 12 bits (4096 values) per pixel. 1 bit is lost in the electrical noise (seen as the sigma value of a dark frame), and 2 more in the suburban sky background. So only 9 bits remain, giving a dynamic range of 512 values, which corresponds to log_2.5(512) = 7 magnitude range. As stars of magnitude 14 can be detected, everything above magnitude 7 is overexposed, and appears dimmer than it should be... This is not a problem, since only 66 Oph is brighter than mag 7 here (next bright star is 7.6).

Field of view and true magnification of Canon binos

Here are my measurements.

Canon 18x50 IS:

• The apparent field of the bino is 62 degres (from Polaris to over Alpha And).
• The real field is 3.85 degres, measured to be midrange from Delta Cas to Gamma Cas (3.60 degres) and Delta Cas to SAO 11444 (4.13 deg).
• Thus the real magnification is 62/3.85 = 16x, not 18x !

Canon 10x30 IS:

• The apparent field is 48 degres (from Polaris to M31).
• The real field is 6 degres (from Beta And to 2/3 between Mu And to Nu And).
• Thus the real magnification of the 10x30 IS binos is 48/6 = 8x, not 10x !

I wonder if this is standard practice among bino manufacturers ?

Observation of Variable and double stars

Yesterday I was able to observe some interesting stars using the Canon 18x50 IS binos. These observations were made on Sep 22 at 20h30 UT.

Delta Cephee

This is a variable star and a double star. Blue compagnon (Mag=6) was clearly resolved, at 41". As variable is orange, contrast is beautiful. Variable star magnitude was equivalent to white Epsilon: 4.2.

According to CalSky, the minima of this star was on Sep 17 at 22.9h UT. As the period is 5.36634 days, the next minima was on Sep 23 at 7h40 UT. My observation was then 0.47 days before minima, very close to minimum brightness.

Update:
Sep 23rd, 20UT, 0.5 days after minima, 4.2.
Sep 24th, 20UT, 1.5 days after minima, about 3.7.




Beta Lyra (Sheliak)

Its luminosity was equivalent to Gamma Lyra (=3.2).
Mimima: Sep 13th at 18.3h UT. Observation at Minima + 9.09 days, or next Mimina - 3.82 days : near maximum.

Update:
Sep 24th, 20UT, 12 days after minima, about 3.6.







Zeta Lyra

This is a double star (47"), clearly visible.

Albireo

Sep 23rd: Easily resolved (35").

Some old pictures

Spring 1977, Grenoble.
60x700 refractor.










Spring 1986, col de l'Arzelier.
Ready to watch Halley's comet !





Mauna Kea, Hawaii, December 1996.












Mt Palomar, Mai 1997.








A more modest observatory near Grenoble.

Old Astro pictures

Here is M57 in Lyra, taken on June 4 1996 from Austin, Texas. 6 minutes single exposure, on Kodak 1000 ISO film. Olympus OM-1 camera, Lumicon Easy Guider.

Here is also M57, taken 11 years later with a smaller telescope, with a single 30 "exposure...on a digital camera (Canon EOS 350D).

This is M13 in Hercules, also 6 minutes single exposure, from Austin.

Comet Hyakutake, near Austin, Texas, April 1996.

Supernovae.

Dim comet.



Edge-on galaxy.

How does the Canon 18x50 IS binoculars compare to the 10x30 IS ?

(Voir ici pour une traduction automatique de cet article en Français)

After searching for years (literally), I was finally able to pick up a good deal on Canon 18x50 IS binoculars out of EBay (thanks to the cheap $!). I have been using the smaller 10x30 IS for 7 years, and always found them very good. But I wanted (of course) bigger magnification, and more brightness.
Here are my impressions after a few weeks'use.




















The optical quality is near-perfect, no problem even on the edge of the image. The stabilization is very good, better than on the 10x30: it starts faster, in less than a second images are rock solid. The button is of the "press to start, press to stop" type : no need to keep it pressed as with the 10x30, this is very convenient.
The eye cups are not as good though. They are too big, and don't fit my eyes, I had to remove them. The 18x50 IS are much heavier than the 10x30, but they are perfectly manageable.

The magnification gain is amazing. I am now able to see much details in passing airliners, that I was not able to see with the 10x30.

The brightness is on par: I can see stars of magnitude 10.5, while with the 10x30 I was limited to 9.5. This indicates they are 2.5 brighter, which is consistent with theory (50*50/30*30 = 2.8).

Watching the moon is amazing: it appears as a big 3D ball. Much details can be seen, such as Mons Pico (even with no shadow), a mountain range only about 20 km wide.

The image hand-help is the same as on a tripod, this is simply amazing.

From time to time, the image appears a little bit blurred or fuzzy. This is probably due to high-speed vibration in the image, caused by the shifting lens correcting shaking. As magnification is high, it is visible from time to time.

On a tripod, with stabilization on, the image appears to slowly drift in one random direction, for about two seconds, than reverses direction, and the cycle starts again in another random direction. I would be interested to know if other owners experience the same thing ?

Overall, these binos are amazing.