est_noise vs. cats

Which one is faster? Which one is better?

Need your attention.

An alternative spectral analysis tool: FAMOUS.

Frequency Analysis Mapping On Unusual Sampling


ftp://ftp.obs-nice.fr/pub/mignard/


What does SNR of the results mean?

Got est_noise running in virtual solaris box.

Previously, est_noise by Langbein compiles successfully but quit when run it. The only error message was "Killed"--nothing else. This proves to be the problem of large expense of ram--est_noise requires 1.7+GB memory to run. I found that the maximum number of observations is hardwired into the fortran code (8103). Considering my CGPS time series are not so long. I changed this threshold to 5002. It works! Great.

买票难

北京站, 为什么把票先打出来再慢慢地卖?难怪代售点没有票.

Still confused with est_noise output.

What are those terms in model.dat file?

I learned the meaning of max.dat from the source code. But the model.dat does not give me much information. I should dip more into the code.

Searching for alternative browser for Solaris X86.

The firefox is good indeed. However, I wanna try more browsers.

Can I input Chinese characters in English GUI?

The hot key cannot call out the Chinese input method toolbar. Can I?

Run English version Firefox in Chinese GUI.

If you choose Chinese GUI for solaris, the Firefox will start with Chinese menu and buttons. It is started with more parameters,

> ps -ef | grep firefox
tianyf 16040 15961 0 08:17:58 pts/14 0:00 /bin/sh /usr/bin/firefox
tianyf 16054 16040 0 08:17:58 pts/14 0:00 /bin/sh /usr/lib/firefox/run-mozilla.sh /usr/lib/firefox/firefox-bin -UILocale
tianyf 16060 16054 1 08:17:58 pts/14 0:31 /usr/lib/firefox/firefox-bin -UILocale zh-CN -contentLocale CN

What if we start only /usr/lib/firefox/firefox-bin ?
Ha.. It's English menu in Chinese GUI now. Good.

est+ from est_noise

#!/bin/sh
#  sample script to run est_noise6ac
#  test data is CARH north from Tom Herring's 'combined' SCIGN solutions


rm -f seed.dat
echo 82789427 > seed.dat     # create a random number for doing hypothesis testing a la Langbein [2004] Figure 4

rm -f junk.in
cat > junk.in <otr                  # format style of inputting specifications (using julian day style; year, month and day could be input, too)
1                   #  Number of data sets  (almost always 1)
1995 1 2005 1       # time span to be considered
y   # rate             #  Estimate the secular rate; almost alway 'y' for yes
0   # num of rate change; If you think there are any rate changes in data, this should be the number of rate changes;   if non-zero, then you'll need to supply the time of the rate change
1  # num of periods;   I'm solving for a 1-year periodicity in the data; this is NOT a band-pass filter
365.25         #  period in days
1              #  Number of offsets in data
2003 357.0     # time of offset due to San Simeon EQ (M6+)
0             # number of exponential exp(-t/tau) or Log (1 + t/tau) or Log (tau + t ) terms for characterizing post-seismic responce. This seems to work okay; outputs searchs for optimal tau on file fort.89
otr      #  format style of input data
CARH.n 
0    #  Number of time-series of auxillary data --- This is useful for analyzing Dilatometer data -- this would be the presure time series
1   # mininum sampling in days  (usually 1 day for CGPS)
n   # This is usually n for 'no'.  With 2-color EDM stuff, sometimes I'd have more than one measurement on a single day and I'd want to combine these data
n  # n for 'no' unless you want to do simulations for hypothesis testing (Figures 2-4 in Langbein, 2004)
0   #  decimation parmeter; choice of 0, 1, 2, or 3.  Use 0 for 'best' results as it uses all of the available data; the other choices toss-out observations but are useful for 'quick and dirty' results-- more exploritory in nature
1 float    #white noise  Initial guess of white noise is 1 mm and you are asking the program to estimate the white noise amplitude
1 float   #power law amplitude; initial guess of PL amplitude is 1 and you are asking the program to estimate a better amplitude
1 float    # power law index; initial guess of PL index is 1 (flicker) and you are asking the program to estimate a better index
0 fix   #Gauss Markov;  fixing the GM time constant to 0; do not make any estimates
.5 2.   # Pass-band limits of the band-pass filtered noise  0.5 cycles/year to 2 cycles/year
1      #  Number of poles in the BP filter; (goes 1, 2, 3, 4, 5, and 6)
0 fix  # Amplitude of BP noise; in this example, it is set to 0 and not estimated
2 fix   # power law index;  Index for a second power law; it is set to 2 (random walk) and not updated in this example
0.0 fix  # power law amplitude; amplitude for a second power law; it is set to 0 and not updated in this example
0  #  Additional white noise to add to data (data.in); usually set to 0 but for cases such as strain or creep data where the power-law index is > 2.5, it becomes critical to add some white noise to the data in order to keep the data covariance from becoming singular.
EOF

time ~/proglib/Strain/Progs/est_noise6ac < junk.in

exit
EOF

Xmanager took up so much CPU and network.

My HP notebook keeps running with high speed fanning. The network usage was usually 20-30%, and the CPU usage about 15%.

crontab cannot find the searching path.

It seems that cron doesn't load the login script (say, ~/.profile); thus, I should manually source it at the first part of cron script file.

No cpu_time() function in f77 of sun studio.

The est_noise6ac.f program failed to compile with f77 compiler of Sun Studio 12. It prints an error saying something about a missing function "cpu_time()" which may be a f90/gcc function. This program was finally compiled with f90 of Sun Studio 12, with the sun performance library (-xlic_lib=sunperf).

The est_noise program took about 1700+MB ram when running.

crontab -e problem for new user in solaris

I found that the "crontab -e" command cannot work for the newly created user "datadm".
It prints "0" and "?", then seems hanging there without ending. It turns out to be the problem of default editor setting. Set it to "vi" will solve this problem. Now, put the following lines input your .profile if you are using "sh".

EDITOR=vi
export EDITOR

That's all. You can now use "crontal -e" to create scheduled tasks.

Large difference of speed between solaris and rhel-5 when running est_noise

The same input data and parameters, it took solaris (dell precision 390 with 2GB RAM) 6 min to complete. While rhel-5 (dell optiplex 755 with 4GB ram) obtained results within fifty-six seconds.

Why so large difference in speed? Is the solaris not suitable for heavy load of numerical computations? Or the difference of RAM matters?

打酱油去。。

算了，快完了，打酱油去了。
顺便做做俯卧撑。。

Source error for sh of solaris.

$ . .shrc
GPSMET_DC_HOME=/export/home/tianyf/gpsf/external/gpsmet_dc: 不是一标识符

$ which sh
/usr/bin/sh

$ uname -a
SunOS gpsdc 5.10 Generic_120012-14 i86pc i386 i86pc

$ more .shrc
GPSMET_DC_DATA=/data0/igs0/pub; export GPSMET_DC_DATA
export GPSMET_DC_HOME=/export/home/tianyf/gpsf/external/gpsmet_dc
export PATH=$PATH:$GPSMET_DC_HOME/sh

Where is the problem? This script works well with bash.

最近无法静心。。

两天来，被贵州那个什么事件搅的无法安心写paper了。持续关注中。。。

疑惑：
1 这种事情有什么好遮遮掩掩的？秉公办理不就行了。否则难逃存在腐败、失职之说。
2 把上万群众说成暴徒。老百姓没事干去政府那干啥？怎么能把责任推给“少数别有用心”、“黑社会”分子？
3 定调不对。说什么“稳定压倒一切”。我们要的是公平、正义。

中国这一锅粥到底糊成什么样了？

切换输入法in xmanager

当通过xmanager连接solaris时，使用ctrl+space切换输入法，结果切换的确实windows的输入法。怎么切换xmanager里solaris的输入法？

1。右键。
2。工具-输入法首选项
3。触发键-添加-（如ctrl+f1）

Done。

还可以设置用，和。翻页----很方便。不错。

Use Chinese fonts in Xmanager.

I got Xmanager to support Solaris Chinese GUI. Haha..

1. Download /usr/openwin/lib/locale/zh.GBK/X11/fonts/75dpi/ to C:\Program Files\NetSarang\Xmanager Enterprise\Fonts. You'd better rename it to 75dpigbk because there is already a direcoty call 75dpi.
2. In Xconfig, configure the default profile to add the 75dpigkb to font searching paths.
3. Restart Xmanager to take effects.

How, at the login screen of solaris, select zh_GBK will work.

It's great. Isn't it?

I switched my internet connection from winxp nootbook to Solaris X86 workstation today, it's important to type Chinese characters in solaris sometimes.