The LIGO Executive Committee Agenda for Monday August 9, 2004 will be:
(Meeting time: 10:30 am Pacific Time)
Special Items:
Minutes of LSC Executive Committee telecon
Friday 30 July 1100 Eastern
Present: Bruce Allen, Patrick Brady, Sam Finn, Joe Giaime, Albert Lazzarini,
Dave Reitze, Peter Saulson (minutes), Bernard Schutz, David Shoemaker, Ken
Strain, Stan Whitcomb Guest: Peter Shawhan (Pulsar reviewer)
Results of discussions with GEO at the GWIC meeting:
a) From now on, a representative of GEO management will serve ex officio on the
LSC Executive Committee. Bernard Schutz joined the Exec Comm in this capacity.
b) There was an agreement in principle to revise the GEO/LIGO MOU to better
reflect the way we've been working together on data analysis. The basic idea of
the MOU is that, for purposes of data analysis, we have one joint data set from
the 1 GEO and 3 LIGO interferometers, and one data analysis team consisting of
all LSC members. Specific language will be drawn up into MOU form.
A question was raised about how to connect this negotiation with
the Virgo MOU, which is close to being signed. It was decided to let the present
language about GEO (that the MOU doesn't cover GEO data) stand as written, to
avoid slowing up the signing of the MOU. Then, when we have signed the new
GEO/LIGO MOU, we can make a simple amendment to the Virgo MOU (and the AURIGA
MOU, too.) We should let Virgo and AURIGA have advance notice of the direction
of these discussions between LIGO and GEO.
AURIGA/LIGO MOU signed at GWIC meeting:
Now, the AURIGA/LIGO joint working group (Cadonati and Prodi,
chairs) will start work on proposing details of joint analysis.
Progress on Virgo MOU and joint analysis plan:
The drafting of the Virgo/LIGO MOU has almost been completed.
Meanwhile, the joint data analysis team has drafted a White Paper, laying out
the first collaborative steps. That draft has now been circulated to LSC and
Virgo scientists.
There was some discussion about whether the MOU's provision for
keeping the Virgo and LSC data analysis teams as distinct entities is a wise
choice. Stan argued that there were several dangers: 1) Data from the
"other" experiment could be treated without proper regard for
instrumental problems, if the analysis is done without strong participation
from people who know the instrument, and 2) If the teams stay very separate,
competing interpretations of the same data could harden before a chance to
compare notes and reconcile ideas. Several other members of the LSC Exec Comm
shared these worries. However, keeping separate identities of LSC and Virgo
analysis teams is a basic feature of the MOU. We need to think about how to
mitigate these problems. There will be time at the LSC meeting to discuss these
issues.
Einstein@Home proposal declined:
Bruce Allen reported that the NSF has declined to fund the
Einstein@Home proposal. Nevertheless, he is confident that the team will be
able to accomplish its key goals. On the scientific side, the code is working
now. There has also been progress on the screen-saver code. APS is committed to
supporting the project even without external funding; they are making available
people skilled in website design and publicity.
Should the S2 Pulsar paper be submitted to PRL?
Peter Shawhan presented the question of whether the S2 pulsar
paper should be submitted to Physical Review Letters. He reported that the
Pulsar group and its reviewers now considered it to be in the right format and
style, and close to the required length. Most discussion centered on the
question of the importance of the result. Several readers of the current draft
feel that it needs to make a better case for the importance of the result.
Bernard Schutz pointed out that the limits on pulsars in globular clusters are
completely new, since the spindown method doesn't work for them; Sam Finn
questioned whether those pulsars represent a distinct class. Most members felt
it did indeed meet the standard of importance required by the journal, and no
one proposed a "strategic" reason to refrain from submitting.
After discussion concluded, there was a vote on the PRL
submission. The plan to submit to PRL passed by a vote of 10 - 1.
Other issues in connection with the S2 Pulsar paper:
After brief discussion, we made the following decisions:
The names of the radio astronomers who are authors should appear
in a separate block, after the LSC names.
The acknowledgment text
should include all of the standard language, including thanks to GEO funders,
in spite of the fact that GEO did not run during S2.
If necessary to save space,
we agreed that it would be allowed to delete the "currently at"
addresses for people no longer in the LSC. But that information should be left
in the manuscript sent to the journal, and only deleted later if the journal
requests it.
New post of Deputy Spokesperson?
Peter Saulson asked for advice on how to reduce the load on the
Spokesperson. There was some discussion of the idea of creating an office of
Deputy Spokesperson. One version of the idea was that the office be set up like
the leadership of, say, the APS, where one is elected as the Deputy
Spokesperson, and then after one term automatically becomes the next
Spokesperson. An argument against that idea was raised: It guarantees turnover,
but the LSC might want to keep a Spokesperson in office for several terms.
Also, there is a virtue to ensuring that Spokesperson and Deputy could work
well together. Other alternatives discussed were: allowing the Spokesperson to
name a Deputy who would serve with him/her for the duration of the term, or
having the Spokesperson delegate some of the most time-consuming particular
chores to individuals of his/her choice. Based on this discussion, Saulson
decided against proposing to the LSC Council a charter amendment that would
create the office of Deputy Spokesperson.
Possible revision to the MOU system:
Time only allowed for a very brief discussion of a possible new
scheme for replacing the current system of MOUs. Under the new scheme,
Technical Working Group chairs and Data Analysis Group co-chairs would assemble
work plans and progress reports for all of the LSC member groups associated
with their efforts. Several members pointed out that the document describing
this new idea is very sketchy, and is especially so on the question of how
these plans and progress reports are read and judged, and what feedback would
come as a result of that process. Albert proposed that the present 6-month
cycle is too short now, and might also be too short under the new system. Dave
Reitze suggested that a small committee be named to flesh out the missing parts
of the new scheme; the Spokesperson will do so.
Additional LSC Exec Comm meeting to be scheduled:
Several members asked that an additional meeting be scheduled, so
that we could have a serious discussion of the question of LSC restructuring
(as well as additional discussion of the MOU plans, if desired.) Such a meeting
will be called soon.
STATUS OF LSC MOUs
(Petrac)
(LSC Research Updates through February 2005, and Progress Reports
through August 2004)
ACIGA
Balearic
Carleton
Florida
LSU
Northwestern
Trinity
(LSC Research Plans through Aug. 2004 and Prog. Reports
through Feb. 2004)
ACIGA
AURIGA
Balearic:
IAP
LSU
National Astronomical Observatory-China (NAO-C):
Northwestern
Orsay Group (aka Groupe Virgo)
Penn State
VIRGO
Wisconsin
SITE TELECONFERENCE (Lindquist)
There was no site teleconference scheduled for Thursday, August
5, 2004.
The list of assigned actions updated through July 22, 2004 may be found Here.
PROPERTY MANAGEMENT (Chargois)
>From: Ed Chargois <chargois_e@ligo.caltech.edu>
DOCUMENT CONTROL CENTER (Turner, Mak)
>From: Linda Turner - turner@ligo.caltech.edu>
> From: Cleveland Mak <mak_c@ligo.caltech.edu>
COST SCHEDULE CONTROL SYSTEMS (Cunningham,
Brambila, Kaufman, Salone)
>From: Esther Cunningham <esther@ligo.caltech.edu>
>From: "Brambila, Ruth" <Ruth.Brambila@caltech.edu>
>From: Gina Salone <gsalone@ligo.caltech.edu>
>From: Florence Kaufman <fkaufman@ligo.caltech.edu>
Financial reports can be found at: http://docuserv.ligo.caltech.edu/~fireport.
(For passwords contact Florence)
SUBCONTRACTS MANAGEMENT (Petrac, Jasnow)
>From: irena@ligo.caltech.edu (Irena Petrac)
>From: Ed Jasnow <jasnow@ligo.caltech.edu>
SUPPORT (Baldon, Kammerling, Lloyd)
>Irene Baldon
>Sharon Kammerling
>Dorothy Lloyd
ADVANCED LIGO (Cost Schedule Control
Systems) T. Frey
>From: Thomas Frey <tfrey@ligo.caltech.edu>
See Advanced R&D Section.
REPORTS (Lindquist)
Prepared a schedule of proposals and reports planned for the remainder of
the calendar year. It looks something like:
We have a lot to do!
CHANGE CONTROL/CONTINGENCY MANAGEMENT
(Lindquist)
There are no open change requests currently in the system.
HUMAN RESOURCES (Akutagawa)
>From: Cindy Akutagawa <cindy@ligo.caltech.edu>
Quality/Safety (Tyler)
>From: Bill Tyler <tyler@ligo.caltech.edu>
No report this week.
.
Summary of Commissioning Activities at
LIGO Hanford Observatory (Landry)
4K IFO
Line noise at 120Hz and 240Hz showed intermodulation with low frequency
resonances such as stack modes, pendular resonances and the microseism.
The addition of transformers to the outputs of the RF distribution
chassis reduced
the intermodulation peaks by 2-4X. More
work on the RF distribution system took aim at 60Hz noise.
A 2V offset was input into the MC additive offset. This detuned mode
cleaner is used as an analyzer of RF sideband phase noise, as described here.
The result is that the measured noise level is above that which can be
accounted for by the crystal oscillator.RF phase noise measurements for the
week were nicely summarized here.
T2P
(i.e. torque-to-position) filters on the beamsplitter were modified to
accomodate the optics unique geometry. Q's were reduced from 10 to 3 to
prevent ringup.
Oscillator RF
AM is not limiting AS_Q.
A dark
noise measurement of the ETM acquision photodiodes was made.
The 4K IOO PSL 24.48MHz EOM was swapped in favour of a broadband EOM unit (with
directional coupler and network analyzer to tune to the appropriate frequency).
The
TCS (thermal compensation system) laser stability was investigated.
2K IFO
The 2k is locking fully again, after concerted recovery work post-vent.
ISCT7
needed to be moved, and REFL beams were subsequently aligned.
PSL tune-ups included the pre-mode
cleaner and the intensity
stabilization servo.
Big Picture: We are pushing hard on four commissioning fronts.
- The HEPI commissioning team and the IFO team are both working to demonstrate
and take low frequency displacement data from a locked X arm with full-up HEPI
isolation operating under both ends (a.k.a. the "one arm bandit"
test).
- Meanwhile the IFO team has also been working on improvements to the MC WFS,
characterizing the main IFO WFS, and lighting up the new thermal compensation
system (which works! see Frolov's entry below).
- We are doing a systematic baseline of radiated and conducted interference,
both to wrap up the RFI remediation at the X and Y ends and to prepare a
"before" baseline for the corner.
- All these tasks are under pressure to finish up in time for tearing down the
corner station electronics, now scheduled for the week of 8/15. Preparations
for the rack move are on track and wiring, power connection, and cross-connect
assembly is in full swing in the new corner station CDS high bay to get ready
for the move.
HEPI Commissioning (Giaime et al)
---------------------------------------------------------------------
The ITMx HEPI system has had its position sensor loops locked robustly and the
system identification measurements have been completed to allow sensor
correction filters to be designed and installed. This will be tried tonight and
we expect to begin testing a single locked arm with both ETM and ITM under full
HEPI control this weekend. Last week the mode cleaner controls were modified
temporarily to reduce the very low frequency laser wavelength fluctuations,
allowing improved sensitivity for a single arm at HEPI frequencies.
Chethan also resolved a mountain of software issues in the IFO and HEPI
controls; see the CDS report for details.
L1 Commissioning and TCS (Frolov)
---------------------------------------------------------------------
Started commissining the TCS. Made first heating test with PRM. The correction
was applied to a single ITM. The thermal effect on power buildup in the PRM
sideband lock is consistent with what is observed at Hanford - factor of 2
increase in power.
HEPI Engineering (Spjeld)
---------------------------------------------------------------------
- ordered ten replacement Parker Servo Valves for HEPI from Zemarc
- evaluated clearance issue of Bellows Shields on the HEPI Actuator
- designed platform for the GS13 witness sensors for HAM crossbeams
- provided additional drawings and documentation for the HEPI web-site
L1 Commissioning and HPLF (Amin)
---------------------------------------------------------------------
HPLF: IPG has indicated that they will be shipping the laser back to LLO
Friday. They are currently conducting a 48 hr burn-in to verify the
functionality of the device. I expect to be unpacking the device Monday morning
provided nothing else goes wrong. This last failure with the laser was
determined to be a manufacturer defect in the output of the fiber block.
WFS Commissioning: Last week and this week Gaby Gonzalez and I tuned WFS 1's
and WFS 2's RF Gain and digital phase offsets. I am in the process of
tuning WFS 1's LO phase adjust. Unfortunately I have run into a couple of
technical difficulties that are being solved.
General computing (Roddy)
--------------------------------------------------------------------
* Current bandwidth usage can be seen at
http://teche.ligo-la.caltech.edu/mrtg/LLO-Router/130.39.245.1_1.html
* Archived bandwidth usage can be seen at
http://teche.ligo-la.caltech.edu/mrtg/archive/
under the dated folder for the week of interest.
* Compiled and installed several more applications in the /apps folder.
Still looking at removing the old /opt from the remaining machines that are
mounting it. Need to relocate the global cshrc and create simlinks so
that user's environment stays the same.
* Installed the old version of GCC (2.95.3) outside of the normal path for when
applications will not compile with the new GCC (3.4.1).
* Working out some emacs problems. It segfaults every time it is run when
it is compiled with GCC 3.4.1.
* Evaluating Sun Management Center to see if it would be useful here. Just
spoke with the local Sun rep since he was on site. He claims that it is
pricy... Have not gotten it to install successfully yet anyway.
* Pulling some CAT6 cable in the Computer Users Room.
* Upgraded the Solsoft software which fixes several past bugs in the software.
It is much improved over the 5.3.1 version. The software also seems to
run better on the Suns than it used to. I am waiting to see how much they
are going to hit us for on our maint. contract since it expires this year.
CDS (Parameswariah)
--------------------------------------------------------------------
Fixed a few path/links bugs in the medm screens. Made sure i checked all the
changes to the cvs repository to future revisions do not lose the changes. Also
added a yellow help button to all ls, asc and sus screens and checked the
changes to cvs. Also checked Tom's updatesnap file to cvs and edited the make
file to install this file.
l1lscreflpwr processor found to be not working for a while, it had lost
connection to the motor controller. This was fixed and the save restore and
burt restore request files on various epics processors were corrected to point
to right database records.
Hepi1 and Hepi2 code with the testpoint transpose bug fixed is installed.
Breaker tripped in MSR. Two power strips in MS5 and one in MS4 are without
voltage. One 20 A breaker in the unlabeled electrical box in the back corner of
the MSR was tripped. Because of this, several key CDS systems went down,
including both frame builders and l0daqctrl. Measured the current through the
breaker that tripped was hovering around 19.5 to 20.4. So I redistributed some
of the sun machines on to the next breaker circuit and reduced the current to
about 16 Amps. After addition the other breaker went upto 12 Amps. The other
two breakers have 4.7 Amps and 8 Amps on them. I think we should take some time
and redistribute the machines/power strips to equal the load.
DAQ Controller brought back up, both frame builders started running and the DMT
machines also came back up. HEPI1, HEPI2 and l1awgl0 also had gone down.
Not sure if this went down earlier. Might have frozen when I brought alive the
reflective memory bypass switch.
Fixed bugs mentioned in the elogs this week.
1) Fixed the L1:IOO_MC_TRANS_DC signal in the L1ADCU PEM chassis. Moved the 4
pin lemo to channel 11 (software channel number 15052) - old channel 9A
(software channel number 15048) and edited the L1ADCU_PEM.ini file. The signal
is not a constant zero signal. This change requested by Rana.
2) Changed the L1SUS3.ini file to point the L1:IOO-MC_L signal to the output of
the L1SUS_MC2_MCL filter module instead of the input. This change requested by
Rana.
3) Set the PREC(Precision field) to 3 on the IO WFS DC READBACKS channels.
l1ascauxl0 processor rebooted and burt restored. Some of the channels in the
IOO-WFS loop were not being saved using burt. Fixed this and created new
saverestore and autoBurt request files.
4) Corrected the swapping of L1:IOO-MC_TRANS_HOR and L1:IOO-MC_TRANS_VERT
database records. They are what they say they are now. I have also fixed the
labels on the medm screen to match this change. - While I was at it, changed
the pico motor control database records from L1:IO_WFSX_XXXXXX to L1:IOO-WFSX_
as per our standard convention. The state code was also changed to reflect
this. This solves the long pending problem of channel IO_WFS channels not
meeting specs. l1iool0 processor rebooted and burt restored. Also fixed the
startup command as it was loading the lots of unwanted code/databases since the
IOO WFS were moved to the ASC rack.
5) PERROR, YERROR, POUT and YOUT : Found that only ETMX and ETMY readbacks to
the dataviewer were only affected for the above optical lever channels. All
other optics with optical levers do not have similar channel swapping problems.
Fixed the .ini files that route the channel to the dataviewers to fix the
channel swapping. Note: This did not affect the servo loops. Only the readbacks
to the dataviewers were affected
<B>STS2 FIR Filter routing is correct: </B>
The STS2 FIR Filter in the corner station is running on hepi1 with the MC1 HEPI
code. Hence the FIR Filter screens L1SEI_MC1_FIR_X_DETAIL.adl and
L1SEI_MC1_FIR_Y_DETAIL.adl are the only screens for FIR filter for STS2 at
corner station.
All the FIR button links on the individual HEPI Screens point to the above
screens only and do not have their own FIR screens. For the filter coefficients
to be loaded, please put them in the L1SEI_MC2.txt only.
The output of these FIR filter banks are all routed to the inputs of the
individual IIR filter modules named "OUT" on the hepi screens.
<B>Hepi1 - RM Position Sensor channels channel hopping </B>
Rich M. and Dan showed that some of the RM Pos Sen input channels on 110B were
being seen on different channels than their original channels. Signal changes
to some channels were also seen on multiple channels.
Looks like we had 110B channel hopping problem. Rebooting the hepi1 processor
fixed it.
CDS software support (Khan)
--------------------------------------------------------------------
1) Developing a perl script that would automate the epics database creation.
2) Laying out the board for the 1-WireDAQ module.
Data analysis (Yakushin)
---------------------------------------------------------------------
Data analysis:
1) DMT version of waveburst is successfully running under Condor. Together with
Sergey Klimenko tuning the new version on S2 data. Due to improvements in the
waveburst algorithm (multiresolution wavelet analysis is now used, triple
coincidence is done inside ETG) this version seems to be significantly more
sensitive than LAL/LDAS one. It also runs on MDC frames about 15 times faster
if one uses just LLO's cluster. I expect even more dramatic improvement in
performance for the production run since a single job that generates triggers
for 50 time lags would be much more computationally intensive than a single job
on MDC frames with just 0 time lag and therefore the relative Condor overhead
should be negligible and the cluster utilization should be much higher. The
amount of generated intermediate data is also dramatically reduced because the
new version does triple coincidence inside ETG (the old one did double
coincidence inside ETG and triple coincidence was applied later) and the data
is written directly to ROOT files (previously intermediate binary files were
used).
2) Preparing for burst F2F and LSC meetings. The objectives are a) to rerun S2
MDC with the new version of waveburst in order to demonstrate the improved
sensitivity; b) produce at least one set of S3 MDC frames (there is already an
order for that from the burst group); c) to demonstrate waveburst performance
on S3 MDC frames.
3) Helping Giovanni Santostasi with his pulsar DMT project.
LDAS admin:
1) Fixed the problem with missing or corrupted S2 L3 frames on LLO's cluster;
2) When trying to run waveburst at CIT, discovered some missing S2 L3 frames
there;
3) The old dataserver, that is used for backups, died. Diagnostic boot showed
problems with memory; however, replacing bad memory did not help. SUN is going
to replace the whole motherboard today.
4) Started publishing to LDR 1% data test for L1 and L3 RDS.
5) Reconfigured the spare t3-13 not to use hot spare drive since we currently
cannot test if a hot spare is good or not.
no report
no report
Simulation and Modeling
(Yamamoto)
Commissioning support ( Matt, Biplab )
Matt and Biplab are working at the LHO site to assist the commissioning. Matt
is mainly working on 4k IFO ASC. Biplab is working on (trying to help to setup
to measure) phase camera images to see the beam profile.
Simulation workshop ( Hiro, Lazz, Matt, Biplab, several from Florida, MIT,
Cascina )
There was a telephone conference discussing about the coordination of the
simulation / software tools workshops. Right now, STAIC and Interferometer
Simulation workshop mainly lead by Virgo exist. The main consensus was that
they will go independently for now with some levels of coordination of the
workshop schedules.
Mirror phase map ( Hiro, Xiao )
Hiro and Xiao are working to create mirror phase maps to be used for FFT. An
algorithm was developed to extend the measured data in the central region to
outside. There is no way to calculate the real phasemap in the unobserved
region, and this is an educated guess with smooth transition from measured to
extended region. The measured data is fit by up to 5th order Wyko terms (const,
x, y, x^2, y^2 and xy, i.e., power with astigmatism) and use it as the
estimation of the outside. Because of limited powers, they do not diverge. The
boundary out side of measured data is estimated by calculating waited average
of near by data. This generates a map which is continuous from the measured
region and is sort of blurred extension of the measured region. This is
smoothly connected ( change weight from 1 at the data boundary to 2 cm out
side) to the smooth distribution using the power term. On top of it, noise
based on the distribution in the central region is added. The systematic error
coming from the uncertainty using this extension to the unmeasured region is
done by comparing the predictions using this phase map and another phase which
uses the average value of the measured region to fill the unmeasured region.
Mach-Zehnder model ( Monica, Hiro )
Monica continued the development of the Mach-Zehnder system. She succeeded to
lock the system. Some part of the transfer functions do not match well with the
measured transfer function. It was suggested to include extra zeros and poles,
because the actual control system is analog and there can exist some poles or
zeros which are not well measured. The displacement noise measured is very
large even at 100kHz, where the simulation prediction is very small. It was
suggested it could be dominated by electronic noise or acoustic noise.
ALFI
Interface improvement ( Melody )
Modifying the ALFI bookkeeping objects to allow displaying the FUNC_X data
variables in a friendlier user interface. The dialog will show the member
variables in the table just as any parameter setting variables (instead of
having to look for them in the C++ code).
New functionality - Bundler ( Bruce )
- Completing new implementation of bundles
and bundlers
in Alfi, and making new
release.
Data Analysis Activities (Lazzarini)
Mendell;
I have been debugging and benchmarking LALFastGeneratePulsarSFTs. The code
works when lookup tables (LUTs) used for trig functions are turned off
(accomplished by setting resTrig = 0 in the input parameters) but the code
produces incorrect results when LUTs are turned on. The bug has been uncovered,
and is due to running off the start of the table. Since these same LUTs
are used in LALDemod, the accuracy of the LUTs was investigated, and the
largest fractional difference between calls to sin(x) and cos(x) and values
obtained using LUTs (with the resolution set to 64) is no more than
0.0015. (This includes the use of SMALL=1.e-9 to handle fabs(x) <
SMALL.) Bottom line: the LUTs work! The bug in LALFastGeneratePulsarSFTs will
be fixed, but since using LUTs does not seem to speed up the code I am
proceeding to use the code for StackSlide Monte Carlo Simulations with LUTs turned
off. Test show that the simulations run 6x faster than before and produce
the same results. I have added to the PULG S2 investigations page an entry on
the accuracy of using LUTs for trig functions.
Shawhan:
* Experimented with new ad-hoc simulated "burst" waveforms,
consisting of white or colored noise bursts, and "whistles" with some
frequency evolution. Some sample waveforms (and audio files!) can be found at
http://www.lsc-group.phys.uwm.edu/cgi-bin/bag-enote.pl?
nb=burs3sim&page=1
* Worked on organizing the "all-sky" burst search using the S3 data.
* Edited several sections of the S2 binary neutron star inspiral paper.
Sutton:
I've been working on a proposal for a suite of simulations for the burst group
to use in the S3 analysis. The goal is to systematically test our
sensitivity over the space of frequency, bandwidth, and duration of possible
detectable signals. I'm still working on a procedure for constructing
injected waveforms with the desired time-frequency content. For
LIGO-TAMA, I've produced the final sets of TFClusters triggers (raw and SG13
simulations), covering the full S2 H1-H2 and H1-H2-L1 coincidence
segments. Over the next day or two I will combine the playground triggers
with their TAMA counterparts for a playground result to be presented at the LSC
meeting. John Zweizig has been helping me with the development of the
long-awaited DMT calibration codes. Finally, I had a telecon for the
TFClusters review in which my draft report was discussed. Overall, that
review is in good shape.
Yakushin:
1) DMT version of waveburst is successfully running under Condor. Together with
Sergey Klimenko tuning the new version on S2 data. Due to improvements in the
waveburst algorithm (multiresolution wavelet analysis is now used, triple
coincidence is done inside ETG) this version seems to be significantly more
sensitive than LAL/LDAS one. It also runs on MDC frames about 15 times faster
if one uses just LLO's cluster. I expect even more dramatic improvement in
performance for the production run since a single job that generates triggers
for 50 time lags would be much more computationally intensive than a single job
on MDC frames with just 0 time lag and therefore the relative Condor overhead
should be negligible and the cluster utilization should be much higher. The
amount of generated intermediate data is also dramatically reduced because the
new version does triple coincidence inside ETG (the old one did double coincidence
inside ETG and triple coincidence was applied later) and the data is written
directly to ROOT files (previously intermediate binary files were used).
2) Preparing for burst F2F and LSC meetings. The objectives are a) to rerun S2
MDC with the new version of waveburst in order to demonstrate the improved
sensitivity; b) produce at least one set of S3 MDC frames (there is already an
order for that from the burst group); c) to demonstrate waveburst performance
on S3 MDC frames.
3) Helping Giovanni Santostasi with his pulsar DMT project.
Lazzarini:
Completed an analysis studying outliers that are produced when the stochastic
background search is performed using what was supposed to be an improved and
unbiased (at least much less biased) PSD estimation schema that excludes the
epoch under analysis by corss-correlation from the PSD estimation. The problem,
now, is that minute-to-minute nonstationarity in the detector noise floor leads
to misestimation of the variance expected in the cross-correlation statistic:
the cross-correlation for epoch k is associated with the average variance
estimated from epochs k+/- 1. This can be overcome by prefiltering the results
to exclude from consideration measurements made when the PSD varied appreciably
over 3 minutes (3 1-minute measurements). The filtering reduces the net volume
of usable data for S3 by 14%.
LIGO Data Analysis System
Software Systems (Blackburn)
Continued working on porting LDAS to Solaris. Was able to bring up the entire
system and run jobs not requiring the frameAPI. All jobs requiring the frameAPI
would core dump when deallocating memory associated with std::string members -
currently investigating this.
Experimented with a modified version of ObjectSpace which compiles for gcc,
SUNpro, and Intel's Compiler. The through put was dismal. On tandem-ii the
system quickly built a queue. Need to investigate what macros are being enabled
or disabled that caused the performance to drop.
Worked on getting the framecpp components to build with gcc 3.4.1 for the DMT
code. This allowed a first look of what needs changing for all of LDAS. Some
packages that LDAS depends upon currently do not compile with the new compiler:
1. Xerces (a patch is
available, but has not yet been tested)
2. STLport - Many changes are
required
3. ObjectSpace - Many changes
are required.
Continued to work on adding a threads running in each API GUI interface to the
controlMonitorAPI.
Continued working on mpiAPI node reuse bug (PR2451). Little progress is being
made on this due to extend of code changes necessary to resolve.
Continued to work on frame/file name extension specification within the
diskCacheAPI's resource file. This has been tested on Tandem and Dev. Will be
testing from CVS with Wednesday night's build of LDAS.
Worked on sub-second resolution checks on Mtime: dded code to support and check
sub second values. Fixed up errors in "general/TimeSpec.hh": (added
header file isolation preprocessor commands, removed redefinition of timespec).
Tested on Tandem 5 and Dev. Will test on Dev from CVS build today.
Closed out PR concerning the documentation of the lam homepage (PR2644).
Closed out PR to create new graph to display active threads in APIs (PR2457).
Closed out PR to fixed closing windows error with time graph and histograms and
addeda dialog to select APIs for histograms if more than 3 are requested in
time graph (PR2327).
Closed out PR to change rebuild frame disk cache function from emergency socket
command to submitting an ldas command and displaying results of the user
command (PR2380).
Closed out PR to fixed cache view printing of narrow bar frame data
(PR1729).
Closed out PR to fix error when requesting GPS time conversion error with GPS
time 1 (PR2621).
Closed out three PRs in the CGI script used to apply for and to administer LDAS
accounts via the web server (PRs 2254, 2519, 2231).
Repeated the system and integration level tests on version 1.1.86. There were
no surprises and the results have been updated on the web and in cvs.
Finalized the report to the LSC Computer Committee recommending a plan for
increasing the length of RDS frames by the start of the next Science Run. The
report in in the DCC (L040081-00).
Hardware Systems (Anderson)
Caltech
-------
(Dan Kozak)
* Continued the transfer of files from HPSS to SAM-QFS (and their subsequent
deletion from HPSS). I'm now transferring the the last category of files:
non-Science run second trends for both LHO and LLO (about 1,000,000 files). I
started with LLO: of the 458620 LLO files, we've transferred 211759 as of this
writing (46%). We have not been keeping up the 2000 files/hour rate that
we had when we started (there was a 4 day period when we were down to
<200/hour) and HPSS has increased the rate of failure from crashing once
every 3/4 days to crashing multiple times a day, so we've lost lots of time
there. I've got Mike Gleicher working on getting the SFS backups in HPSS
to work correctly so they don't fill up the /opt/sfs_arch disk (the most
readily understandable problem of the couple that I'm seeing).
* Restarted various parts of HPSS that crashed so that the above could continue
(this took up a _lot_ of my time this week).
* Replaced 2 failed drives in Booth 3510s with disks from our spare 3510. Put
drives received from Sun into spare 3510. Gave control of spare 3510 to
Ben Johnson for testing purposes.
* Participated in a talk with individuals from CACR about our experiences with
SAM-QFS.
(Al Wilson)
* Continuing working on the dead ASA units.
* Node8 Failed the burn in test. Contacted ASA tech and followed their
suggestion. Will burn in over the weekend.
* At the same time I asked about our replacement MB for PCDEV1 they said it
will ship by Friday and apologized for the delay.
(Stuart Anderson)
* Continuing to test SAM-QFS 4.1.2 and report minor issues to Sun.
* Finished the initial development of an enhanced SAM-QFS prestager script that
monitors users data requests and attempts to pre-load data from tape to disk
before users make actual file requests.
* Updated a few packages in /ldcg
MIT
---
(Keith Bayer)
* Added Lindy to MIT cluster.
* Working on condor node adjustments.
* Moving data on pcraids.
* Investigated ram problems on SF880.
Livingston
----------
(Igor Yakushin)
* Fixed the problem with missing or corrupted S2 L3 frames on LLO's cluster.
* When trying to run waveburst at CIT, discovered some missing S2 L3
frames there.
* The old dataserver, that isused for backups, died. Diagnostic boot showed
problems with memory; however, replacing bad memory did not help. SUN is going
to replace the whole motherboard today.
* Started publishing to LDR 1% data test for L1 and L3 RDS.
* Reconfigured the spare t3-13 not to use hot spare drive since we currently
cannot test if a hot spare is good or not.
Hanford
-------
(Greg Mendell)
* Continued working on various issues involving RDS generation and planning for
S4. The main progress to report this week is that the LDAS createRDS jobs and
the driver script (after resetting the frameLength in the resource file) have
been shown to run without complaint on 32 s raw frames. This is a minor
test, since the content of the frames themselves were not checked, but it
verifies that 16 s is not hard-coded anywhere as the raw frame size.
(Ben Johnson)
* Installed Matlab 7 on ldas-pcdev1, connecting with the site's license file.
* Reconfigured t3-25 to _not_ have a hotspare.
* Created logical disks on test 3510, identical to LHO/LLO 3510
configuration. I'll experiment with the disk verification methods.
* The cluster is heavily loaded again. AC is working ok.
* Added Keith Thorne to the cluster, assisting him with setup for data
analysis bootcamp.
General Computing (Wallace)
MIT:
(Keith)
-Installed femlab for urop on straw.mit.edu
-Backed up failing laptop drive for user contacted dell / replaced drive etc...
-Investigating dell laptop overheating troubles
-Investigating ibm mini-pci wireless card failure
-Helped install/configure laptop for new postdoc
Livingston:
(Shannon)
-Current bandwidth usage can be seen at
http://teche.ligo-la.caltech.edu/mrtg/LLO-Router/130.39.245.1_1.html
-Archived bandwidth usage can be seen at
http://teche.ligo-la.caltech.edu/mrtg/archive/
under the dated folder for the week of interest.
-Compiled and installed several more applications in the /apps folder. Still
looking at removing the old /opt from the remaining machines that are mounting
it. Need to relocate the global cshrc and create simlinks so that user's
environment stays the same. -Installed the old version of GCC (2.95.3) outside
of the normal path for when applications will not compile with the new GCC
(3.4.1).
-Working out some emacs problems. It segfaults every time it is run when it is
compiled with GCC 3.4.1.
-Evaluating Sun Management Center to see if it would be useful here. Just spoke
with the local Sun rep since he was on site. He claims that it is
pricey... Have not gotten it to install successfully yet anyway.
-Pulling some CAT6 cable in the Computer Users Room.
-Upgraded the Solsoft software which fixes several past bugs in the software.
It is much improved over the 5.3.1 version. The software also seems to
run better on the Suns than it used to. I am waiting to see how much they
are going to hit us for on our maint. contract since it expires this year.
Hanford:
(Christine)
- Network usage can be seen at
http://apex.ligo-wa.caltech.edu/~christin/mrtg/
198.129.208.1_198.129.78.122.html
- I've been working with people at PSU in preparation for the upcoming summer
camp activities.
- The antenna extenders for the wireless access points have arrived, I've been
installing and testing them to find the best location for them to be mounted.
- Misc. user support.
CIT:
(Mike)
-Worked on updating all NTSRV's with OS patches and anti-virus updates and
looking into the logs for intrusions/OS errors.
-Burned last months NTSRV ghost images to DVD and ran end of month ghost backups.
-Updated software and security patches on 4 laptops before they were shipped to
Hanford, for the LSC conference.
-LLPDMWorks Server: I installed IIS and WebPortal and setup all users accounts.
This was a very time consuming project. -Other Misc. onsite/phone user support.
(Lisa)
- Began writing a script to automate cleanup of the dhcp tab.
- Did monthly backups.
- After patching the apache/php server on vega, the php application does not
run correctly. Spent quite a lot of time trying to resolve that.
Vega runs a tracking tool that is critical to the internal operations of
GC. This problem is still unresolved.
Mail Stats 7/22 - 7/28/04
Messages Accepted:
Spam Rejected:
Viruses Rejected:
False Positives:
Total Mail Thru:
% spam:
(Veronica)
Most time was spent on catching up with tasks accumulated during my time away
last week:
- LSC website: Updates to the August meeting website. Posted the talks given at
the GR17, as well as papers for the LSC review. howto.
- LIGO website: Posted updates to other webpages at the LIGO website. Helped
Mike with using the digital cam and wrote up a cheatsheet for cam use to be
taken to the LHO. Continued helping Sharon get up to speed on web updates. We
worked on updating her and Barry's webpages, as well as on general html howto.
(Larry)
-Spent time going through purchases. A number of things on order have not
arrived when they should so those items are being tracked down.
-Reviewing the user accounts and making a list of old accounts to be removed
after approval by LIGO management.
-Working on a new laptop build. Getting the specialty applications installed is
definitely taking more time than planned.
-Working with Christine, Ed, Mike and others on the logistics for the LSC mtg.
-Began installing the Spyware detection s/w ona couple of machines. So far
there have not been any clean machines.
-Worked a number of equipment issue including: new installations, moving
equipment for different people, some hardware debugging, turning old equipment
over to Ed C. to be taken care of.
-Assisted a number of users with different applications. Continuing to assist
Sharon in her transition, this will be an ongoing group effort.
From: "Thomas Frey"
<tfrey@ligo.caltech.edu>
Progress Period from 07.30 to
08.05
OUT OF THE OFFICE ON
VACATION JULY 30th
·
·
See http://www.ligo.caltech.edu/~tfrey/index.html for a
complete listing of all project related cost and schedule data.
Accomplishments:
·
·
Sub-system
PLANNING activities
§
§
Executed meetings with Carol and Subsystem Teams Tuesday through
Thursday.
§
§
Continued work on preparing web space for posting Adv. LIGO reports.
("The Whole Enchilada")
§
§
Continued work on updating the FY05 non-labor cost breakdown, based upon
progress data through June 30. (www.ligo.caltech.edu/~tfrey/rd2005/)
§
§
Resolved technical issue related to the export of cost loading data from
Primavera. Our database is so large that CSV exports, with 14 years worth
of data by quarter, are hitting a critical mass. To alleviate this
challenge, we have to divide all values by 100 with out decimals. This
will result in an accuracy variance of ~1000 USD.
§
§
Started input of changes / preparing RFIs / responding to RFIs, as a
result of the meetings with Carol and the Subsystem Teams.
·
·
ROSTER DATABASE:
§
§
Finished input of additional authorship list corrections and changes
provided by Peter Saulson.
§
§
Finished work on "electronic proofing" of the authorship
data. This is a result of our meeting with Stan and Phil last week.
§
§
www.ligo.caltech.edu/~tfrey/authorship/compare.htm
§
§
Continued to work with Irena to provide Barry with information regarding
FTEs and Institutions.
·
·
COST BOOK DATABASE:
§
§
Executed meeting with Carol, Phil, and
§
§
Started preparing list / statistics of accounts and activities as
requested by Carol, for
From: Larry Jones <ljones@ligo.caltech.edu>
Advanced LIGO Seismic Structure
SEI Structure:
Configuration design is complete and all design requirements are met with
the exception of a few minor issues which were compromised. Detailed drawings
are in process; 68% of the approx. 120 parts drawings required are ready for
supervisory review. Release for fabrication of piece parts will be in three
groups, by structure stage. These are scheduled for September 6, 20 and 27. The
pods drawings release is scheduled for August 16, the spring test tool drawing
release for August 23, and the springs and flexures drawings release for August
30.
ASI has provided a verbal report of the contract's Cost To Complete (CTC), with
breakdown details to follow. This is a significant increase over the increase
reported on
Delivery at LASTI of the BSC structure prototype is now scheduled for
Actuators:
Vacuum bake and RGA scanning were completed on the large and small prototype
actuators in the new Vacuum Bake Oven C at the 40M lab. Flag hydrocarbon
outgassing levels measured were among the lowest we've measured, at 1.8E-11
tl/s for the large actuator and 9.5E-12 tl/s for the small actuator. These
readings include the background outgassing from the chamber, RGA head and
related plumbing as well. The large actuator was shipped to Kyle Ryan at LHO
for thermal vacuum testing.
PSI is proceeding with fabricating the 6 large and 6 small actuators for the
BSC prototype structure.
Branford Wire, a possible contractor for precleaning the mag wire, is willing
to consider changing cleaning solutions for effective removal of the paraffin
dry lube used by MWS Wire for payoff ease. The supplier of the dry lube is
being contacted for his recommendations for cleaning; Branford's source for
cleaning solutions is also being contacted.
Displacement Sensors:
Nothing new
Seismometers:
Nothing new
Galling/Dusting Test:
Nothing new.
Other:
Planning is proceeding on the design and sourcing of a larger vacuum bake
oven for the quad structure.
From: Janeen Romie <romie_j@ligo.caltech.edu>
AdLIGO Suspensions
Distributed the Advanced LIGO Quad Installation and Alignment
Fixtures Product Design Specification, T040151 to Ken Mason, Ken Mailand, Mike
Smith and Doug Cook, who will be doing the designs.
Working with procurement on the blade RFQs.
Participated in telecons with the design team on the catcher/upper structure
interface.
Working with
Working on the July Primavera updates. I hope to have them to Thomas on Monday,
Aug 9th.
Worked with Carol and Thomas this morning on the suspensions budget and
schedule.
From: Helena Armandula <ahelena@ligo.caltech.edu>
Advanced LIGO Coating Development
Received sapphire and fused silica substrates coated with doped Ta2O5. (Formula
#4)
We have 3 different types of doped Ta2O5 coatings, Formula 1, 3 and 4 applied
to both, silica and sapphire substrates.
The coating of the last substrates with the doped Ti2O5, Formula 4, took place
in the large coater at LMA, qualifying this large chamber for the manufacture
of doped coatings, if needed, for Advanced LIGO mirrors.
From: Bill Kells
<kells@ligo.caltech.edu>
1. Working on issues concerning the Sph/FS down select. In particular we are
trying to coordinate measurements on The sapphire absorption inhomogeneity with
Stanford and our OTF. I (and Liyuan) are gearing up (after the absorption
issues are settled) to study HR surface point scatter. We have a design and
setup ready.
2. I have been studying Radiation pressure noise issues for advanced ifos, in some detail from a fresh perspective.
This is what I hope (very maybe) to present at the LSC. At least it will be a
good opportunity to sound out the ideas.
From: Michael Smith
<smith@ligo.caltech.edu>
OPTICAL LEVER
FOCUS LENS
I am still in the process of learning how to use Zemax. I created an optical
model for the AdLIGO optical lever receiver lens, using the Zemax optical
design program. I developed a simple assembly and alignment procedure.
The model agrees with the ABCD matrix model I had developed previously. The
housing was modified to avoid vignetting of the extreme ray bundles, as
discovered by the optical ray trace layout. The program was used to optimize
the focus, and spot diagrams enabled a simple determination of the focusing and
decentering tolerance of the first two lens elements; the focus of the
objective lens and the lateral decentering of the second lens element are
excessively critical. A re-design is in process to reduce the tolerances.
From: Peter King <pking@ligo.caltech.edu>
Shally Saraf and Carl ? came down from
Stanford to work with a different pre-modecleaner. They had some
difficulty getting a good output from the pre-modecleaner lent to them some
weeks earlier. The pre-modecleaner input and output mirrors appear to
have been damaged, as there is a fair amount of scattered light around the
incident spot. It is not obvious how the damage came about, either during
shipping or some other means. Meanwhile, they have borrowed another unit.
From: Riccardo DeSalvo
<desalvo@ligo.caltech.edu>
1. Francesco
Working on mirror suspension flex joint development.
I had a look to Xavier's last year files and now the material and the element
for the mesh of the ANSYS model applied top our flex joint design are correct,
however the vibrational modes coming out from the simulations aren't realistic.
I hope I'll understand what I am doing wrong as soon as I find Calum Torrie.
I am also analytically deriving the frequency of the modes we are interested
in, I made a simplified model of the joint as a double pendulum with spring
attached to simulate the stiffness of the material. I found the dependence of
the frequency on the stiffness and the applied load, a quantitative analysis
should be possible thanks to N.Lobontiu's paper about flexing beams with
fillets, I am still working on it.
Simone
Baked the maraging training flex joints in a oven at 435 C for 100 hrs under
Argon atmosphere, and preparing the test setup, including a sturdy support
stand and payload, vacuum compatible for later in vacuum measurements.
The hooking setup is ready. Starting Maraging training tests asap.
Yanyi
This week I have finished the PCB layout for the control circuit, which
consists of the LVDT driver, the gain stage, the voltage saturation alarm and
the power supplies, and created a list of all the components and chose them
from the digikey catalog.
Maddalena
I'm just arrived from
Barbara:
In this week I started assembling the vacuum cabling for the stepping
motor and for the PZT. We also decided to modify the circuit for the PZT.
We realized that the sperical optics is thinner than the mex-hat one.
We are machining modifications to accept both optics thicknesses. In the next
days, as soon as the mechanical modifications will be ready we will be able to
mount the spherical optics and starting with the optical part of this project.
Juri
working on integration of thermal noise with different beam profiles
Jessica and Randal
Since we have arrived in
Justin
With Alessandro in
For additional information about this report, contact whitcomb_s@ligo.caltech.edu