Posts about sciblog (old posts, page 1)

2009-11-22 some unix tips

2010-11-11 13:40:37
  • find / -nouser te donneras tous les fichiers dont le nom du propriétaire est inexistant de la table /etc/passwd
  • batch converting
    for draw in `find /path/to/wiki/data -name \*.draw`; do
        file=`dirname $draw`/`basename $draw .draw`
        if [ -e "${file}.gif" ]; then
            echo "Converting ${file}.gif to ${file}.png"
            convert "${file}.gif" "${file}.png"
        fi
    done
  • You may always pipe the output of commands through grep to find specific words, but it can also be used to find files that contain a text string:
    grep -lir "some text" *

    The -l switch outputs only the names of files in which the text occurs (instead of each line containing the text), the -i switch ignores the case, and the -r descends into subdirectories.

  • compression
    in : tar cvf nom.tar dir/*
    out : tar xvf nom
    liste : tar tvf nom
  • du -sk : occupation des fichiers sur un disque
  • a2ps archives/unix.txt -Prsi2 imprime joliement...
  • dos2unix fic1 fic2 transforme fic1 en fichier texte unix nomme fic2
  • Pour connaitre le taux d'occupation de l'UC d'une machine, utiliser la commande sar. Exemple : sar -u 1 5 donne la consommation UC pendant 1s et 5 fois de suite

2009-11-15 MoinMoin: howto install a new theme

2010-11-11 13:35:33
  • locally
    scp Downloads/moniker18_2.1.1.zip  perrinet@195.221.164.4:/var/www/moin/perrinet/data/plugin/theme/tmp
  • on the server
    cd /var/www/moin/perrinet/data/plugin/theme/
    export USER=www-data
    export GROUP=www-data
    export INSTANCE=/usr/share/moin/htdocs/moniker
    unzip moniker18_2.1.1.zip
    cd moniker18_2.1.1
    cat read\ me\ on\ installing.txt
    cp -r moniker /usr/share/moin/htdocs/
    cp moniker18.py ../../
    chgrp -R $GROUP $INSTANCE
    chgrp -R $GROUP ../../moniker18.py
    vim ../../../../../../perrinet.py # set moniker18 as default

2009-08-29 Tips on Filesystems, security and al on mac os x

  • mount with AFP sharepoints from the command line (thanks to this hint) :
    # mount_afp [-i] [-o options] afp_url node
    mkdir /Volumes/truc
    mount_afp afp://user:PASSWORD@server/truc /Volumes/truc
    umount /Volumes/truc
    rmdir /Volumes/truc

finder nags when changing a file's extension

  • read the default value (should be 0)
    defaults read com.apple.finder FXEnableExtensionChangeWarning
  • change it:
    defaults write com.apple.finder FXEnableExtensionChangeWarning False

Make spell check show only desired languages

HFS+

  • Mac OS X Filesystems
  • One feature of HFS volumes is that fle are referred to as links: for instance you can read a PDF file while changing the name at the same time. No problem!
  • Safely remove '._' files created by HFS(+)
    find . -name '._*' -print0 | xargs -0 rm

Spotlight

  • to remove the remaining index files on the volume using the command:  sudo mdutil -E /Volumes/volume_name  utilitaires CLI :
    sudo mdutil --help
    mdutil: unrecognized option `--help'
    usage: mdutil -pE volume ...
            mdutil can be used to manage the metadata stores used by Spotlight.
            -p              publish metadata for the provided volumes.
            -i (on|off)     set indexing status for the provided volumes.
            -E              erase the master copy of the metadata stores for the provided volumes.
            -s              print indexing status for the provided volumes.

System & Security

nmap

  • to scan for open ports on a remote machine :
    sudo nmap -T3 -vv -sS -p 1-65535 -P0 google.com
    Starting nmap 3.81 ( http://www.insecure.org/nmap/ ) at 2006-04-05 16:32 CEST
    Initiating SYN Stealth Scan against google.com [65535 ports] at 16:32
    Discovered open port 80/tcp on google.com
    SYN Stealth Scan Timing: About 2.18% done; ETC: 16:55 (0:22:26 remaining)

Dash board

  • Don't use Dashboard? No particular reason to leave it running, consuming memory. Following http://www.macosxhints.com/article.php?story=20050723123302403, you can turn Dashboard off by doing:
    defaults write com.apple.dashboard mcx-disabled -boolean YES
    killall Dock
  • Unsurprisingly, you change YES to NO to re-enable Dashboard:
    defaults write com.apple.dashboard mcx-disabled -boolean NO
    killall Dock

Unix - X11

Change login window on (Snow) Leopard

  • disrupted by the look of the plasma flames? think it looks like a cheap star trek sundae? check http://paulstamatiou.com/2007/10/31/how-to-change-leopards-login-wallpaper :
    cd /System/Library/CoreServices
    sudo rm DefaultDesktop.jpg
    #sudo mv DefaultDesktop.jpg DefaultDesktop_old.jpg # if you want to keep the Aurora stuff (it stills around, do a 'locate Aurora'
    sudo ln -s /Library/Desktop\ Pictures/Nature/Stones.jpg DefaultDesktop.jpg

2008-12-05 duplicate files

2011-07-06 14:00:04

You may find yourself overwhelmed by files and in the need to keep the filesystem organized. If deleting is the best option, you may consider these 2 options:

Dupinator

Dupinator, tries to find duplicates and to report them in order to clean-up the organization of your files.

changelog

It works by:

  • launched via command line by passing a set of directories to be scanned
  • traverses all directories and groups all files by size
  • scans all sets of files of one size and checksums (md5) the first 1024 bytes
  • for all files that have the same checksum for the first 1024 bytes, checksums the whole file and collects together all real duplicates
  • deletes all duplicates of any one file, leaving the first encountered file as the one remaining copy

2008-11-20 Impact Factor

2010-11-16 16:57:41

Most researchers nowadays are judged based on their publication list and ---as a shortcut--- quantifically rated by their cumulative Impact Factor. How efficient is this method?

  • This paper studies the assumptions underlying the journals' impact factor and the open access initiative:
    • M. Taylor, P. Perakakis, Varvara Trachana. The siege of science , URL . Ethics in Science and Environmental Politics, 8(1):17--40, 2008 .

2008-08-29 keyboard

Touches spéciales (unix) sur clavier Mac francais

Merci a http://busy.lab.free.fr/mac/ !

backslash \ shift + option + /
pipe, or | shift + option + L
tidle ~ option + N
simple quote ' 4 key, that's to say a regular ' on any keyboard I guess.
opening brace { option + (
closing brace } option + )
opening square bracket [ shift + option + (
closing square bracket ] shift + option + )

Apple also got the habit to give it's own special names to some keys:

  • ctrl is called, well, control. So far, so good.
  • the alt key (with a funy sign under it) is called option.
  • the key with a funny sharp and the Apple logo is called command (for we are supposed to be so creative Apple users I think), or sometimes more logically apple.

keyboard shortcuts

  • in Finder (and most other applications)
    • ⌃⌘Z to zoom the window
    • ⌘ + ` to switch between your windows
    • ⇧⌘[ and ⇧⌘] to switch tabs
    • ⌃⌘D to show the dictionary
  • in Firefox:
    • Fn + Alt + Left arrow gets you to the home page.
    • Command + L : link bar
    • Command + K: search box
    • Ctrl + Tab: Next Tab
  • see Mac OS X keyboard shortcuts

2007-10-22 Deliverable M9-3: Workshop for definition of a detailed version of the V1 hypercolumn model

The INCM was holding a workshop on the V1 hypercolumn model on the 22nd and 23rd of October 2007. The purpose of the workshop was to promote the coordination of modeling efforts within the consortium and in particular to organize collaboration in WP9T2 and associated work-packages. It is labeled as deliverable M9-3: "Workshop for definition of a detailed version of the V1 hyper-column model" for WP9-2 (workpackage 9 task 2), but is also linked by its subject to WP5. In contrast to the previous meeting, we started with brief presentations of the results from each group to expose the different efficient aspects of each model. This allowed us in the second half of the workshop to converge to some main issues and prioritize the neural features that are the most important for the efficiency of V1. As for the format, we proposed that PhD students and post-docs should have the opportunity to present this work to give them experience and reduce the burden on busy chiefs. We felt it was important to involve as many in this workshop as possible as we approach a crucial stage in integrating the different models.

Program

Afternoon of Monday, 22 October 07: Framework

12:30-14:00

Registration and Lunch

14:00-18:00 (First session)

Introduction: coordination of modeling efforts for WP9T2

14:00-14:10

Guillaume Masson (INCM): Welcome

14:10-14:30

  • Laurent presented some examples of using the benchmark structure proposed in NeuroTools and the new SpikeList object. The benchmark will be the privileged way of spreading the different benchmarks to all partners and is a practical way of describing a benchmark as a list of experiments, storing the result of the experiments, distribute and run the experiments on a cluster and finally for plotting figures. In a second part, Laurent presented the SpikeList object as an answer to the difficulty of finding a common format for describing lists of spikes. Some experimental example and an interface with PyNN was shown as a proof of concept. Jan suggested that the internal representation could be sparse. This should be easily done by using the format methods (See SpikeList).

14:40-15:10

  • Andrew and Jens presented a system's approach to modeling the visual system. Partners in FACETS need to exchange knowledge on their modeling progress and this approach allowed to construct a model from composite parts (retina, LGN, V1) from the “LEGO bricks” of the different groups. A practical example with INRIA/INCM retina and a INCM V1-layer 4 was shown. An LGN brick is to be developed. Brick for higher areas (MT) can also be developed.

15:20-16:00

Coffee

16:00-16:20

  • Andrew proposed a format specification for benchmarks : 1) the input is provided as zipped PNG files, storing is done in the FACETS knowledge base (every stimulus has a unique URL), 2) A specification scheme for writing benchmarks and a whole experiment was presented with a proposed format using a XML interface.

16:30-16:50

  • Adrien presented his physiologically realistic retina, emphasizing on the contrast gain control. His model uses the INRIA simulator but with an open architecture and specification compatible with the FACETS specification. The retina uses a feed-back on the bipolar cells to achieve realistic contrast gain control which was compared with classical experiments (Shapley and Victor, 79, Enroth and Cugell XX). A possible extent is to study the importance of the spike profile at image onset which reveals first the luminance image and then the edges (Enroth Cugell 83 Bernadete-Kaplan 99). He also approached a mean field approach valid for gratings.

17:00-17:20

  • Klaus presented the retina used at the TUG. It is based on a spatio-temporal decorrelation model from Dong and Attick (1995) for the linear part and on a gamma renewal process (Gazères, 1998) for the spiking part. It is available to FACETS partners in the SVN1 and will converted to the format specified in the VisualSytem class (See https://www.kip.uni-heidelberg.de/repos/FACETSCOMMON/facetsmodel/LGN/trunk/graz ).

17:30-18:00

General Discussion

  • conclusions on benchmarking progress and decisions about their future development and use.
  • We concluded on benchmarking progress and decided about their future development and use. In particular, the unification proposed by the Benchmark class and the SpikeList object from Laurent and the specification proposed by Andrew seemed to fit to the needs of the partners and we agreed to use the proposed schemes.
  • We agreed on further definition of the visual benchmarks WP9T2-VisionBenchmarks (while respecting the standards from FACETS_Benchmarks). This will be implemented in a coming deliverable.
  • The question if we can reach the goal a single, common framework for multiple V1 models in FACETS was left open due to the wide variety of approaches in the consortium. However, the discussion suggested some collaborations of groups on some specific scientific questions.

Morning of Tuesday, 23rd of october 07

09:00-13:00 (Second session)

Progress in Modeling V1

  • 09:00-09:20*
  • Mike described current progress towards a neural model of motion perception in V1/MT, a collaborative project with CNRS/INCM. The model is based on the recurrent neural circuitry between a V1 hypercolumn and area MT, and uses a Kalman-Bucy approach to estimate the velocity of the moving object by integration of local motion information. The model will be suitable for testing using the FACETS visual motion benchmark stimuli, in particular the CNRS/INCM data on motion integration for smooth eye pursuit.*

09:30-09:50

  • The KTH model was presented by Jan, with the latest changes and additions and some preliminary results. HH neuronal models, hierarchical columnar structure. The horizontal connectivity is specified by the LISSOM model while the vertical connectivity is inspired by biology.*

10:00-10:20

  • Klaus presented the TUG model with the latest changes and some preliminary results. He illustrated the use of Izhikievitch neurons, the data from Alex Thomson (and not from Binzegger) by showing some results of the columnar model in particular by studying the influence of NMDA.*

10:30-10:50

  • Jens presented progress of the INCM/ALUF cooperation emphasizing on progress in the specification of the thalamo-cortical projections. He presented results of applying the different benchmarks proposed to reveal the functional properties of the model. It emerged that the inhibition scheme used allowed a normalization of the input at the network level. This was put into evidence using the spike-triggered conductance profile which were consistent with some recent data from Rudolph et al. (2007).*

11:00-11:30

Coffee

11:30-13:00

  • Round-table discussion: questions, problems, and issues of modeling V1. Moderator: G. Masson
  • Retina model: shouldn't we use a more standardized input? How should we set up background noise? Biologists need to specify what they intend by background (on-going) activity to be simulated in large scale neural networks.
  • Back to back cooperation between experimentalists and modelers for defining benchmarks in connections with dissemination of experimental data.
  • Link with more high-level task, as the one presented by Mike and in WP9T3 or the work done in collaboration INCM-INRIA. Since they use the same benchmarks (such as motion integration), these approaches can better defined the computational rules to be implemented in large scale neural networks. One good example in the role of asymmetric diffusion of information/activity in the network, thanks to feedback from higher areas.

13:00-14:00

Lunch

Afternoon of Tuesday, 23rd of october 07

14:00-16:30 (Third session)

Scientific questions coming from the Biology of V1

14:00-14:20

  • Julian presented a review of different results on the physiology and anatomy of the cat retino-thalamo-cortical projections. In fact, though the question of the architecture of these projections was questioned during the workshop, little is known with general agreement. Correlating different works from the literature, a detailed quantification was reviewed suggesting a disagreement between anatomy and physiology. Presenting the work of Ringach (2004), he concluded by showing own simulations suggesting that the properties and diversity of the receptive field of cat's area 17 simple cells may be captured by a wiring scheme based on the specific quantization of the parameters of the retino-thalamo-cortical pathway.*

14:30-14:50

  • Cyril reviewed different models of the emergence of orientation and direction selectivity before emphasizing on the results of different groups on the role of conductance profile in this function. This revealed a diversity of behavior between push-pull model where inhibitory and excitatory profiles are in overlap and other configuration This was put in light with results obtained at the UNIC.*

15:00-15:20

  • Alex presented preliminary results of center-surround interactions using VSD optical imaging in the primate V1 cortex. In the retinotopic position of the center the response to the center appears with decreasing latency for increasing contrast. The response of the 80% contrast surround reaches the center at a latency equivalent to approx. 15% contrast, leaving open the question of the interaction of these two information streams. Preliminary results show suppression for high contrasts but facilitation for low center contrast. Further analysis (of latency, propagation) suggests the functional role of horizontal propagation in this configuration.*

15:30-16:00

Coffee

16:00-16:30

General Discussion Moderator: Yves Fregnac (UNIC)

16:30-17:30

Outcome: Planning of Implementation plans / priorities for WP9T2.

  • Several actions need to be taken. 1) We keep the idea of one annual meeting on V1 modeling. The meeting shall be held in June instead of October, to prepare for Annual reports and implementation plan. We will post a call for the organization of the 2008 meeting. We should also try to bring more biologists in these meetings. 2) We will organize a phone/video conference once every 3 months to exchange information and compare outputs for each benchmark steps. 3) We shall provide a timeline for delivering benchmark tools and objectives, as well as deadline for collecting results for testing models. Such a timeline will be added to the D9-2 in which we will describe the different benchmarks. 4) We will set a discussion list on the FACETS Wiki website to propose new questions from modelers to biologist and vice-versa. The idea is to put information for which there is a general agreement rather than having an on-going forum. Answers shall be concise, with reference to published work and or available data. 5) We shall promote active collaboration between sites with the objectives of common publication of one specific aspects of visual tasks to be develop in FACETS (on-going activity, local cortical point spread function ….)

Questions

Below are some questions modelers (please add to this list) would like the biologists at the meeting to answer during the round-table discussion (11:30-13:30 on 23 October):

  • What is the purpose of the Y-type pathway input to layer 4 of cat area 17?
  • Is the tuning of cortical neurons dynamic or not (e.g. for orientation)?
  • Can simple or complex cortical cells be directionally selective but untuned for orientation?
  • Are inhibitory neurons in cortex generally tuned for orientation or not?
  • Do inhibitory fast-spiking (FS) neurons have a higher spiking threshold than excitatory regular-spiking (RS) neurons?
  • Is modelling corticothalamic feedback essential for models of V1?
  • What common mistakes do modellers make that annoy you and fellow biologists the most?
  • How can modellers best help the biologists?

For the sake of fairness, we would also like some questions from the biologists for modellers to answer during the same discussion session.

Organization

When and where?

The date for the workshop is Monday, 22 and Tuesday, 23 October 07. It will take place at the INCM in Marseille as last meeting (see Marseille_November2006 ).

Who is attending

Please https://facets.kip.uni-heidelberg.de/internal/jss/AttendMeeting?meetingID=28 register your attendance.

  • monday lunch 16 people (everybody except Anders Lansner)
  • monday dinner 15 people (everybody except Anders Lansner and Andrew Davison)
  • tuesday lunch 16 people (everybody except Andrew Davison)

facilities

we will have

  • a beamer
  • no internet connection (ask if you need one)
  • lunch and coffee breaks!
  • ( the video conferencing system was not needed anymore)

more info

2006-11-30 Post-doctoral Position in Computational Neuroscience: "Functional, Large-scale Models of Visual Motion Perception"

alert The position has been filled

We are currently inviting applications for a postdoctoral position in computational neuroscience to study functional, large-scale models of visual motion perception. The post is for up to 3 years in the DyVA team at the INCM (CNRS) in Marseille, France and will be funded within the european FACETS consortium.

The project will involve developing, designing and implementing large-scale neural networks of the primary visual areas. The goal is to bridge theoretical principles of computations, physiological data (e.g. optical imaging) and behavioral data (e.g. eye movements) to bring new understanding on the neural computations underlying the perception of motion. We are particularly interested in the dynamics of motion integration, focusing on responses to complex motion stimuli in areas V1 and V5/MT. Our theoretical approaches span from statistical inference to adaptive distributed representations to provide insights into these parallel, dynamical processes. We use state-of-the-art simulation software to validate these results thanks to our computer facilities, with the goal of transferring this technology to aVLSI chips developped within the consortium.

The DyVA team is a young CNRS pluridisciplinary research team (Head: Dr. Guillaume Masson) integrating research on eye movements and visual perception at both behavioral (psychophysics, motor control) and physiological (optical imaging, electrophysiology) levels. The computational project will be conducted by Dr. Laurent Perrinet in close collaboration with other FACETS teams (Pr. O. Faugeras, INRIA,Sophia-Antipolis, Pr. Ad Aertsen, Freiburg, Pr. W. Maass, Graz, Pr. Y. Fregnac, CNRS, Paris). The work conducted in Marseille will involve a PhD Student and a Computer Engineer together with the post-doctoral fellow. Close interactions within the team with the biological tasks conducted (optical imaging in awake monkey) are promoted.

The postdoc will work as part of a European wide research team based on a new EU-funded Integrated Project entitled "FACETS: Fast Analog Computing with Transient States". The FACETS project is a major 11 million Euros, four-year research project funded by the European Union. The stated objective of FACETS is to explore and exploit the computational principles that constitute the basis of information processing in the brain. The project involves all facets of neuroscience from experimental neuroscience ("in vivo"), the construction of models and analytical descriptions for neural cells and networks ("in computo"), to the construction of very large scale neural circuits in VLSI technology ("in silico"). The FACETS consortium includes fifteen of the major laboratories in Europe in these areas. One of the goals of the FACETS consortium is to develop a large-scale model of the primary visual cortex and explore its computational capabilities for solving low-level motion tasks involving short- and long-range lateral interactions.

Expertise in computational modeling as well as mathematical and programming skills are required. A keen interest in neuroscience is necessary to ensure the tight coupling of the computational approach with experiments done in the lab. The position is available immediately, the appointment will be for a fixed term of up to three years, ending 31st August 2009.

Marseille is a major french town by the mediterranean sea with a lively atmosphere and cultural life. The DyVA team is located in a CNRS campus with excellent research facilities. Net salary is about 1800€/month, full social coverage included.

Interested candidates should send their application materials - cover letter, CV, statement of research interests and research experience (no .doc file, please) or informal enquiries to Dr. Laurent Perrinet ( mailto:Laurent.Perrinet@incm.cnrs-mrs.fr ).

This work was supported by European integrated project FP6-015879, "FACETS".

2006-11-20 V1 hypercolumn Coordination Meeting, 20th - 21st Nov 2006

Subject: Coordination meeting of the WP9T2 and WP5T3 tasks.

Main points and decisions

  • review / demonstration of coding strategies by different partners. This definition has impact on the "Meta Simulator",
  • review of existing V1 models inside and outside FACETS (see also !Review_of_V1_models),
  • definition of the common benchmark FACETS_Benchmarks to qualitatively compare different models and strategies in order to break the model complexity barrier, as we began to do for the simulator,
  • definition of the architecture of the FACETS model of an "hyper-column" by defining the priorities of different partners: effect of including different neuron types, including layers, specific lateral connectivity.
  • Partner 13 (Plymouth, Mike Denham) will provide an hypercolumn for D25.
  • Partner 6b will provide a retina before Dec. 15 for implementing benchmark zero.

Program

(actual program changed slightly)

Monday, 20 November 2006

Monday, 20 November 2006

morning, 09:00 - 13:00

Panel discussion : Reviewing essential knowledge from biology

* Christophe Lamy - Alex Thomson (ULON) : Diversity in the architecture
(layering, cell types and dynamics) of cortical columns
* Fred Chavane (INCM) : Essentials of the functional architecture
and cortical lateral connectivity
    * Cyril Monier (UNIC) : Models of V1 and orientation selectivity
(also see
    Review_of_V1_models)
    Round-table : what shall we keep for modeling/hardware ? what are the
priorities ? Moderator: A. Lansner

13:00

Lunch

afternoon

Visual benchmarking for FACETS V1 models

* Laurent Perrinet
(INCM) : Short description of visual function benchmarks for FACETS
* Jens Kremkow (INCM): Contrast gain control with layered neural
network
* Guillaume Masson (INCM): Using IO data for tuning models ?
* Adrien Wohrer (INRIA) : (CANCELED) Statistics of inputs for
benchmarking V1 models)Round-table : Implementation plans for
benchmarking. Moderator : O. Faugeras

Tuesday, 21 November 2006

Tuesday, 21 November 2006

morning, 09:00 - 13:00

Overview of V1 models from FACETS: diversity and common strengths

* Mike Denham (UP) : The english model
* Anders Lansner (KTH) : The swedish model
* Malte Rasch / Wolfgang Maass (TUG) : The austrian model
* Outcome : Definition of a FACETS common model of V1. Moderator : L.
Perrinet

13:00

Lunch

afternoon

Planning of Implementation plans * benchmarking the existing
models
    * implementing the FACETS model for D25
    * link between WP9 and WP5 on the track to the "detailed model" of
V1 computing.
    Outcome : to set deadlines and implementation plans. Moderator : G.
Masson

venue

who, when

  • complementary to the FACETS Web Site here are the timetable for participants (opportunities to share a taxi, there?):
  • Olivier Faugeras will attend on monday (no hotel)
  • Andrew Davison will arrive on Tuesday, leaves wednesday afternoon (hotel on tuesday)
  • Cyril Monier (hotel from sunday to wednesday)
  • Olivier Marre Pierre YGER will arrive on Sunday, leave wednesday morning (hotel from Sunday to Tuesday)
  • Anders Lansner + Martin Rehn arrive Sunday evening and leave Tuesday (flight around 5 pm). (hotel from sunday to monday)
  • Malte Rasch will arrive on Sunday 19.11 (hotel from sunday to monday)
  • Christophe Lamy will arrive on Sunday @ 11.30pm leave monday (hotel on sunday)
  • Michael Denham arives monday at 2pm (st charles), leaves tuesday at 18.09 (st charles) - (hotel on monday)