Space Warps Talk

Collaborative modelling - Spacewarps Lens candidate ASW0004dv8

  • Capella05 by Capella05 moderator

    With the latest version of SpaghettiLens it is now possible to collaborate with each other to make more refined models of the good lens candidates we are finding here on Spacewarps.

    So, in keeping with the ethos of the Zooniverse, we are going to work together to model a single candidate, and at the end of it, we are writing a Zooniverse Letter to chronicle our attempts (PI by @DrPhilMarshall ) 😄

    This is the first of many modelling mini-projects, so there may be bumps along the way, but at the end, you will be able to model a lens candidate worthy of an expert 😄

    Want to get involved? Read below…

    How do I get started?

    First off, watch the tutorial here:
    http://mite.physik.uzh.ch/tutorial/videos/all/tuto_player.html

    Then…

    Join a Google+ hangout with some of the Moderators and Scientists and have a modelling master class!

    PM myself @capella05 or @psaha to set up a date and time – we will then post details on the board as to the next hangout so everyone can join!

    Remember - Everyone is welcome 😃

    (You can skip this bit if you wish, but I always find them really useful)

    Finally…

    Just click on the link below to modify the initial model, or the children modeled by others !
    Even if you do not want to contribute to the letter, you can still refine the model!

    enter image description here

    enter image description here

    Model click left to refine the model, go to the bottom of the page and click 'Revise'

    The Analysis…

    As we are going to write a letter about our exploits, we will have regular hangouts to discuss the models - All you need is a mic and earphones - we will do the rest.

    At the end we will have one or two convincing models to publish, and everyone who has contributed will be credited 😃

    When you have modelled the candidate, please post the link below along with any comments.

    Posted

  • psaha by psaha scientist

    See also this thread.

    Posted

  • Budgieye by Budgieye moderator

    Spaghetti lens modeler http://mite.physik.uzh.ch/lmt.html

    Here are my thoughts on ASW0004dv8

    I see a long blue arc running from about 8 to 1, well centered around a yellow elliptical galaxy. The arc is buried within the disc of the galaxy. The arc at 9 is about 5" from the centre of the lensing galaxy. The arc at 12 is about 6" from the galaxy, so there is a slight asymmetry. Perhaps the light path is being affected by one of smaller ellipticals in the N . The colour of the arc is light blue. I would prefer to see a darker blue, but perhaps the light is being affected by being blended with the light of the galaxy.

    There is no definite counter-arc. There is a vague area of blue. The light from the counter-arc is hidden behind the galaxy nucleus.

    There are several other blue objects nearby which may or may not be lensed, I'll ignore them right now.

    The is a line of repair to the 1 of the galaxy, so I suspect the area has had a cosmic ray hit, maybe the original should be looked at.

    deleted my own attempts. These below are based on Capella05 model

    http://mite.physik.uzh.ch/data/004571 I moved the saddlepoint slight towards 12, to the vague blue patch. It seemed to make no difference.

    Posted

  • c_cld by c_cld

    ASW0004dv8
    Thanks Budgie for your observations.

    I used Aladin with the CFHTLS-WG on the coords 36.039756 -10.968812.

    My Aladin measures distances are

    south arc point: 3.3 arcsec PA=110

    north arc point:3.3 arcsec PA=346

    mid point :3.06 arcsec

    The arc is sligthly flattened but without assymetry for me. It shows an extension of 124 degrees. Its center is a little west offset from elliptical center.

    The nearest galaxy in north is at 3.41 arcsec at PA=324.7

    My revised model from last http://talk.spacewarps.org/#/boards/BSW0000004/discussions/DSW000063c is

    http://mite.physik.uzh.ch/data/4543

    It could be nice to have some measurement tool (compass, distance ) to check the synth image...
    The southwest brigther elliptical (in foreground?) seems without deflection action on the lensed source.
    From eyeball the redshift of the source is at max 0.6 and the lens around 0.2 - 0.25, but i've no photometric redshift estimation tool (by hyperZ ? ).

    To be followed up...

    Posted

  • Capella05 by Capella05 moderator in response to Budgieye's comment.

    Yes, we are going to use my initial model as the base for this collaborative model.

    I specifically made it very basic so we could all have a go fine tuning it!

    Posted

  • ElisabethB by ElisabethB moderator

    My first attempt

    Posted

  • Capella05 by Capella05 moderator

    Moved the minima to the NE of the galaxy (as per Budgies initial model) and added a mass point.

    The model is still suggesting quite a prominent counter image at 2:45

    Revised model

    Posted

  • ElisabethB by ElisabethB moderator

    And I got this one after tweaking Capella's model 4516 : model4570

    I made the max. smaller, moved the top min. a bit to the right and moved the lower saddle point in a bit.

    The arc looks okay, but agreed : not too happy about the bright counter image

    Posted

  • Budgieye by Budgieye moderator

    I will delete my first attempts, and concentrate on Capella05 initial model.

    Can you mention what you are doing to the model, and what change you obtained?

    Posted

  • Capella05 by Capella05 moderator

    This is the first attempt at collaborative modelling, so there may be a few hiccups!

    I am trying to think of a way of keeping things logical, so if we can try the following:

    • Only have a single model in a post.
    • List the parent of the model.
    • The ID of the model (Like what Els did)
    • As Budgie suggested, the change you have made to the model and what it achieved.

    For Example:

    • Parent - 4516
    • Revised Model - 4568
    • Action - Moved the minima to the NE of the galaxy (as per Budgies initial model) and added a mass point.
    • Result - The arc is more clearly defined from 7 to 11:30, not too happy about the predicted counter image at 2:45

    Posted

  • Budgieye by Budgieye moderator

    parent http://mite.physik.uzh.ch/data/004570

    revised http://mite.physik.uzh.ch/data/004585

    action I added the mass of the big yellow galaxy to the right.

    result It made no difference

    So I did it again, and made an absolutely huge circle!

    http://mite.physik.uzh.ch/data/004617

    result: and it made no difference.

    Posted

  • Budgieye by Budgieye moderator

    enter image description here

    Posted

  • c_cld by c_cld

    Parent 4568

    revised 4628

    action

    1. moved the minima inside the gal. quite opposite to the arc ;

    redshift lens 0.25 source: 0.65 (the gal. lensing looks like between 1237673699667280320 photZ 0.369 and 1237673699667279895 photZ 0.225 in the near north field )

    result: the arc get some blobs, which are not so clear in SW pic, even in CFHTLS-WG. Use "brightness and contrast" button to have a fainter arc image...

    The other galaxies are not relevant to the model which is very sensible to the position/distance of minima and center gal.

    Posted

  • Capella05 by Capella05 moderator

    • Parent - 4628 by C_cld
    • Revised - 4648
    • Action - tidied up the spaghetti, moved the left saddle slightly south, added an additional mass point.
    • Result - The left arc is more defined, not too sure the additional mass point has had any affect on the model.

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator

    I started this one from scratch. The annotated image looks a bit weird but I got rid of the counterimage

    model4681

    Posted

  • Capella05 by Capella05 moderator

    • Parent 4681 by ElisabethB
    • Revised 4686
    • Action - re-positioned the maxima, tidied up the spaghetti
    • Result - we no longer have the point at the galactic core, still indicating a counter image at 5. We need to make the arc more pronounced and extend it to at least 12 o'clock.

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator

    • Parent 4686 Capella05
    • 4711
    • Action :moved the max a little bit to the N.and moved the top min. a bit.
    • Result : arc from 7 to 12

    Posted

  • Capella05 by Capella05 moderator

    • Parent 4711 ElisabethB
    • Revised 4737
    • Action: added two mass points
    • Result: made no difference whatsoever!

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator in response to Capella05's comment.

    I know ! 😄 I have been trying that out as well !

    Posted

  • Capella05 by Capella05 moderator in response to ElisabethB's comment.

    When Rafi is back on Tuesday we really need to ask him about the mass points!

    Posted

  • c_cld by c_cld in response to Capella05's comment.

    restarted from your 4648

    http://mite.physik.uzh.ch/data/004755

    action: continue smoothing, using "add ruler" button for having nice circular contours

    keeping all neighbors point masses

    glass configuration:redshifts lens 0.25, source 0.65 ; pixel radius 11; nmodels 200

    result: better arc from 8 to 12, thiner with less blobs

    4741

    need: arc closer to the pic from PA 146 to PA 346; role of mass points inside/stradlle/outside contours lines?

    Posted

  • ElisabethB by ElisabethB moderator

    Started from scratch again and put the saddle point at the bottom of the arc and to my surprise, it got a reasonable result.
    4758

    enter image description here

    Posted

  • Budgieye by Budgieye moderator

    Oooh, ElisabethB, that looks strange

    parent http://mite.physik.uzh.ch/lmt.html?rid=4758

    revised http://mite.physik.uzh.ch/data/004809 ElisabethB

    action: moved saddlepoint close nucleus at about 5, moved arc markers further apart

    result: bright spot at 11, arc has bright spot at 1 but that might get masked by a satellite galaxy, counter-arc is masked by galaxy nucleus.

    enter image description here

    enter image description here

    I did it again using the lens distance of 0.2 - it looks the same

    Posted

  • Budgieye by Budgieye moderator

    parent http://mite.physik.uzh.ch/lmt.html?rid=4809

    revised http://mite.physik.uzh.ch/data/004836

    action: pushed red dot marking center of galaxy a bit left, also pushed saddlepoint a bit left

    result: counter-arc is hidden behind galaxy, arc is fainter

    enter image description here

    Posted

  • KhalilaRedBird by KhalilaRedBird

    I tweaked what I found but would really like to try again from scratch. [http://mite.physik.uzh.ch/data/004853] (http://%20http://mite.physik.uzh.ch/data/004853)

    I moved the maxima and minima slightly to where I saw the strongest blue of the arc, and I did see a counterarc very faintly and following the curve of the lensing galaxy. I moved the nodes on the magenta curves until, for the main arc, they paralleled the arc (away from the lensing galaxy, as best I could. This involved bringing them closer to the arc.

    The resulting image resembles what I see as the stronger (leftmost arc, with a single point (minimum) for the counterimage. I left it at that until I have a better feel for what I'm looking for/at. I seem to have been drawn to doing curve-fitting to parallel the arcs.

    A bit late, I find I should have noted what model I started with -- but I changed it so much it is probably irrelevant.

    I have given my name as Khalila RedBird, which is my religious name. My legal name is Sandra Lee Harris.

    Posted

  • KhalilaRedBird by KhalilaRedBird

    Images for my first try, 004853, as described in my previous post.

    http://mite.physik.uzh.ch/lmt.html?rid=4853

    http://mite.physik.uzh.ch/data/004853

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 4758

    Result : 4865

    Actions : moved max. a bit to the right. Moved the min. further apart. Moved the lower saddlepoint out a bit.

    Result : no more counterimage, but a small central arc.

    Posted

  • Budgieye by Budgieye moderator

    There would be a faint arc at 5 further away though. But faint.

    Posted

  • c_cld by c_cld in response to ElisabethB's comment.

    My comments on your model 4865 with two saddle points in the arc.

    For me, even if the output is an arc, it does not represent the shape of the arrival-time surface in the physical world. I would say it's a #simfail.

    You could get an arc with the simplest contour of a saddle in the upper end of arc and a minima in lower end but what will be the physical meaning of such a reconstruction, apart similarity of geometrical shape?

    I would say the same for input images where are drawn contour lines crossing each other: in the arrival time the self crossing are only saddle points. As Saha wrote we draw the skeleton of images, which I interpret as a non-contorsionist one.

    Could @psaha clarify this point as to be on the right track for collaborative modeling drawing contours on physical grounds?

    Posted

  • Capella05 by Capella05 moderator in response to C_cld's comment.

    When Prasenjit gets back from holiday we will be holding a hangout to discuss what criteria we will use to determine if a model is successful or not.

    At this point it is more about working out the processes we will use to go about collaborative modelling, and generating enough models that we can use for the data analysis hangout.

    Posted

  • Capella05 by Capella05 moderator in response to KhalilaRedBird's comment.

    Welcome KhalilaRedBird!

    I hope you will have fun modelling with us - let us know if you need anything!

    Posted

  • c_cld by c_cld

    From revision 4755

    newly revised 4916
    rotating a bit the saddle point in the core of the elliptical.

    The superposition of the input image and output is

    ASW0004dv8 - 4916

    The counter image of the arc in blue can't be seen trough the elliptical disc!

    Posted

  • rafael_kueng by rafael_kueng in response to Capella05's comment.

    Yep, it's confirmed. Unfortunately there is still a bug with the pointmasses. I'm working on it as soon as I got the other workers updated and running, so you have more cpu power available.

    Posted

  • rafael_kueng by rafael_kueng in response to KhalilaRedBird's comment.

    just as a note: if you start off modelling on the results / data page of an existing model, Sl will automatically keep track of the parent.

    by the way: welcome to modelling!

    Posted

  • rafael_kueng by rafael_kueng in response to C_cld's comment.

    nice picture! would such a function be useful to have in SL? maybe with a slider to adjust the transparency?

    I was thinking about it, but I'm not sure how to automatically align the images. If you could use it, let me know, then I can invest some more time in trying to get this thing implemented.

    For all of you: if you find a bug, or anything you don't like, or that could be better, either let me know by pm, or a post to the spaghettilens thread, or if you happen to have a github account you can create a ticket yourselves right here:
    https://github.com/RafiKueng/SpaghettiLens/issues?state=open

    Posted

  • c_cld by c_cld

    Thanks Rafael.

    No problem for me to put the input/synthetic image in GIMP2 without doing any scale/rotation change.

    I suggest you ask the group for delivering this superposition (subtraction ) with a slider in SL to show where blobs are well modeled or not.

    About alignment I think it's not a problem if there is a slight offset. Proposition: could you visualize this offset in putting a reticle in the synthetic image on the location of the red input point?

    Posted

  • KhalilaRedBird by KhalilaRedBird

    Trying again to get the hang of this.

    • Parent 4516
    • Revised model http://mite.physik.uzh.ch/data/005130
    • Action: Conformed spaghetti loops to parallel curves of the arc; brought loops closer to arc; moved saddle point away from lens to reflect faint counterarc
    • Result: the arc is more oblate than circular, as it is in the original image; there is suggestion of faint counterarc

    It was necessary to watch the tutorial again, this time with headphones, as the audio was much less intelligible the first time around.

    Posted

  • KhalilaRedBird by KhalilaRedBird in response to rafael_kueng's comment.

    Thank you! I'm playing this by ear and intuition at this point. I always was more a qualitative than quantitative scientist -- which didn't help much back at MIT in the '60s.

    Posted

  • rafael_kueng by rafael_kueng in response to KhalilaRedBird's comment.

    yeah, the tutorial.. It will become better.. It's on the todo list to...

    Posted

  • rafael_kueng by rafael_kueng in response to C_cld's comment.

    ok, fine. it worked because you created a nice input/spaghetti image 😃 But since one could arbitrarily zoom in/out to create the input image, this doesn't work out of the box... So since image scaling is arbitrary, it's fine if you do it while overlaying, as long as you keep the aspect ratio! You should never rotate thou.

    Not that you did it, just to mention it for others applying this technique.

    Posted

  • Capella05 by Capella05 moderator

    • Parent 5130 by KhalilaRedBird
    • Revised 5237
    • Action: moved the left saddle point slightly south, the right saddle point marginally towards the maxima (galaxy), tidied up the spaghetti.
    • The mass distribution map looks cleaner, less of an arc at 3.

    Posted

  • c_cld by c_cld

    After the Rafael'post on its go ahead with pixrad 12 models,

    here is model 004916

    revised with pixrad 12:

    ASW0004dv8 - 5310

    ASW0004dv8-5310

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 5310 by C_cld

    New : 5505

    Actions : moved the saddle point and the max. to the right. Moved both the min. closer to the saddle point.

    Result : fainter arc and counter image.

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 5505

    Result : 5515

    Actions : moved the min. a bit more to the saddle point.

    Result : another faint arc but a slightly stronger counterimage

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 5515

    Result : 5520

    Actions : moved the saddle point and the max. a bit

    Result : another faint arc and counter image

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 5520

    Result : 5523

    Actions : moved the saddle point and max. out a bit.

    Result : faint arc, but a more pronounced counter image

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 5523

    Result : 5527

    Actions : some tweaking

    Result : very interesting (I think) : a very faint arc, a faint counter image and a weird image that could be hidden in the glare of the central galaxy.

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 5523

    Result : 5529

    Actions : again some tweaking (sorry)

    Result : very interesting

    enter image description here

    Posted

  • Capella05 by Capella05 moderator

    • Parent : 4516
    • Revised: 5593
    • Action: Moved the maxima and saddle points.
    • Result: The arc on the left still needs to be fine tuned, but the mass distribution is the cleanest I have seen so far.

    enter image description here

    Posted

  • Capella05 by Capella05 moderator

    • Parent : 5593
    • Revised: 5594
    • Action: Moved the left saddle point slightly NE
    • Result: The arc on the left looks more refined, mass distribution is still looking good!

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 5594 by Capella05

    Revised : 5653

    Actions : saddlepoint and max. moved out a bit, 2 min. closer together and the saddlepoint at bit lower

    Result : fainter arc and counterimage

    enter image description here

    Posted

  • c_cld by c_cld

    • Parent: 5594
    • revised: 5811
    • Action: tweaking only redshifts to force deflector mass change (zs 0.36, zl 1.1 )
    • Result: little change of arc tickness?

    5811 5811

    Superposition/ subtraction of image output from image input is as good in 5594 as in 5811, adequate location of counter image in core galaxy.

    Conclusion: what else could be pulled out with this software to reinforce lens candidate? a bit more flatness of arc? or a model comparison test with a different software?

    Posted

  • drphilmarshall by drphilmarshall scientist, admin

    Guys! I'm back from various travels/travails, looks like you have some very promising models here.
    Notes:

    • I agree, there is a hint of a counterimage between the two bright lens galaxies, but it's very faint. It would be best if we could subtract off a model for the light from these two objects - but in the absence of that, a view of the g-band only image would be useful.

    • The superposition by @C_Cld is really helpful: I would prioritise the development of this sort of presentation if possible, @rafael_kueng - you were quickly into the regime where quite subtle distinctions between model-predicted arcs need to be made!

    • I agree we should include both mass components - probably with masses roughly in proportion to the apparent brightnesses of the lens galaxies (insofar as that is possible!)

    • Are we getting the detailed structure of the arc correct? A g-band only view would probably help here too.

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 5593

    New : 6479

    Actions : moved max. out a tiny bit, moved mins out a tiny bit

    Result : fainter arc with brighter bits and fainter counter image.

    enter image description here

    Posted

  • Capella05 by Capella05 moderator

    Hi All!

    I am busy collating all of our models / comments / actions / parents / children etc... into a single spreadsheet in anticipation of @psaha return.

    Since we will be having a hangout with the next fortnight* to learn how to analyse our own (and others!) models - perhaps we should hold off posting more models? The reason why I say this, is that we have been very prolific with our modelling attempts, and we need to start narrowing down the list 😃

    Or, after analysing our models, we may decide to do something totally different!

    Feel free to go mad on modelling any other candidate - I am sure @rafael_kueng appreciates our beta testing 😄

    *Keep an eye on this thread for the hangout details 😃

    Posted

  • drphilmarshall by drphilmarshall scientist, admin in response to Capella05's comment.

    Good idea, @capella05. Will you be using a Google spreadsheet so everyone can follow along?
    I will investigate the image viewing capabilities I referred to above, it'd be great to try and get a better view in among those bright lens galaxies.

    Posted

  • Capella05 by Capella05 moderator in response to drphilmarshall's comment.

    Will do @drphilmarshall !

    It will take me another few days to compile everything - but will have it done by the end of this weekend. Then I will work on converting it into a Google spreadsheet.

    Posted

  • drphilmarshall by drphilmarshall scientist, admin

    Great. Will you collect the submitted synthetic images too? I can imagine it being useful to have a flickbook of those, to compare quickly against the data image.

    BTW, have you seen the "Open in Dashboard" button at the top right hand corner of the subject page? Check it out on the ASW0004dv8 page. (You have to do a tutorial first - but when you next click the button, you get straight to the subject.)

    I tried to see if the g-band image showed any sign of a faint counter-image, but its a bit too noisy I think:

    http://tools.zooniverse.org/#/dashboards/spacewarp/520ebbc4e847261af90001c7

    I played around a bit with the color image stretch parameters, but no sign of a counter-image there either. Did I miss something?

    Posted

  • c_cld by c_cld

    Dashboard is great but difficult to fine tweak the sliders.

    Looking at bands by different stretches with Aladin, there could be a hint of a counter image at 3:30. It doesn't correspond to its location in models and is very uncertain.

    My (g-r) image keeps some gray patches at 2:30 and 3:30
    ( ASW0004dv8 )
    ASW0004dv8

    No distinct features seen in the arc...

    Need a better resolution!!

    Posted

  • Capella05 by Capella05 moderator

    Yes, I will collect the synthetic images along with the spaghetti image. Once I have everything together I will research flickbook - at the moment I just flick between tabs on my browser to compare images!

    I am with C_cld with regards to the dashboard - useful, but very tricky to fine tune the slides. Bit off topic, but it would be great to have a tutorial explaining the different filters and what we would expect to see in each of them?

    Posted

  • Capella05 by Capella05 moderator

    I have finished collecting all our modelling attempts of ASW0004dv8 😃

    34 models in all, and I must admit we have several really good ones 😄

    Can't wait for round 2!

    Posted

  • Budgieye by Budgieye moderator

    Total change: moved counter-arc to blue patch to the south, results not too bad, mass is OK. I assume that part of our nice blue arc is masked by the satellite galaxy.

    ElisabethB did something similar
    enter image description here

    enter image description here

    http://mite.physik.uzh.ch/data/007084

    Posted

  • Capella05 by Capella05 moderator in response to Budgieye's comment.

    Hi Budgie,

    We are holding off submitting any new models on this candidate until we can have a hangout and analyse the existing models.
    We need to get an idea of where our models are failing, and where they are really good

    I am in contact with @Psaha and hopefully we should have a date and time settled soon!

    I am all of us are really eager to get started on Round 2 😄

    Posted

  • drphilmarshall by drphilmarshall scientist, admin in response to Budgieye's comment.

    Hmm - the colour of that southern blue blob looks different to that of the arc, to me. I think it's a different source, one being weakly lensed rather than multiply-imaged. Good to see that your model rules it out by predicting a lot of ring-shaped brightness that we don't see! This is important actually: the predictions that the model makes about lensed features that are not there in the data are very useful in ruling out possibilities.

    Posted

  • drphilmarshall by drphilmarshall scientist, admin in response to Budgieye's comment.

    I had a thought about the collected models: if you can agree on which are the good ones, than you should be able to use all of them to estimate the lens model parameters. I don't know what the parameters of the model are, but a measurement of Einstein radius would be very nice! The spread in estimates over all good models will give you a rough idea of the uncertainty on the parameters too.

    Having said this, itll be important to keep models of each type together before estimating parameters. For example, it would make sense to keep all the single clump models together, and then the double clump models in a different group.

    Posted

  • Capella05 by Capella05 moderator

    We are having a hangout tomorrow evening - Fri 2013-08-23 18.00 UTC!

    Please see this for Thread details 😄

    Posted

  • Capella05 by Capella05 moderator

    Hello All 😃

    We had a really good hangout last night, and decided on a action plan moving forward. The good news is - we have a new release of SpaghettiLens later this weekend, and the bug with regards to 'Mass Points' will be fixed - woohoo 😃

    As for the rest...

    Rafi and Prasenjit: are to analyse our Round 1 results and categorise them according to a 'Traffic Light System' - that is:

    • Green - looks good ;
    • Amber - Lets discuss this / a possibility / needs improvement ;
    • Red - eeerrrr, No. The laws of physics do not support this

    Capella05 and C_cld : Once the categorisation is done, they will 'whittle' down the Round 1 candidates that will make the letter.

    ElisabethB and Budgieye : Will make the final approval of the models chosen from Round 1.

    All modellers : Will model the candidate using Mass points (where appropriate!), and post there result in this thread using the following format:

    • Parent - 4516
    • Revised Model - 4568
    • Action - eg. Moved the minima to the NE of the galaxy (as per Budgies initial model) and added a mass point.
    • Result - eg. The arc is more clearly defined from 7 to 11:30, not too happy about the predicted counter image at 2:4
    • Need - eg. A more defined arc to the left

    Capella05 : Will maintain the spreadsheet - I will ensure new data is incorporated. Once I have figured out "Zooniverse letters", I will start writing up a proof version of Round 1.

    Spreadsheet here

    Thank-you to everyone who has contributed so far, we make an awesome team 😄

    Posted

  • c_cld by c_cld

    Parent: 6479

    Revised model: 7200

    Action: Moved the minimum at 4 of the left elliptical (somewhat at the circle center of the arc) ; put the saddlle at the edge of the right elliptical; no change on the contours of arc

    Result: arc well defined, a bit too extended, curvature flatter; counter image seems too prominent but located in core right elliptical

    Need: lower counter image and arc extension

    Posted

  • c_cld by c_cld

    Round 2 after new release of SpaghettiLens with mass points: thanks @Rafael

    Parent: 7200

    Revised model: 7258

    Action: Added mass circles centered on each of the two main ellipticals of the group

    Result: arc well defined thinner, good curvature; counter image faint located in core right elliptical

    Need: looking at clumps by comparison with alternatives

    Posted

  • psaha by psaha scientist

    Another model 7334 based on 7242 taking the large galaxy near the arc to be the main lens, and the other galaxy at lower right to be a point mass.

    Posted

  • Capella05 by Capella05 moderator

    • Parent - 6479
    • Revised Model - 7350
    • Action - Added a mass point to the right elliptical, moved the saddle point slightly to the right. Re-centered the maxima.
    • Result - Arc still looks good, extends from 7 till 1. Mass distribution and contour maps are smooth.
    • Need - The arc needs to extend from 8.

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator in response to Capella05's comment.

    Re 7350
    I've tried to get the arc from 8 to 1, but I'm getting a bit frustrated. Nothing works ! 😉

    Posted

  • Capella05 by Capella05 moderator in response to ElisabethB's comment.

    Lol - glad I am not the only one! I may start making some rather drastic models 😃

    Posted

  • ElisabethB by ElisabethB moderator

    in the hang out the mass points were really visible in the mass distribution, but I haven't been able to duplicate that effect.What did I do wrong ? ! 😄

    Posted

  • c_cld by c_cld

    Parent: 7258

    Revised model: 7403

    Action: keeping a point mass on the right elliptical only, leaving the minimum as it was

    Result: no significative change: arc well defined thinner, good curvature; counter image faint located in core right elliptical

    7403

    Clone: 7405 by tweaking redsifts: much the same as 7403

    7405

    Posted

  • c_cld by c_cld in response to ElisabethB's comment.

    May think that your point mass is outside the mass output image, but is well present in the config file:-)

    My concern is that we don't have reference points between images (at least some reticle for saddle and minima )

    Posted

  • drphilmarshall by drphilmarshall scientist, admin in response to C_cld's comment.

    Hi @C_Cld,

    Definitely agree about the sliders - could do with some finer control there. I went back to the colour view, and tried harder to bring out a counter-image between the pair of lens galaxies. I think I see a very faint smudge there, and changing the displayed colour of the lens galaxies does help a bit in this. Pale orange seems to be a good choice! I varied the non-linearity parameters first, then set the color contrast, then went back to the nonlinearity parameters. This is ssomething like the best I can do, I think:

    http://tools.zooniverse.org/#/dashboards/spacewarp/520ebbc4e847261af90001c7

    What do you think? Can you see a counter-image, or am I kidding myself?
    I'd be interested to see models that predicted counter-images at this position, anyway - ie quite close in to the lefthand lens galaxy, and just below the line connecting the two lens galaxies.

    Posted

  • drphilmarshall by drphilmarshall scientist, admin in response to drphilmarshall's comment.

    Actually, having said that, and looking back, I think many of your models do have counter-images in about the right place - is there any way of showing the lens galaxy positions on the synthetic (predicted) images?

    Posted

  • c_cld by c_cld in response to drphilmarshall's comment.

    Hi Phil,

    Seeing your dashboard I agree with your hint of a counter image at 3:30 oclock of the left galaxy. I had the same assumption with my (g-r) image posted in page 6. That's why I modeled with a red minimum in this location: models
    7200 , 7258 , 7405.

    Lens galaxy positions are seen as white pixel in "Original SourceDiffPlot" image when it was added as mass point like in model 7258. One counter image is close by (the red min), the second one at saddle point.

    In model 7405 (left mass point removed and no white pixel!) there is a third counter image far right int the disk of galaxy.

    That's all for now, even if I wonder how to rule out a tidal trail ASW0004dv8

    Posted

  • Capella05 by Capella05 moderator in response to drphilmarshall's comment.

    I think I can see your counter image - definitely more noticeable in the dashboard!

    We did speak with Rafi about the possibility of having the synthetic image superimposed over the actual image, but in the short term, this is not possible. What Rafi will be able to do, is to allow the spaghetti to be overlayed on the synthetic image. This will at least give us a frame of reference, when analysing the synthetic image. Us modellers will just have to pay more attention when placing the spaghetti, so it is as accurate as possible!

    Posted

  • Capella05 by Capella05 moderator

    • Parent - 7350
    • Revised Model - 7421
    • Action - Moved the right saddlepoint slightly inwards towards the maxima.
    • Result - Arc has improved, extends from 7:30 till 1. Mass distribution and contour maps are smooth.
    • Need - Starting position of the arc still need tweaking.

    enter image description here

    Posted

  • c_cld by c_cld in response to Capella05's comment.

    I did overlay the various output images scaled identically, but I agree it would be faster to get one at once for interpretation.

    By eye visualizing these overlays, there is still the problem of the right curvature (or flatness): an output circle could fit easily an input arc on one point but with greater deviations on three points.

    example: here the center of the input osculating circle of the arc on 3 points is offset by ~1 arcsec from the center of the galaxy, but I can't eyeball it in only output images.

    The difficulty is to get the right identical scaled input and output images before stacking them.!

    Posted

  • drphilmarshall by drphilmarshall scientist, admin in response to C_cld's comment.

    OK, I see. I guess I want the green saddle point image to be more central, because it'll be the brighter of the two counter images (the red minimum should be even more central and demagnified...). Here's an attempt, building on your model 7258 (same arc image positions and lens galaxies), that has an arc with a nice shape to it: 7422. What do you think? I expect something like this model has been submitted before - I just felt like I should have a play with SL 😃

    Posted

  • c_cld by c_cld in response to drphilmarshall's comment.

    It seems a good improvement for me and I follow on this:

    Parent: 7422

    Revised model: 7459

    Action: moved saddlle inside; reduced left point mass; entered redshifts source 0.7 (blue galaxy), lens 0.4

    Result: reduced extension arc with good curvature, counter inside disk, mass distribution more circular and smoother

    Need: don't know

    7459

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 7459

    Revised model : 7479

    Action : added a lot of mass points

    Result : fainter arc, but with some weird effects, arc from 7.30 till 12 - mass distribution and contour maps look okay.

    Need : smoother arc and arc starting from 8

    Posted

  • c_cld by c_cld

    @Capella05 Please modify my update of spreadsheet round2 (can't have the correct format ). Thanks

    I've put a superposition of "Original SourceDiffPlot" with "Contour Plot" and added a red circle and its center as a star (red minimum).

    I suggest that @Rafael should give us something similar but it'll remain difficult to compare outputs without reticles (points of references) as each model has its center on the red minimum which is variable.

    Posted

  • Capella05 by Capella05 moderator

    Hi Claude, Thanks - will update the spread sheet tomorrow.

    Posted

  • rafael_kueng by rafael_kueng

    Hi all

    I just scanned the thread and saw my name mentioned a few times, and wanted to let you know that I'm at the moment very busy writing up my master thesis (about SpaghettiLens and the sim modelling challenge). I will have the first draft in approx 2 weeks, and the presentation and defense of my master thesis on the 26th of Sept.

    Unfortunately, I won't have much time to implement new features until that date. I will catch up afterwards and (try to) give you all the features you need 😃

    Posted

  • ElisabethB by ElisabethB moderator

    Good luck with the master thesis, Rafi ! 😄

    Posted

  • rafael_kueng by rafael_kueng

    thanks 😃

    I've got a small present for you all, check new syn Imgs. Hope thats approx what you were talking about 😃 At least it's what I could do without needing much time.
    Later this weekend I'll rerun all the models in the coll model thread, to recreate the syn imgs in the same style. Just be patient..

    Posted

  • c_cld by c_cld

    Hi Rafael,
    It's very kind of you for taking time to put these marks. I've resent my model 7459 and got a fantastic output http://mite.physik.uzh.ch/data/007820
    which I used for stacking on input image after a small scaling. Looks very great as you could see in the spreadsheet.

    All the best for your thesis 😃

    Posted

  • Capella05 by Capella05 moderator

    Good Luck Rafi! I will try your change out tomorrow 😃

    To everyone else, sorry I have been so absent this week, I will update the spreadsheet tomorrow.

    Posted

  • Capella05 by Capella05 moderator

    I have finished updating the spreadsheet with all the new models.

    I am going to spend the rest of the afternoon, learning Python to import more data 😃

    Posted

  • Capella05 by Capella05 moderator

    I have also added the contour and distribution maps to the round 2 candidates - let me know if it helps!

    Posted

  • ElisabethB by ElisabethB moderator

    Wow ! Great job ! tx Capella !

    Posted

  • ElisabethB by ElisabethB moderator

    I started again with the original model 4516 and introduced lots of mass points.

    Contour map and mass distribution look okay. Arc from 8 to 12 and a very faint counterimage.

    The only (?) weird thing is that the arc is not aligned with the minima and saddlepoint.

    result : 8425

    enter image description here

    Posted

  • ElisabethB by ElisabethB moderator

    Parent : 8425

    Actions :

    • moved bottom saddle point slightly to the left
    • removed one mass point
    • moved top saddle point to the right
    • moved left min. to the bottom

    Result : 9381

    Arc from 8.30 to 12.30. Contour map +/-, mass distribution okay.

    Posted

  • drphilmarshall by drphilmarshall scientist, admin in response to ElisabethB's comment.

    I guess the mis-aligned synimage and image points is a bug, @rafael_kueng?

    BTW What parameters are we keeping track of here? It would be great to be able to present some measurements, Einstein radoius or other mass distribution parameters in particular. I guess that would be conclusion 2 of a Letter, conclusion 1 being that you can find good models for the lens that predict the arc shape in some detail (and that predict a faint counter-image). Great stuff!

    Posted

  • rafael_kueng by rafael_kueng

    hmm.. It doesn't look like a simple bug on the first sight.. I'm sorry, I'll only investigate in detail after 26th (my defense and masters exam, still have to write up a lot, prepare a presentation and learn for the exam 😦 I'm slightly panicking atm..) Afterwards, I'm completely yours again 😃

    Those parameters are not saved at the moment, but I can easily generate a interface that lets you get those. Again, in a few days.. 😉

    Posted

  • Capella05 by Capella05 moderator in response to rafael_kueng's comment.

    Good Luck Rafi - holding thumbs for you. See you again after the 26th!

    Posted

  • Budgieye by Budgieye moderator

    Best wishes from me too.

    Sorry I haven't been posting. This is my busiest week of the year.

    Posted

  • Capella05 by Capella05 moderator

    After chatting with Els last night, I felt I should post a quick update as to the progress of our mini-project.

    Things are happening behind the scenes:

    • @psaha is reviewing our round 1 candidates.
    • I am in the process of updating the spreadsheet - this will be
      complete by the end of the weekend (I see some of the image links are broken - I will fix them at the same time).
    • The next step is posting a list of the best Round 1 candidates, so we know what we are doing right and where we are going wrong. Knowing this we are better equipted for the rest of Round 2.
    • Once Rafi is back with us and has made some bug fixes, we will proceed with Round 2.

    Sorry for the delay - but I feel we will get better results if we wait for all of the above, instead of just randomly generating models.

    I am really excited to get stuck in again!

    Posted

  • Capella05 by Capella05 moderator

    The spreadsheet is now fully updated - I have also resized all the images so they are of the same scale (easier to make comparisons).

    Will let you guys know when we are up and running - in the meantime, are there any new modelers out there that want to take part?

    Posted

  • ElisabethB by ElisabethB moderator in response to Capella05's comment.

    Great job, tx Capella ! 😄
    And I can't wait to take this project one step further. But only after Rafi gets his master thesis out of the way of course ! 😄

    Posted

  • rafael_kueng by rafael_kueng

    @ mis aligned bug:

    This is not a bug in SL. This is really what comes out of the simulation. I don't fully understand why, but the masses cause this misalignment.

    Posted

  • Capella05 by Capella05 moderator

    Just a heads up - I plan on writing the letter next week as I have time off work. So, if you guys could do the following:

    • If you really want to do another model, you have until Friday 😃 Then I will be locking the thread, until the letter is published.
    • As previously discussed in the last hangout, the letter will be mainly based on: How did we go about collaborative modelling // How did we determine what models were successful and what models were failures (along with examples of both) // the final best model along with all the relevant data // positives and negatives of collaborative modelling // and finally feedback from you guys are to whether you see a future for collaborative modelling. So, please send me your feedback with regards to the above by email or PM. If you would like to suggest an additional topic / category, please do the same. If I could have it by the end of the weekend that will be great.
    • Claude - I will be sending you the list of the examples of good / bad examples early next week, if you could narrow them down as previously discussed.
    • Prasenjit and Rafi will help me determine the final candidates that will be included in the letter.
    • Once I have finished the first draft (with your approval) I will submit it to Dr Phil for review etc... Once corrections have been made etc, it will get published.

    Finally, thank you everyone for all your help. We just have one final push to go 😃

    Posted

  • ElisabethB by ElisabethB moderator

    I just had to try another one ;-D

    Contour map looks weird, mass distribution okay.

    Parent = 9381

    Result : 10004

    Posted

  • c_cld by c_cld

    After reading Psaha on spreadsheet, I remodeled and forgot to post it. Here it is

    removing point mass on lens max, but increasing contour size

    making huge point mass on right

    parent =7820

    result = 9951 more circular and clean mass distribution output

    Posted

  • Capella05 by Capella05 moderator

    Locking thread until the letter is completed - see you all in a week 😃

    Posted

  • Capella05 by Capella05 moderator

    Hi Everyone....

    Yes, I know it has been longer than a week - I forgot how long the scientific process takes. Sorry!

    The letter has progressed well and we are just a smidgen away from a final version 😄

    Will post more closer to the time, but I felt you had to know, that a lot of progress has been made, and your models are not in vain!

    Posted

  • JeanTate by JeanTate in response to Capella05's comment.

    The letter has progressed well and we are just a smidgen away from a final version 😄

    That was "a month ago"; is there now a final version? Has it already been published?

    Posted

  • Capella05 by Capella05 moderator in response to JeanTate's comment.

    I apologise - I have been in sub-saharan Africa for the past month - where internet is somewhat erratic.
    It is all but done, but considering I only got back today, it may take a day to post.

    Posted

  • JeanTate by JeanTate in response to Capella05's comment.

    Happy New Year, Capella05, and welcome back!

    Looking forward to the details ...

    Posted

  • Capella05 by Capella05 moderator

    Done 😃

    Letter

    Posted

  • Capella05 by Capella05 moderator

    Now I need sleep - thank you all!

    Posted

  • JeanTate by JeanTate in response to Capella05's comment.

    Very nice, Capella05, well done! 😃

    Posted