SpaghettiLens - Updates, Bug fixes etc...

Hi all.

I'll open this thread to announce news on the SpaghettiLens modelling program. So keep an eye in here 😃

Today a big update happened:

Input image is finally saved, a new save model dialog
Input pane got a speed up
Log in process streamlined (saving defaults in a cookie, [Enter]ASW000XXX[Enter] gets you started)
Ability to hide all spaghetti (see display settings)
Made the result view prettier 😃

Posted July 8, 2013 12:58 PM

SpaghettiLens v1.4.1 is online now:

On the result page (like http://mite.physik.uzh.ch/data/3955) you have now more options under the "actions" section (at the bottom):

you can directly try for yourself to model this image ("start from scratch")
or you can try to improve an already existing model ("revise this result")
you see if this model has any 'parents' (is a revised version) and / or 'children' (has been revised)

the second point can be used to continue to work on own models, and also to try to improve others models. (No worries, in any case there will be a new result and the old one won't be touched)

Posted July 24, 2013 5:06 PM

by psaha scientist

To supplement the SpaghettiLens tutorial, here are some models of very clean systems.

These are lensed quasars that were discovered in the 1980s and 1990s and have been studied extensively, in the course of which they have been observed at much higher resolution by HST. That makes them much easier to model than new discoveries on Spacewarps. They may look like toy lenses compared to Spacewarps candidates 😃

There are six examples here, and they are like archetypes. Pretty much all lenses where a single galaxy dominates are some kind of variation on one of these six types.

Straight doubles like HE1104-180
(model) are the simplest. In a quasar you just get two pointlike images. But if the source is a galaxy the images, especially the more distant one (a minimum) stretches into an arc.
Bent doubles like SBS1520+530
(model.) The bend is a result of one or more other galaxies perturbing the gravitational field of the main galaxy.
Symmetric quads like the Einstein Cross Q2237+530
(model.) If the quasar had been placed in exactly the right place, we would have seen an Einstein ring. But as the source is slightly displaced, we see a quad. Note the slight departure from symmetry -- that is the clue to which images are minima and which are saddles. Again, had the source been a galaxy, we would have seen four arcs, perhaps joining into a partial or full Einstein ring.
One-near three-far quads as in B1422+231
(model.) Here the source is further away from the spot needed for an Einstein ring, and the displacement is along the long axis of the gravitational field.
Three-near one-far quads RXJ0911+055
(model.) Here is source is displaced along the short axis of the gravitational field. Note how different the spaghetti loops are from the one-near three-far case. This is the trickiest configuration of the six.
Asymmetric quads PG1115+080
(model.) This configuration arises when the source position is displaced from the spot needed for produce and Einstein ring, but that displacement corresponds to neither the long axis or the short axis of the gravitational field.

It is not necessary to remember all this information in order to make models, but with practice one develops some intuition for the six configurations. The one important thing is that one-near three-far is very different from three-near one-far.

You can experiment with these systems using the new `revise this result' feature on SpaghettiLens. You may want to turn down the radius in pixels from 10 or 12 down to 8, to get results faster.

Posted July 27, 2013 1:44 PM

by rafael_kueng

SpaghettiLens v1.5.0 online:

more natural Synthetic Image

unfortunately I could only update one of the working machines, so if many users are modelling you could get some bigger wait times. I'll be able to upgrade the rest next Monday. Please don't run models with pixradius 12 if not necessary. (do one with pixradius 8, you can load the result and improve pixradius next Monday) I'm really sorry for this inconvenience. The reason is that the computer system here at university needs an update, and i don't have access to it.

please let me know it the default values for the image brightness are to low or to high

the original SynthImage can still be seen, but it's hidden in the results page under "advanced output images"

Posted July 31, 2013 4:23 PM

by rafael_kueng

not really an update, but some good news:

I got the worker nodes updated, so you now have approx. 8 times the power available (4 additional 4 core machines @ 2.8GHz doing the workload, besides the old 8 core @ 1GHz machine)
You can expect it to be approx twice as fast for doing one model, and up to 6 person can now get models at the same time without waiting in the queue.
So go ahead with pixrad 12 models 😃

The bad news: these machines are only borrowed.

Posted August 8, 2013 4:39 PM

by rafael_kueng

SpaghettiLens v1.6.1 is online since yesterday 😃

Here is a list of changes:

external masses are working
better ext_mass ui element
better ruler2 ui element
ratio arcsec/px := pxScale now parameter
tooltips
keyboard shortcuts 1234,qw,asd,zxc
naming and new placement of output images
div. svg ui bugfixes, strange behaviour when ruler ontop...
added tools app section in django
1. tool: ResultDataTable
first traces of an api..

Posted August 25, 2013 10:04 AM

by rafael_kueng

ResultDataTable tool:

http://mite.physik.uzh.ch/tools/ResultDataTable

This is a tool for people trying to get all the additional hidden data for results / models.

Basically you use it by adding things directly to the url:

?id=10,2,3-8,11 the ids of the results/models you want to retrieve, komma separated, or ranges. NO empty spaces.
add &type=csv if you'd like to get a csv (Excel) table instead of html.

All the options are explained if you go to the url without any arguments.

There is one restriction: Browsers usually don't support urls that contain more than 2048 chars. That means you can't retrieve more than ca. 300 unique ids in one run. Let me know if this becomes a problem.

Posted August 25, 2013 10:13 AM

by rafael_kueng

For people wanting more control or an automated way of getting things, there is also a data api (application programming interface) you can access with any programming language that can do post requests. Results are json strings (javascript objects)

http://mite.physik.uzh.ch/api

It can't do much at the moment, only return data for results. If there is demand, I'll extend the api. (or you do it yourself, the whole project is open source 😃 )

An example in python 2.7 how to get and print all the data for result id 3000:

import requests as rq                   # maybe install it first with 'pip install requests'
url = 'http://mite.physik.uzh.ch/api'
pdata = {
  'action': 'getResultData',
  'rid':3000                            # rid is the result / model id
}
result_data = rq.post(url, pdata).json()

# 'result_data' is a python dictionary containing all the data as key/value pairs.

for key, value in result_data.items():
  print 'key:', key, 'value:', value

Posted August 25, 2013 10:45 AM

by rafael_kueng

SpaghettiLens v1.6.2 is online

list of changes:

SynthImage has now input points drawn on top of it (min, max saddle as colored dots and PointMasses as crosses)

Posted August 30, 2013 3:41 PM

by rafael_kueng

SpaghettiLens v1.6.3 is online

minor service update, some changes on the backend for offline plot generation

Posted September 4, 2013 5:12 PM

by rafael_kueng

Those who were wondering:
The university did a computer system update, and I failed to get the message.. So all the simulation worker were stuck and didn't produce any results, starting from yesterday evening till now (CEST). It is fixed and working again.

Posted September 18, 2013 12:51 PM

by c_cld

Today Oct 5 and yesterday, SpaghettiLens seems endless even for very low estimated duration. Is it a server overloaded queue or an issue in SL? Closing tab (and loosing input) appears the only solution. Hope SL will be up running next week.

Posted October 5, 2013 4:05 PM

by Capella05 moderator

It is down for everyone - I will email Rafi on Monday.
I didn't want to disturb his weekend 😃

Posted October 5, 2013 4:41 PM

by rafael_kueng

Oh sorry for the downtime! It's up and running again. All the servers hung up. I'll invesigate the problem..

Posted October 8, 2013 2:05 PM

by rafael_kueng

Hi there

Sorry for being rather quite the last few weeks. There's a lot up at the moment, I'm quite busy working as a teaching assistant at the university at the moment.
I saw there was a downtime today, because of some networking errors from the network administrators. I installed a work around until Sysadmins get the system back up running. It should work as usual.

Best Rafael

Posted November 11, 2013 9:12 PM