Thursday, June 23, 2016

.astronomy8 unconference sessions #astroatfootie #astroblog

Unconventional Astronomy Outreach #astroatfootie
* Instead of inviting the public to come to us, astronomers should go where the people already are: sports events, concerts, ... people waiting in line who don't have better things to do.  Airports (once you get clearance!).  
 * Toronto ideas: Yonge-Dundas Square, Buskerfest, street fests, free outdoor movie nights, SF fandom conventions, ...
 * Astronomy on tap: ticketed in Austin, free in NYC.
* Astronomers in a pub, unannounced except to owner, wear distinctive outfits "people in orange hats are astronomers; ask them questions".  Or, wear butler outfits / black tie / glow necklaces (1 BSc, 2 MSc, 3 PhD, 4 tenure, etc).
* Astrotrain ... on subway ... ask an astronomer anything.
* Need to be friendly and approachable and willing to say "I don't know".
* Shopping malls have a high fraction of people who aren't looking for you, and don't really care.
* Work half-and-half with scientists & street theatre performers
* Choose your audience, then figure out how to reach them.  Match the outreach personnel to the audience.
* Have a simple script: Come to our open nights. Take math, physics, chemistry, computer science.
* "That's a great question. Have you thought about being a scientist? Because anybody who asks good questions would make a good scientist."
* If your audience is HS students, from communities underrepresented in universities: teen-group concerts, parents/grandparents/younger-oriented events like ice-skating, "Wiggles",
* If your audience is educated: airports.
* Get feedback from happy-buckets: one smiley, one meh, one sad: put a token in your choice.

Academic Blogs #astroblog
* What are they good for?
* In CS, it's tutorials.
* Mine is http://pathallresearch.blogspot.co.uk/
* If you think you have something to say, you should blog.  It's good writing practice.
* Blogs can be cited in the journals.
* Having students select Astrobites.com articles - using it as an accessible journal club for intro astro grad courses.

Other Unconference Session Summaries
#lieflat - see their Google doc
#libdotastro - astronomy libraries ... learn their goals, then ask for your needs in that framework
#astroatfootie (pop-up science) - get feedback and give numbers to funding agencies
#recruit - .astronomy needs a FAQ; answering questions interested parties have; nomination form; rename hack-day to make-day.
#supergit - clarified to  diagram
#divdot - why is astronomy less diverse than some other sciences?  We're not communicating that astronomers learn many skills which are widely applicable in many careers.
#dotmeme - watch the twitter!



.astronomy8 Day 3 Morning Sessions including lightning talks

Software & Data Publishing with the AAS - Tom Robitaille & August Muench

* Software is key to science, and is done largely by junior scientists.
* Need to recognize this in e.g. hiring committees.
* AAS now has a software policy, including accepting software papers (no minimum length) which do not include novel research results, but do include novel or important software
* Recommend publishing software itself, w/open source license, with a static snapshot archived on e.g. zenodo or figshare or dataverse. Those websites will give you a DOI (Digital Object Identifier) which should be permanent.
* When citing software, cite the paper (for credit) and the DOI referrring to the version of the software used (for reproducibility).
* In aastex6, use \software{Code \citep{http://dx.doi.org/...}}

* How could the articles AAS publishes make better use of data objects produced by the community?
* Figure Sets (incl. interactive, and videos)
* Data citation and linking to archives via DOIs

  
Libraries for Astronomers - Daina Bouquin

* Libraries can be incubators for innovation ... from the Atlas of New Librarianship: Libraries empower people to positively impact society by facilitating knowledge creation.
* What are a community's values and aspirations ... how can libraries help you meet them?
* You want to do awesome stuff with library resources; Libraries want to show off what you do.
* Concepts to Guide your Library: Conversation, Ownership, Adaptability, Support.
- Conversation: learning is social; not everyone learns just from exposure to artifacts; instead, build knowledge through collaboration and exposure to new ideas and processes.
- Adaptability: tools will change quickly, so help students learn to problem-solve, to ask questions, to not be intimidated by change.
- Support: safe place to say "I don't know"; to encounter diverse viewpoints.
* Physical resources: Leap Motion Controllers, digital cameras, ...
* Training topics: github, jupyter, python, visualization, ...
* https://github.com/CFA-Library


Palomar Transient Factory Schools - Adam Miller
* PTF has flexible scheduling for moving objects, variables, transients, etc.
* New transients are identified (sometimes classified) within 20 minutes of observation
* Machine learning has been critical
* P48 for the survey, P60 for followup photometry, P200 for followup spectroscopy
* P48 was the POSS telescope; ZTF will fully fill the 48 deg^2 FOV ...15x faster than PTF
* See link above for Time-Domain-Astronomy summer schools
* Summer school participants are mostly graduate students, undergrads w/letters of ref, & postdocs

* General Structure for these sorts of schools:
- science talk immediately followed by a hand-on session working with an iPython notebook on that science topic.
- each problem builds on, and utilizes, skills from previous problems
- lectures and exercises dynamically updated each year based on student feedback
- Marketing is important: know who you would like to attend the school, and invite them
- Size is crucial: ~50 is OK, ~100 probably too large
- Summer programs should supplement traditional courses
- Exclude organizers who don't care and won't put in time
- Build the program top-down: consider the entire school as a single unit and design an arc for it
Q&A:
- Jump over much of the learning curve by involving professional educators - in this case they're mostly involved with the concurrent undergrad school.
- Matthew Graham, La Serena Data School, limited to 40 places (8x oversubscription), including 3 mixed-team project days, working under a professor who "knows what the answer will be".


Lightning talks:
* Demitri: SciCoder.org 8: Yale University - mostly astronomers - 5 days - August 1-6, 2016, $180
* Demitri: nitelite fits viewer for mac
* Demitri+Niall: Astrotweeps - tweet for a week about your science and the astronomy community
* Niall: astronomy outreach at massive sports events or concert events or shopping malls
* 3js is utterly brilliant for 3D visualization (rendering engine, browser adaptable)
* Erik: JWST_DADF at maillist.stsci.edu, github.com/spacetelescope tools incl. specviz MOSviz ...coding sprints using Trello.com
* Jamie Ferguson, University of Glasgow, HCI, Making Astro More Accessible Thru Tech
* Simon skymapper.anu.edu.au uvgriz multi-epoch 2" seeing survey of entire southern sky, releases mag 9-17 in 2018+, <19.5-21.5 2019+
* Alyssa adsass.org ... paper density on the sky works now; next, click on image in a paper and see it in online telescope context: Journal Images in the Sky on zooniverse.
* Kelle:
AstroBetter Blog and Wiki is a way YOU can publish.
Arce.li is a "journal" to aggregate online content and get indexed by ADS.
SciencBetter Consulting, LLC: https://twitter.com/ScienceBetter
* Daina: .io domain profits go to Britain and not to the Chagos Islanders; also beware .book


























Tuesday, June 21, 2016

.astronomy8 day 1 unconference session notes

#dynamap session (first half)

For MANGA IFU survey in SDSS-IV, need to display maps (of few 1000 pixels or less) of galaxy properties.
Want to do it dynamically in the browser.
Switch between images and spectra.
Overlay different properties.
User-choosable colourmap and binning scale.  Save and share plot commands.

FITS viewers: firefly (IRSA, LSST); astroviewapp.com (very controllable, not intuitive); others; all are "nearly ready"

Javascript libraries of potential use: highcharts.com, openlayers.org, leafletjs.com


#intfigs session (second half)

If you want to submit an interactive figure, contact the journal on submission.

What will the data object look like that you, the author, will provide for an interactive graph?
What about standards for the readers, so they know they can view the interactive figures easily?

Brushing is (e.g.) selecting some points in one plot, and highlighting them in a different plot.
Brushing can be done in Crossfilter and D3, and in Bokeh. Bokeh is a Python interactive visualization library that targets modern web browsers for presentation.

AAS plans to move data tables and interactive figures outside the publishing firewall. Hooray!


#astroally session

On twitter, if you're interested in social justice in astronomy, and science in general, you should follow Sarah Tuttle and Chanda. And Katie Mack, Jessica Kirkpatrick, and Sarah Ballard. See also Kelle Cruz' MyMentorNetwork Twitter list.


Sessions I didn't attend
#introjulia : for numerically intensive computing in python syntax
#cooltools : https://dotastronomyteam.slack.com/messages/cooltools/ (not public, sorry)
#pywhereisit: astropy.userecho.com ...spectroscopic tools a bit lacking; discoverability issues


.astronomy8 Day 1 lightning talks

Duncan Forgan & Aleks Scholz
#foundthem https://arxiv.org/abs/1605.02947
Open science: scientists communicate with the rest of the populace, during and after the research has been done.

Ashley Villar
astrobites.org run by graduate students who summarize arXiv for undergraduate students! FTW!  And would love to host useful intro documentation for undergrads, written by anyone!

Edward Gomez
There is no such thing as a stupid question (the comic book): http://bit.do/dotastro-comicbook
Scientists ask questions until they are satisfied by the answers.

Coleman Krawcyzk
Tactile outreach for the visually impaired

Matthew Graham
NOAO data lab datalab.noao.edu as a Science Exploration Format: images, catalogs (DEC outputs, DESI inputs, etc.)  Software will be public.  Similar to tools proposed by ESA, GAIA, LSST, SKA.

Justyn Campbell-White: Classifying Stellar Bubbles
Using Citizen Science Data for PhD project.

Adam Avison
http://jb.man.ac.uk/pynterferometer (note 30% chance of crashing on a Mac)  Simulating results from a given interferometer setup. 

William Roby
A history of JavaScript. Dark Ages of 1990s due to fragmentation. Today: ES6 improvement; 'strict mode'; standardized API. Very interesting for programmers. Don't be afraid to try it.

Edward Gomez
Asteroid Day is June 30th, 2016. http://asteroidday.lcogt.net  Allows you to schedule your own observations of asteroids, stitches them together into movies!



.astronomy8 Second Morning Session

Ali Swanson - Citizen Science with the Zooniverse: turning data into discovery

People are uniquely capable of classification.

Anyone can participate; many different people contribute, then engage in a broader conversation.

Snapshot Serengeti: 1 million classifications in first 3 days.  Took 6 months of a full-time developer to set up.  See below for how to DIY more easily!

By not allowing "I don't know" as an answer to what animal is in an image, can identify images with unknown animals as those where people's answers vary wildly.  Populace agrees with experts 97% of the time; experts with experts 98% of the time; half of the disagreement is on cases experts called 'unclassifiable'.

Zooniverse Project Builder: DIY projects!  www.zooniverse.org/lab

Break complex tasks down into single steps (as easy as, swipe left or swipe right!).

Data (images) hosted on Amazon S3, though not necessarily Zooniverse's Amazon S3.


James Gilbert - Gaffa tape and string: Professional hardware hacking (in astronomy)


What makes a hack a hack?  A need for results but a lack of time.  E.g. funding proposal deadlines.

Worked at AAO & Oxford on multiobject spectroscopy.
* Starbug w/vacuum suction proceeded only after a proof of concept hack was demonstrated.
* Echidna spines demonstrated just in time.

Managers - don't be afraid of the hack spirit!

 
Geert Barentsen - Open Science with K2

Open science is the movement to make scientific research, data and dissemination accessible to all levels of an inquiring society. -Wikipedia.   Such a society might have more understanding of science, and give more support to it.

Using proprietary data as a reward system bears significant costs. We should be able to have more than one paper on the same data or discovery.

Don't fear your colleagues; fear a lack of public support for science.

Kepler got rid of its proprietary data period.  Currently running with K2.  Data available 2 hours after downlink.  K2 is a hack that uses photon pressure to keep the spacecraft pointing stable (though only along the ecliptic, and only for ~80 days at a time per field).  K2 funds ($10-150k) to do the same science.  Large programs required to release value-added data and software.  Early-career researchers benefit the most from K2's openness.  Ranked Excellent in all 3 categories in NASA senior review.  NASA's TESS will have open data from day one. 




















.astronomy8 Day 1 - Morning

Becky Smethurst: Introduction
If you're reading this, no matter when, you're part of the .astronomy community.  Be a sponge and learn from the community.

Sarah Kendrew: .Astronomy and JWST
* Aug. 2017: Early Release proposal call (500 hrs, no proprietary period)
* Nov. 2017: Cycle 1 proposal call
* Late 2018: Launch
* Launch + 6 months: Early Release observations done; Cycle 1 observations start

Tools for research: jwst.stsci.edu
* online instrument simulators, new exposure time calculator capabilities, new data analysis tools & astropy, WebbPSF

Bruno Merin: ESASky: http://sky.esa.int
* For Astronomers and the Public
...got distracted looking up Hall's Arc; see my twitter @PatrickBHall ...

Carl Ferkinhoff, Hardware .astronomy
1) low(er) cost astro instruments
2) invest undergrads in development
3) open-source it all
4) develop principles of open-source hardware astronomy
hardwareDOTastro.github.io

ROBH.aT idea: Lots of campus telescopes out there ... robotic/automated observing is the future ... skills learned in roboticizing telescopes are applicable to industry... can we roboticize for $5k/scope?
1) Automate Dome Control
2) Automate the telscope: ROBH.aT
3) Network automated scopes for studies of exoplanets, variable stars, SNe, LSST followup, others?
Q&A: LCOGT has a uniform set of telescopes; some groups have thought about networking a set of heterogeneous telescopes, incl. itelescope.net

Monday, June 20, 2016

Philipp Plewa: D3.js (Data-Driven Documents) Java Library: Let's Make a Star Chart

D3 is a data visualization library (not a chart library like plotly.js or Vega-Lite).

D3 maintains a mapping between your data and graphic elements.

Can use SVG (Scalable Vector Graphics).  Or, canvas (raster-based, faster for many elements).

Examples: https://bl.ocks.org/
Editable examples: http://blockbuilder.org/

Good colours: http://colorbrewer2.org/

Star chart with subtle colour and twinkling: www.bit.ly/217O2C5


Lisa Ballard - Space Hacks with WordPress and Python

Wordpress started for blogs, but can be used for websites.
Mobile-optimized wordpress themes and plugins.

Can post to your own wordpress site via python! Use:
from wordpress_xmlrpc (Wordpress XML-RPC plugin) [Remote Process Controller]
Used Jinja2 templating interface: http://jinja.pocoo.org/

Exoplanet Ball Pit: can you scale exoplanet sizes from Earth to Jupiter and fit them all in a reasonably-sized area?

WikiSky
virtualsky.lcogt.net
Info on specific astronomical objects: www.strudel.org.uk/lookUP

Day Zero Lightning Links (.astronomy8)

Arna Karick: http://techsavvyastronomer.io/
...including some tutorials for useful software packages.

Jonathan Fay: www.worldwidetelescope.org - now sponsored by the AAS.  Including a free Windows application, a web client, and tools for integration into your website.

Carl Ferkinhoff: Open Science Framework: online lab notebooks: https://osf.io/
Lots of functionality: read-only versions, integration with github, ...

Chris North: www.chromoscope.net
...lots of multiwavelength capability, less zooming capability.

Becky Smethurst: to let you know your python code has finished running: os.system('tput bel') ...details on twitter a few weeks back




Demitri Muna, Languages 101 (.astronomy8)

Opinion: for a cross-platform user interface, use a web interface.

In python, you need to use a third party library to create a user interface.

MVC Design: Model * View * Controller

Model: the representation of the data you're working with (typically a database)
View: how you present your model data (e.g. a web page)
Controller: retrieves data from model and passes it to view for display (e.g. a python script)

Model doesn't care how data is viewed; View doesn't care how data is stored.

Flask & Django are python frameworks that employ MVC design.

Avoid PHP - it grossly violates MVC design.

Create maps, add markers and pan/zoom, with JavaScript library Leaflet: http://leafletjs.com

JSON (JavaScript Object Notation) to exchange data with another program.  In python, "import json" and use json.dumps() & json.loads()

If you need a database, use only one of these two:
*PostgreSQL - must be installed on a server; can handle many users, easy support for indexing points on a sphere, e.g. cone searches (q3c)
*SQLite - already installed on your computer.  For your convenience.

* Object Relational Mapper (ORM) - sits between a database and objects/classes
* You interact with an ORM in an individual Session

* SQLAlchemy: Demitri, Brian C, ...
import sqlalchemy
... # create org and student objects
session.add(org)
session.add(student)
session.commit() # saves data

***Advice: Learn to write in an object-oriented coding style***

* APIs: Dan F-M, Tom R, Brian C, Demitri M, ...
* To let other people build their own applications on top of your data, you will need to design an API, e.g. using Flask, and host it on a server.  Example: http://api.sdss3.org/api_spectrum.html
* In principle, the View can stay the same as the Model changes, but the Controller must change if the Model changes.
(The API can stay the same as the data format changes.)

* JavaScript visualization: D3.js is the premiere data visualization library for JavaScript.  Very low-level and therefore complicated.  

* Convert C/C++/Fortran to JavaScript: http://emscripten.org
e.g. CFITSIO conversion: https://github.com/astrojs/fitsjs

Question: How to figure out how new a computer needs to be to run a particular application?
Answer: everything discussed here can be run on a Raspberry Pi $30 computer... can get a free Windows license for same.








Dr. Arna Karick: What is a hack? (.astronomy8)

A hack is a small project that solves one problem, designed and executed as quickly as possible.

Not just technical -  can be artistic (e.g., a play).

A mix of people makes for a good hack day.

Teaching hacks: an online resource about how to do a thing can be very useful.

Interesting existing hacks: those of
Erik Tollerud (Astro Object Databases)
Ruth Angus (ArXiv Beta)

Useful websites:
Slack
Trello
LightTable
CodePen
Plotly
CartoDB & LeafletJS [mapping!]
D3js.org





.astronomy 8 (Dot Astronomy Eight) Intro

Here I am in rainy Oxford, along with numerous other astronomers, for the .astronomy 8 conference!

Notes on Sarah Kendrew's retrospective:

DotAstronomy meetings started in 2008 in Cardiff.  (I bet it rained then too.)

DotAstronomy is about: how can we use the internet to do astronomy? (Including scientific research, public outreach & engagement, and public education, not to mention blurring the lines between those topics.)

At .astro meetings, people often talk on subjects outside their main research interests, in part because the meetings are not focused on a single research area, and include non-astronomy speakers as well.

dotastronomy hacks are about creativity, or software, or science, or any combination thereof. Hacking is a way to learn. Hacks of the past:  Chromoscope - optimize.js - d3po.org [interactive plots] - OS X widgets for Ned Wright's Cosmology Calculator.




Tuesday, March 15, 2016

I like my coffee dark... like my skies.

Ever sit down at the kitchen table with your morning coffee and find that an 8-meter-diameter telescope in Hawaii is right at that moment observing something you told it to observe?

Me neither.  Until this morning.  (The magenta squares mean observations are "ongoing".)


Soon the secrets of the spectrum of SDSS J114546.21+032251.9 will be mine!  Bwah ha ha ha ha!