Profile

Cover photo
Doug Burke
Works at Smithsonian Astrophysical Observatory
11,527 followers|200,621 views
AboutPostsPhotos
People
Have him in circles
11,527 people
Christina Heikkila's profile photo
Herb Sawyer's profile photo
Phillip Ibarra's profile photo
Michael Rainey's profile photo
Sayed Ahmed's profile photo
Sergio Alonso's profile photo
Kalle Last's profile photo
Emir Şevkioğlu's profile photo
Lee Gray's profile photo
Work
Occupation
Astronomer
Employment
  • Smithsonian Astrophysical Observatory
    Astronomer, 2001 - present
  • Institute for Astronomy, University of Hawaii
    Astronomer, 1999 - 2000
  • Astrophysics Research Institute, Liverpool John Moores University
    Astronomer, 1997 - 1999
Story
Tagline
An astronomer who spends way too much time on the computer.
Introduction
I am an Astronomer working at the Chandra X-ray Center, which is part of the Smithsonian Astrophysical Observatory, in Cambridge, MA. Research interests include observational cosmology, galaxy clusters, AstroInformatics and the semantic web.
Basic Information
Gender
Male
Other names
Douglas Burke
Apps with Google+ Sign-in
  • Plants vs. Zombies 2

Stream

Doug Burke

Shared publicly  - 
 
When writing last week's 'simulating and fitting 2D data in Sherpa' IPython notebook, I realised that I can try out the object API of the pyBLoCXS module (Bayesian Low Count X-ray Spectral), and so compare the error estimates from the "standard" X-ray Astronomy techniques (covariance and confidence) with that from a Monte Carlo Markov Chain (MCMC) analysis. So I threw this notebook together today (mainly to show how to do it, rather than discussing why or interpreting the results).

As a reminder, the other notebooks in the series, of using Sherpa for modelling and fitting data in Python - can be found at  https://github.com/DougBurke/sherpa-standalone-notebooks
In my last notebook, I showed how to estimate errors for a simulated 2D data set, using the object API of Sherpa. This analysis lies in the frequentist camp, so to show that Sherpa has no biases (the rather-poor pun intended), this notebook is going to highlight a more Bayesian style of analysis ...
1
Add a comment...

Doug Burke

Shared publicly  - 
 
Another IPython notebook showin off some of the features in Sherpa, our Python fitting and modelling package. This time I talk about how to build it so that you can use the models from XSPEC and, thanks to a discussion with a colleague, how you can write a model that extends (or changes) the behavior of an existing model (rather than having to rewrite it from scratch).
I was recently asked by a colleage whether Sherpa allowed you to subtract a model. This is "easy" - since you can create model expressions using + , - , * , and / - but it turned out that what was really wanted fitted into my current documentation obsession, writing user models in Sherpa.
1
1
Boris Borcic's profile photo
Add a comment...

Doug Burke

Shared publicly  - 
 
To round the week off, here's a notebook (building off of yesterday's one, which is linked to at the start of this one, for those who need some context) showing how to create plots with Sherpa when using the low-level API for fitting and modelling. It also shows that you can still use matplotlib commands with these "pre-canned" plots.

Another "Yay" for IPython notebooks.
In the last two notebooks I've created, I've shown how to create a user model (using add_model rather than load_user_model ) and use the "low-level" interface$^\dagger$ to fit a data set (that is, without using the routines in sherpa.ui or sherpa.astro.ui ). One thing I haven't done is show how ...
1
Add a comment...

Doug Burke

Shared publicly  - 
 
A colleague - on reading my recent post on how to add a user model to our Python-based modelling and fitting environment (Sherpa) - asked why I was bothering with the CDF rather than use the PDF. This is a good example of expanding the example, since it covers writing an "integrated" model and how to write a "guess" function.

It also meant I could avoid writing the documentation I'm supposed to be writing, by writing other documentation!
In the previous notebook - User models in Sherpa - I showed how you could write a user model to fit the Cumulative Distribution Function (CDF) of a data set with a Gamma distribution. Rather than use the load_user_model function, I showed how to write the model taking advantage of the Sherpa ...
1
Add a comment...

Doug Burke

Shared publicly  - 
3
Add a comment...

Doug Burke

Shared publicly  - 
 
 I keep on forgetting to mention when I post a new entry in my exploration of using Haskell - in particular, IHaskell https://github.com/gibiansky/IHaskell/wiki which builds off of the Jupyter Project work - for Astronomy.

Previous entries can be found at my cheapo-blogging-platform™ (ake GitHub + markdown): https://github.com/DougBurke/astro-haskell/blob/master/README.md
GitHub & BitBucket HTML Preview. Enter the HTML file to preview: or prepend to the URL: http://htmlpreview.github.io/?https://raw.github.com/twbs/bootstrap/gh-pages/2.3.2/index.html. or use this bookmarklet while browsing GitHub or BitBucket: HTMLPreview. Developed by niu tech | Contribute on ...
1
Add a comment...
Have him in circles
11,527 people
Christina Heikkila's profile photo
Herb Sawyer's profile photo
Phillip Ibarra's profile photo
Michael Rainey's profile photo
Sayed Ahmed's profile photo
Sergio Alonso's profile photo
Kalle Last's profile photo
Emir Şevkioğlu's profile photo
Lee Gray's profile photo

Doug Burke

Shared publicly  - 
 
Continuing my IPython notebook blitz, explaining parts of Sherpa, our Python fitting and modelling package which is on GitHub (https://github.com/sherpa/sherpa/). This on e shows how you can simulate and fit a 2D (i.e. image) data set using the object API, and do some error analysis on the parameters. It probably best makes sense when read as part of the series, which can be found at https://github.com/DougBurke/sherpa-standalone-notebooks
Thanks to some inspired thinking by a colleague - left anonymous, to protect the innocent - I am going to try the name "the object API" - and variants thereof - to describe what I've previously called the "low-level API". That is, the API you get from using the Sherpa classes directly, ...
1
Add a comment...

Doug Burke

Shared publicly  - 
 
More publicity for the IPython notebooks I've been writing recently to show how to use our Python-based fitting and modeling library now that we've made it an "open" project (it was always GPL based, but now the code is on GitHub and we welcome contributors).
I have started to collect together a few IPython notebooks highlighting a few features of Sherpa - in particular the standalone version. They can be found on GitHub at https://github.com/DougBurke/sherpa-standalone-notebooks/...
2
Add a comment...

Doug Burke

Shared publicly  - 
 
Finally got around to writing up a simple example of adding your own model to Sherpa. In this case I wanted to fit a cumulative distribution with a Gamma CDF. It also shows an example of using the under-documented "direct access" API.

This works with the version we distribute as part of CIAO as well as the version on GitHub https://github.com/sherpa/sherpa/.
I was recently working on a way to charactrize the draws from a MCMC chain - in this case generated with pyBLoCXS but that's only relevant here because we are interested in Poisson-distributed data, hence the Gamma function - and came up with a simple user model for Sherpa.
1
Add a comment...

Doug Burke

Shared publicly  - 
 
I'm in a link-sharing kinda mood today: #DataScience for #Astronomy school in La Serena.
AURA Observatory in Chile LSST Corporation National Optical Astronomy Observatory Virtual Astronomical Observatory National Science Foundation (U.S.) CONICYT (Chile) Universidad de Chile-CMM/NLHPC Millennium Institute of Astrophysics Universidad de La Serena ...
1
Add a comment...

Doug Burke

Shared publicly  - 
 
Interested in being the new Director of Citizen Science at Chicago's Adler Planetarium? Then I've got the link for you...
Director of Citizen Science The Citizen Science Department at the Adler Planetarium seeks to engage everyone with real science, whether online or in person. The department at the Adler Planetarium is home to the Zooniverse, including a development team who design and build novel citizen science projects and, with colleagues, the infrastructure that supports them. They have played a leading role in making Zooniverse the world’s largest and most su...
1
Add a comment...

Doug Burke

Shared publicly  - 
 
I've had some spare time on my hands recently, thanks to the snow we've been getting in the Boston area. So, I decided to use it to continue my exploration of communicating with notebooks (in the IPython sense of notebook).

Here's a link to an IHaskell notebook (so the IPython notebook machinery but using Haskell instead) which shows some simple Astronomical calculations (in this case, Cosmological ones, such as the angular-diameter and luminosity distances), and trying out one of the "units" packages for including units in a calculation. Comments can be made on the GitHub page https://github.com/DougBurke/astro-haskell/issues  (or here ;-)
GitHub & BitBucket HTML Preview. Enter the HTML file to preview: or prepend to the URL: http://htmlpreview.github.io/?https://raw.github.com/twbs/bootstrap/gh-pages/2.3.2/index.html. or use this bookmarklet while browsing GitHub or BitBucket: HTMLPreview. Developed by niu tech | Contribute on ...
1
Add a comment...