Differences

This shows you the differences between two versions of the page.

--- ideas_page [2019/01/29 00:38]
216.165.21.2
+++ ideas_page [2019/03/12 00:04] (current)
wkerzend
@@ Line 31: / Line 31: @@
 ----
-==== Moving TARDIS to Python 3 ====
-**Difficulty:** Medium
-**Astronomy knowledge needed:** None
-**Mentors:** @wkerzendorf, @unoebauer
-**Programming skills:** Python
-**Related TEP:** [[https://github.com/tardis-sn/tep/blob/master/TEP012_gui_overhaul.rst|TEP012]]
-**GSoC Application Tag:** python2
-**Description:** TARDIS is built on Python 2. Python 2 reaches end-of-life in 2019 and with-it several of the developers of the python scientific stack are switching to Python 3 only. TARDIS will need to transition to Python 3 as fast as possible and this project will lead to this transition.
-**Your first objective if you choose to accept the mission:** Identify places where python3 will make the code simpler and issue PRs that show python3 code that is simpler than the current code.
 ==== Expanding the Integration-Testing Framework ====
@@ Line 56: / Line 39: @@
 **Astronomy knowledge needed:** None
-**Mentors:** @orbitfold
+**Mentors:** Vytautas Jancauskas, Ulrich Noebauer
 **Programming skills:** Python
@@ Line 68: / Line 51: @@
 We have two types of tests: unit tests that verify small portions of the code and full-scale integration tests. Both of these test types are implemented with a framework. But the integration tests are difficult to use.
-  * currently we mainly run the integration tests on an external server. We want to integrate them into our general TRAVIS continuous integration routine. We'd also like to expand that and check also different properties of our model against reference data (e.g. electron densities, ionization fractions, dilution factors)
+  * currently we mainly run the integration tests on an external server. We want to integrate them into our general Azure Pipeline continuous integration routine. We'd also like to expand that and check also different properties of our model against reference data (e.g. electron densities, ionization fractions, dilution factors)
   * hand in hand with expanding the verification process we will improve the reporting process which should contain detailed plots and comparison results. A framework exists but is currently not actively used.
-**Your first objective if you choose to accept the mission:** Mark the TARDIS full test as slow and make it easy to enable it's execution only with a commandline option to `py.test` ?????
+**Your first objective if you choose to accept the mission:** Setup the integration tests to run once a week on Azure pipelines given the examples at https://tardis.readthedocs.io/en/latest/running_tests.html
 ----
@@ Line 79: / Line 62: @@
 ==== Atomic Datasets ====
-**Difficulty:** Hard
+**Difficulty:** Medium
-**Astronomy knowledge needed:** Medium
+**Astronomy knowledge needed:** Low/None
-**Mentors:** @lukeshingles, @shaching
+**Mentors:** Luke Shingles, Mark Magee
 **Programming skills**: Python, Parsing, Databases, SQLAlchemy
@@ Line 101: / Line 84: @@
 We created a package that compiles all of this information into a database named Carsus (see http://carsus.readthedocs.io/en/latest/). For this year we aim to strengthen the link between Carsus and TARDIS and also include more atomic data into Carsus.
-**Your first objective if you choose to accept the mission:** Use the script (https://github.com/tardis-sn/carsus-db/blob/master/scripts/create_hdfstore.py) to generate an HDF store - then implement a meta table in the HDF file that stores the units of the columns. ????
+**Your first objective if you choose to accept the mission:** download the atomic data from http://kookaburra.phyast.pitt.edu/hillier/web/CMFGEN.htm and read in the tabulated data contained in si2_osc_kurucz.
 ----
@@ Line 111: / Line 94: @@
 **Astronomy knowledge needed:** Low/None
-**Mentors:** @unoebauer, @wkerzendorf
+**Mentors:** Yssa Camacho, Stuart Sim
 **Programming skills**: Python
@@ Line 117: / Line 100: @@
 **Related TEP:** [[https://github.com/tardis-sn/tep/blob/master/TEP007_isotopes_decay.rst|TEP007]]
-**GSoC Application Tag:** nuclear decay
+**GSoC Application Tag:** new-model-format
 **Description:** The main task that TARDIS accomplishes is to take a model of the exploded star and then simulate the transport of light resulting in a prediction of the observed spectrum. There are currently multiple formats to input the model. For this year, we aim to streamline this process and provide a single format for input. We will also provide converters for the legacy formats.
-**Your first objective if you choose to accept the mission:** ???
+**Your first objective if you choose to accept the mission:** Combine the functionality from read_cmfgen_density and read_csv_isotope_abundances to read a single files that can interpret columns from both functions (https://github.com/tardis-sn/tardis/blob/master/tardis/io/model_reader.py).
 ----
@@ Line 127: / Line 110: @@
 ==== Jupyter notebook widget for TARDIS ====
-**Difficulty:** Medium
+**Difficulty:** Hard
 **Astronomy knowledge needed:** None
-**Mentors:** @wkerzendorf, @unoebauer
+**Mentors:** Yssa Camacho, Laud Bentil
-**Programming skills:** Python
+**Programming skills:** Python, Jupyter
 **Related TEP:** [[https://github.com/tardis-sn/tep/blob/master/TEP012_gui_overhaul.rst|TEP012]]
-**GSoC Application Tag:** gui
+**GSoC Application Tag:** jupyter-widget
 **Description:** Often we need more information about the model and the calculation than the mere spectrum. For example, we frequently need to investigate in detail how a specific spectral line feature forms, which ions and which specific line transitions contribute. For exactly this purpose a Qt-GUI was developed (see [[http://tardis.readthedocs.io/en/latest/gui.html#gui-explanation|here]]). It allows the user to easily analyse TARDIS runs and extract important physical information without knowing the exact inner data structure of TARDIS.
@@ Line 143: / Line 126: @@
 QT for this type of application is no longer the ideal framework. Many applications now work with Jupyter notebook and we want to have a functional GUI that acts as a Jupyter widget. Once this is implemented, we would like to have several example notebooks that exist within the documentation so that it is easy for users to try it out themselves.
-**Your first objective if you choose to accept the mission:** Make a Jupyter notebook and embed it in the documentation that showcases the current way of running TARDIS and plots the spectrum. ???
+**Your first objective if you choose to accept the mission:** Make a Jupyter notebook and embed it in the documentation that showcases the current way of running TARDIS and plots a spectrum.
 ----
@@ Line 153: / Line 136: @@
 **Astronomy knowledge needed:** None
-**Mentors:** @wkerzendorf, @unoebauer
+**Mentors:** Wolfgang Kerzendorf, Vytautas Jancauskas
-**Programming skills:** Python
+**Programming skills:** Python, conda
-**Related TEP:** [[https://github.com/tardis-sn/tep/blob/master/TEP012_gui_overhaul.rst|TEP012]]
-**GSoC Application Tag:** gui
+**GSoC Application Tag:** conda
+**Description:** TARDIS is a tool to analyze transient phenomenon like supernovae. It is important for researchers to quickly install this tool to react to newly observed objects. Anaconda offers a quick and easy way to install python packages and its dependencies. For this project, we want to make it easy to install TARDIS via Anaconda.
+**Your first objective if you choose to accept the mission:** Make a conda packages for TARDIS following the instructions on https://conda.io/projects/conda-build/en/latest/source/recipe.html.
+----
+==== Enable the Dask framework for parallel execution for TARDIS ====
+**Difficulty:** Hard
+**Astronomy knowledge needed:** None
+**Mentors:** Tyler Pritchard, Christian Vogl
+**Programming skills:** Python, dask
+**GSoC Application Tag:** dask
+**Description:** Exploring different simulations of supernovae and comparing them with observations is one of the important tasks of TARDIS. This means that TARDIS needs to be run many tens of thousand times with different inputs. The Dask framework will allow TARDIS to be excecuted on multi-core systems and super-computers. This project aims to combine TARDIS with dask to make it easy to run many iterations of TARDIS.
+**Your first objective if you choose to accept the mission:** Use dask to run distributed TARDIS instances in parallel on your system. Use http://distributed.dask.org/en/latest/client.html as a guide to make a simple example.
+----
+==== Improve the C/Python interface ====
+**Difficulty:** Hard
+**Astronomy knowledge needed:** None
+**Mentors:** Wolfgang Kerzendorf, Christian Vogl
+**Programming skills:** python, cython
+**GSoC Application Tag:**c_interface
+**Description:** TARDIS is a hybrid code built on C/Cython/Python. Currently, there is a lot of interface code to port the python data structures from Python via Cython to C. This makes the code cumbersome to test and try out new things. There are now several new ways that promise to make this easier. We want to try to use cFFI and Apache Arrow to do this.
+**Your first objective if you choose to accept the mission:** use cFFI to provide an automatic interface for testing the parts in the montecarlo package instead of ctypes.
+----
+==== Profile TARDIS ====
+**Difficulty:** Easy/Moderate
+**Astronomy knowledge needed:** None
+**Mentors:** Wolfgang Kerzendorf
+**Programming skills:** python, cython
+**GSoC Application Tag:**asv
+**Description:** TARDIS is a code that prides itself on being relatively fast to compute a synthetic spectrum. We are also continuously adding additional microphysics in the code which sometimes requires additional calculation. It is important to understand how much this microphyiscs adds to the runtime of the code. For this we want to implement a benchmark in asv (airspeed velocity) that can automatically generate a report for us.
+**Your first objective if you choose to accept the mission:** implement a simple benchmark in asv.
-**Description:** TARDIS is a tool to analyze transient phenomenon like supernovae. It is important for researchers to quickly install this tool to react to new observed objects. Anaconda offers a quick and easy way to install python packages and its dependencies. For this project, we want to make it easy to install TARDIS via Anaconda.
-**Your first objective if you choose to accept the mission:** Make an offline conda packages for TARDIS.

TARDIS GSoC 2019

User Tools

Site Tools

Differences

Page Tools