This is a page to record decisions on what is in or out for a Minimum Viable Project.
The purpose of a Minimum Viable Product should get a practical, extensible, maintainable, well-documented product into the hands of users as fast as possible. There is a delicate balance between what is too minimal to be particularly useful and what is too ambitious to be viable. As a worksheet we can begin to construct a table of features/capabilities and categorize them in terms of importance to the user and viability for an early release based largely on existing code and/or calibrations and/or reference files.
MVP worksheet
Feature/capability | Importance for MVP (must, should, nice) | Difficulty to deliver in MVP (high, medium,low) | Comments |
---|---|---|---|
user-friendly APIs | should | high | This is at least for a quick/dirty run. This might be similar to what Nimish Hathi mentioned. This concern is also including how data and processed data are encapsulated in variables inside the working environment (e.g., Jupyter). For example, considering the aXe outputs (check this ticket in Jira for some info ASTROGRISM-45 - Getting issue details... STATUS ), there are many files produced and this is not obvious how to access these saved outputs. It would be nice if we have a wrapper for reading these outputs back into the working environment, and we design how a user can access mostly used information easily such as x.trace, x.wavelength, x.flux. |
modularity | must | medium | This would be useful especially when we think of extendability either adding new grism definition from different facilities, or user customized objects from base classes. |
Interactive GUI | nice | high | Interactive graphic user interface will help users to quickly examine visually. |
documentation | must | medium | This is important, as we all can agree |
minimal code comment standard | should | low | We should discuss about what would be the minimal requirement for code commenting. Have these requirements noted down, and make sure that all codes complied to the minimal standard before accepting any push to the main code body. We might assign someone specifically to go through all codes and check for the compliance. |
Tutorial / cookbook | must | low | This should explain itself how it import it is. |
minimal functionalities in prototype | must | medium | At minimum, the prototype should be able to take inputs = {a grism image or a set of dithered images, other association files... such as direct image, background image, etc.}, and give outputs. There should be APIs for users to easily re-run with different parameters (as compared to aXe, all outputs will be in OUTPUT folder, a user has to manage saving this folder separately in a different name, and re-run almost the whole codes just to produce outputs with different parameters. From my aXe experiences, I think extracting grism spectra composes of i) locating a spectrum, ii) compute trace and wavelength given a spectrum, iii) extract (in aXe sense, this is the SPC files), iv) post-extraction calibration (e.g., aperture correction, combining spectra with outlier detection and rejection algorithm in case use avoid performing extraction on a drizzled image, or flux scaled to photometric points... aXe does not perform these steps. |
Identify associated data sets | e.g. Find and download direct and dispersed images that overlap on the sky via an archive query | ||
Organization and bookkeeping | Conventions for file formats (in and out) metadata in files, file names, directory structure, output files (e.g. column names and units) | ||
Geometric transformations | Outline all of the variants and what the use cases are (e.g. elaborate from Nor's presentation) | ||
Astrometric registration | Align dithered observations | ||
Simulations | Create a simulated 2D dispersed spectrum from a 1D spectrum and image morphology | ||
Background subtraction | What are the different background components & approaches to estimating/subtracting for HST instruments? | ||
Flatfielding | This can be subtle; the same approach can't be used in all circumstances. Maybe multiple user stories are needed? | ||
2D spectral extraction | Maybe multiple stories with different approaches to getting a 2D extracted dispersed spectrum? | ||
Co-adding | Maybe several stories with different approaches to co-adding spectra taken at different orientations? | ||
Find an isolated emission line | EM2D use case | ||
Create an emission-line map | Create a 2D emission-line map from spectra taken at different orientations | ||
Fit a set of templates | Varying flux and redshift |