synspec by tcjansen · Pull Request #17 · tcjansen/GSoC

tcjansen · 2019-07-15T20:34:18Z

A place for comments on "synspec" - simulating spectroscopy with synphot tools.

Co-Authored-By: P. L. Lim <2090236+pllim@users.noreply.github.com>

… out binset and forced tapering in Observation calls

…ot finished)

…ce.py

…unction name

pllim · 2019-07-15T20:35:56Z

README.rst

@@ -1 +0,0 @@
-Browse the branches to see the different notebooks I've been working on!


Why do I seem to see this diff in multiple PRs? Is this intentional?

Ah, that's because I don't copy the README over to the new branches

tcjansen · 2019-07-15T20:45:44Z

Copying some questions from Slack:

wl_in_center = ((1.1 - 0.33) / 2 + 0.33) -- why not apply u.micron here instead of later?

I decided not to put it there because I couldn't figure out how to get np.arange() to play nicely with unit quantities in this line:
spec_wavelengths = np.arange(0.33, 1.1 + delta_wavelength, delta_wavelength) * u.micron

For a bandpass that is equivalent with "observing through no filter", why not use Const1D with amplitude=1? Does it have to be a box?

Nope - Box1D was just the first model I saw that worked :P. But Const1D will work better, thanks.

pllim

Some minor comments. I'll defer to @bmorris3 and @eteq for the science side of things.

pllim · 2019-07-15T20:39:33Z

synspec_examples.ipynb

+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "WARNING: W35: None:5:0: W35: 'value' attribute required for INFO elements [astropy.io.votable.tree]\n",


These warnings show disappear if you use yet-to-be-released astropy 4.0 😬 . In the meantime, if they bother you, you can filter them out with https://docs.python.org/3/library/warnings.html#overriding-the-default-filter .

synspec_examples.ipynb

bmorris3 · 2019-07-23T06:24:09Z

synspec.py

+    binset = np.array([first_wl])
+    binwidth = first_wl / R
+
+    while binset[-1] <= waverange[-1].to(u.angstrom).value:


I haven't though very hard about this yet, but is there a way to write this without a while loop, to keep it speedy?

Will think more about this!

bmorris3 · 2019-07-23T06:24:45Z

synspec.py

+    Returns a spectrum with bins defined by a constant bin width specified
+    by the user.
+    """
+    if not waverange:


I think it's a little clearer what you mean if you write this as if waverange is None:

pllim · 2019-07-23T15:41:50Z

synspec.py

+import numpy as np
+
+
+def from_R(source_spectrum, bandpass, R, waverange, force='none'):


R is also another high level programming language. And typically uppercase variable signifies a constant, so I propose renaming R to resolution, spectral_resolution, or something like that.

pllim · 2019-07-23T20:33:33Z

synspec.py

+    binset = np.array([first_wl])
+    binwidth = first_wl / R
+
+    while binset[-1] <= waverange[-1].to(u.angstrom).value:


First of all, I don't think this can be vectorized easily as binset[i] depends on binset[i - 1]; and also we don't have to worry about speeding it up until it is proven to be a bottleneck; and even then Cythonizing it might be good enough.

However, I am concerned that this algorithm does not round-trip. For example:

waverange = [5000, 5500] * u.AA R = 100

Following the math, I think binset here is actually bin edges. And I think you mean in your docstring that you want the bin's central wavelength divided by its bin width is R, correct? So, having calculated binset with this algorithm, my attempt to round trip as follows:

dw = binset[1:] - binset[:-1] # Bin widths wcen = binset[:-1] + dw / 2 # Recalculate bin centers observed_R = wcen / dw

And I get 200.5, not 200.

So then, I tried to calculate the bins by hand, knowing the starting wavelength and R (knowing ending wavelength does not help much at the start here, but it lets you know when to stop the bins). However, I got stuck after the first bin. This is because, given that every bin is defined as wcen / dw = R, and I know the edge of the first bin, for the second bin, I know neither the wcen nor its dw in advance, and it seems to me that I have to solve for two unknown at once. Am I going crazy here?

Then I looked around in specutils and only found spectral_resolution() that calculates the R given the wavelengths, but not the other way around. This makes me think that maybe requiring R to be constant throughout is not practical in the first place?

🤔

pllim · 2019-07-23T20:42:49Z

synspec.py

+    return Observation(source_spectrum, bandpass, binset=binset, force=force)
+
+
+def from_wave_array(source_spectrum, bandpass, waveset, force='none'):


I don't see from_wave_array being used in the notebook, is this function necessary?

tcjansen and others added 25 commits June 4, 2019 16:12

Initial commit - I'll give future branches more descriptive names

eec1f6c

Added comparison plot, more explicit variable names

b597924

Query sdss filters from SVO, get rid of files

fad52b0

Added Gaia query for hat-p-11 parameters

01619ef

Cite Gillon 2017 for trappist-1 parameters

b6a355e

Added preamble with links to parts of the tutorial

397d3a2

Use pmv_kpno to simulate the atmospheric transmission

c8fc9d5

Uncommented some things (oops)

dfc48f2

Added Kepler (ie space-based) example observation of the same stars

e1cd58a

Adjust 1-1 plot size for visability

e10eb67

typo fix

6d163ed

Co-Authored-By: P. L. Lim <2090236+pllim@users.noreply.github.com>

changed import style

c8fb9b4

Co-Authored-By: P. L. Lim <2090236+pllim@users.noreply.github.com>

Addressed pllim's comments: fixed typos, changed variable names, took…

463965a

… out binset and forced tapering in Observation calls

push changes

f015cb9

small changes made for consistancy

beed31b

Added a query to SkyCalc for modeling the atmospheric transmission (n…

c700ee9

…ot finished)

not sure what to do with skycalc results?

a8049ed

Skycalc query fully implemented

6021ffa

Changed kepler example from count rates to counts

255a413

Error when loading skycalc data on a different machine

186e909

Skycalc error resolved - had to do with strict network firewall

69d4570

Moved the skycalc querying to its own script, atmospheric_transmittan…

0cfbb40

…ce.py

Implemented tynt for querying filter transmittance

805c5e9

Renamed atmospheric_transmittance.py to skymodel.py and changed the f…

6cd0d13

…unction name

Initial commit

4fae02e

pllim reviewed Jul 15, 2019

View reviewed changes

tcjansen added 2 commits July 15, 2019 17:19

Changed bp model from Box1d to Const1D, plotting binflux in steps

5aadf3c

Muted warnings, set Observation(force='taper')

c331ff0

tcjansen added 5 commits July 15, 2019 18:06

Zoomed in on spectral features to show binning

6497ffe

Minor changes

5c2e3c8

Added snr examples

1685ea6

Added snr examples

c321987

Wrote first draft of synspec, implemented it into the tutorial

3a501c3

bmorris3 reviewed Jul 23, 2019

View reviewed changes

Minor changes

53c2dab

pllim reviewed Jul 23, 2019

View reviewed changes

		@@ -1 +0,0 @@
		Browse the branches to see the different notebooks I've been working on! No newline at end of file

		import numpy as np


		def from_R(source_spectrum, bandpass, R, waverange, force='none'):

		return Observation(source_spectrum, bandpass, binset=binset, force=force)


		def from_wave_array(source_spectrum, bandpass, waveset, force='none'):

Conversation

tcjansen commented Jul 15, 2019

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tcjansen commented Jul 15, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pllim left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

tcjansen commented Jul 15, 2019 •

edited

Loading