[x] Run-on sentences, needs a bit of wordsmithing and a bit more lead in.
[x] not sure about including figures from other papers directly.
[x] open with something about how visualization is about representing data on the page, the describe state-of-art, then Du Bois, then our model will explain / describe them all
[x] when introducing new cites lead with idea not with name
2.1
[x] when introducing new cites lead with idea not with name - is painfully clunky here
[x] still not sure about copy-pasting other peoples figures, but do not think figure 4 adds anything over figure 3.
[x] Flip discussion of keys order to be your way of understanding it and then contrast it to the way Munzer does it (and why it is insufficient)
[x] put fiber bundles at the top and then argue everything else is a special case?
2.2
[x] This is jumbled and needs to be expanded into two or three paragraphs.
[x] VTK very much explicitly carries structure in their data model (as shown by the image from the mayavi docs you have above).
2.3
[x] break list out into a list?
[x] does this subsection need to exist? - every how to write guide says yes
3.
[x] not sure this figure is helping at all
[x] start with incomplete sentence
[x] "is generally" -> "can be understood in general as"
[x] too much jargon in the opening paragraph, explain what each layer means in laymen terms.
[x] start with A: E->H earlier, then talk about how it will be decomposed?
3.1
[x] fiber bundles are good for all data, not just visualization data
[x] was this meant to be a definition list?
[x] the indentation / paragraphs are wonky, not sure which are real and which are formatting errors.
[x] still no discussion of why we should be bothering with this structure at all
3.1.1
[ x] "to formally describe naming the components of the fiber we use Spivac's ... ". Start with what we want, follow with who's work we are using.
[x] still no discussion of why we should be bothering with this structure at all
[x] timestamps are not R+ (the world existed before Jan 1 1970)
[x] Temperature is a restricted set of R (for K it in R+)
[x] U vs F between (7) and (8)
3.1.2
[x] Flip order "to do we use <tool/formalism/idea>"
[x] Again, what do monoids / measurement scales get us? What part of the process can we talk about now that we could not before?
3.1.3
[x] Not sure what the opening sentence means.
[x] the connectivity / continuity of the data is much deeper than the ability to segment for parallelization.
[x] again, why do we care? What can we talk about that we could not talk about without this concept?
3.1.4
[x] Why do we care?
[x] use this language to describe:
[x] numpy array
[x] xarray array
[x] xarray dataset
[x] pandas series
[x] pandas dataframe
3.1.5
[x] this one comes closest to saying why we care!
3.2
3.2.1
[ ] being picky, not sure what negative color is. Maybe limit all of these to the interval? - was common notation last time O looked
3.2.2
[x] indentation looks off
[x] L196 "associated point k"?
[x] I do not understand the last paragraph
3.2.3
[ x] what about vector formats?
3.3
[x] put word 'section' in to xrefs?
[x] sentence on L219 - L221 does not make sense, what is "all of the data" in this context?
3.3.1
[ x] I'm not sure that referencing the names different visualization libraries give to the visual parameters actually makes it clearer.
3.3.2
[x] this is the best written section!
[x] make reference to the monoid section above
[x] maybe expand sentences to full paragraphs?
3.3.3
[ x] I'm still not convinced that the concept of 'glyph' or 'mark' is useful because it gets into hair-splitting / word games.
[ x] "To define a mark, using the language of Munzer..."
[ x] this part of the model is reflected in the literature as glyphs [cite]
[ x] in special cases Q constructs the glyph
3.3.4
[x] if you have RGB that this is an image not a heatmap, I would replace every reference to "heatmap" with "image".
3.3.5
[x] I could follow this because I knew what it was meant to say, needs to be expanded.
[x] "\hat(Q) allows us to map the mathematical model more directly to Matplotlib..."
3.3.6
[x] not sure with K2 \inculde K1 is out of scope?
3.3.7
[x] not clear
3.4
[ x] concern about re-using \alpha, \beta from S ?
[ ] L323-LL326 This needs to be un-packed (is unclear what it is)
[x] why are you telling us about Tori? why are you telling us about mobius strips?
4
[x] \hat{Q}?
[x] need better notation that \nu=transforms etc to associate the code with the math (= is too overloaded already).
[x] "The equivalence classes A' naturally map to Python classes, parameterized by..."
4.1
[x] need to pass something to self.data.view() to pick the right correct sheaf (this is critical!)
[x] L368 -> making excuses for the way Matplotlib currently works, something like "obvious optimizations are being ignored to clearly show the structure".
[x] a post-proposal task is to implement both Line and Point without inheriting from an existing Artist
[x] do not be forgiving on view.get()
[x] L377 paragraph not clear. Make sentences shorter and expand to two paragraphs
[x] "the only difference" -> "the critical difference"
4.2
[x] if you are going to show the None for the key here, need to account for itin the Artist examples
[ ] maybe do this before Artist?
4.3
[ ] maybe do this first?
[x] list to describe data is missing words
[x] leave the notion of base / fiber schema
[ ] The fiber should be instance, not class, level state
[ ] show wrapping dict of numpy arrays
[x] and a dataframe
[x] not sure we need tau to be a method?
[ ] show downselection to a sheaf (zooming - passing axes view limits back and getting more data)
[ ] wrap analytic functions?
4.4
[x] include the fiber meta-data in the case study
[x] the signature of view needs to be consistent with section 4.3.
[x] do not do kwargs.pop. It is a bad idea in general and only adds complexity to this example.
[x] do normalization in __init__ not in _make_bars - why?
[x] do not do deeply-nested fallback behavior. It is both bad design and makes this more confusing than it has to be.
[x] also do not mutate input
[x] and dict.setdefault exists if you must
5.
[x] when describing what you are doing start from one end and systematically walk to other.
5.1
[x] effectiveness is fully out of scope for this model. While you cam separately develop logic around the encoders, this math does not care
2.1
2.2
2.3
3.
3.1
3.1.1
3.1.2
3.1.3
3.1.4
3.1.5
3.2
3.2.1
3.2.2
3.2.3
3.3
3.3.1
3.3.2
3.3.3
3.3.4
3.3.5
3.3.6
3.3.7
3.4
4
4.1
self.data.view()
to pick the right correct sheaf (this is critical!)view.get()
4.2
None
for the key here, need to account for itin the Artist examples4.3
tau
to be a method?4.4
view
needs to be consistent with section 4.3.__init__
not in_make_bars
- why?dict.setdefault
exists if you must5.
5.1
5.2
6.