Slide adjustment

[story645]

feedback:

1. Motivate ahead of time that this is about the architecture of the library, looking at matplotlib as a middle layer
   1. atoms for API/library building
   2. maybe a layers of librarys figure> backends->mpl-> seaborn/deeplabcut/mne
   3. little bits circuit, pipes game
      3. replace the source and sink w/ data and graphic data:
         • [x] add another data point
         • [x] change to something in a different latitude (rather than JFK or LGA)

topology:

• [x] add a shuffle data example in between for topology
• [x] replace the little arrows w/ one little arrow going back to a phantom row w/ question marks

layout

• [ ] make first very sketchy
• [ ] make graphic sketch, but the data table not
• [ ] make grapihic and data not sketchy with arrows

motivation for functionlism

linked-chart -

• [x] maybe flip the direction of the arrows
• [x] maybe arrow to legend
• [ ] note, if I change one bar, how does that back propogate inclusion
• [x] maybe condense to one and add inclusion graph as part of topology motivation - two fiber bundles
• [ ] maybe add a set of columns arrows - maybe tear apart rows and columns as lead in to equivariance

big arrow slide

1. maybe frame it as how do we decompose this arrow? what is a good way to decompose it?

equivariance:

1. maybe kelvin to celsius instead

math:

• [x] first slide all data + all graphic, then isolate for the rest
• [x] add sections & labels immediatly, discuss as fiber bundles
• [x] fix location of arrow
• [x] maybe move S arrows in
• [ ] replace V* sub axes w/ rectangles
• [ ] make many and put in faded out big dot
• [ ] distribute arrows all over S
• [ ] move tau arrow in E
• [ ] Q^hat_{point} # make clear that it assembles into the glyph, is specification of artist
• [ ] move V* slide to appendix
• [ ] rho is the what color at this pixel function?, Q manufactures the rho
• [ ] maybe lable functions w the parameters/arguments

narrative

• [ ] back out into why topology, equivaraince, then into what, then how
• [ ] Topology eg. Butler's vision with some modification (framing for related works section on rewrite)
  • [ ] maybe add complex nested topology, explain what topology means in those contexts (table, time series, (lat,lon))
• [ ] equivarance (many people with modification) (framing for related works section on rewrite)
  • [ ] add a slide on fiber
• [ ] we tie in with the visualization space
• [ ] flesh how the functional decomposition makes this work \nu, Q, \tau, \rho -> make this bit clearer - maybe add a slide on arrows-> functional paradigm -> identify the
• [ ] the meta-graph [not sure if that is the right term] of inclusions of fiber-bundles which allows us to encode what is shared and what is not.

appendix slides

• [ ] walking through the math formally

[story645]

Frame in terms of ownership - first slide is some viz, second slide is arrows forward to data, third is arrows back from data -> this should be an utter mess of arrows on purpose: talk about/annotate/title is how mpl currently owns this mess and also combines these arrows in weird ways, new API is about disentangling what benefits do devss get? 1) the own part of the data backward pipeline 2) by disentangling the visualization steps, easier for devs to drop on their own 3) shared colormaps/norms/aesthetic mappers - mpl doesn't really allow sharing em (find the issue #s?)

[story645]

function tree versus UML spaghetti flow chart transition into math?

[story645]

• Not really/consistently is a throwback to slides from last year's talk