Final changes for glvis-4.3

[tzanio]

📆 Target release date: July 15, 2024

🟡 In Progress:

• Support for integral finite elements #280
• Visualization of quadrature functions #286

🔴 TODO:

• Command-line option to ensure minimal level of refinement to large meshes don't show as linear by default -- see Higher auto-refinement limits #285 ← @tzanio
• Support for visualization of particle data?
• Update CHANGELOG ← @tzanio

🟤 Optional:

• Updating GLVis mesh vector or scalar solution without drawing a new mesh #272
• Update CI to test installation with both GNU make and CMake #241
• Additional color palettes? #268
• Change colorbar number format #260
• Future Improvements 🚀 #133

⚪ After the release:

• Update website, https://github.com/GLVis/web
• Update glvis-js and glvis.org/live
• Add CI testing with GitHub Actions glvis-js#7
• Update pyglvis

🟢 DONE:

• Update version number from 4.2 to 4.3
• Vector field options in 2D with single colored arrows above #278 ← @v-dobrev will review
• Switch from <key>A</key> to A in README.md? See this issue

⚫ Postponed:

• Calling VisualizationSceneSolution3d from own application #274
• adaptive geometry refiner
• raytracing?!
• better vector graphics output
• improved testing with random keys
• Configurable line width with modern OpenGL #232
• Refactoring to remove global state #170
• [WIP] Improved transparency with dual depth peeling #166

[acfisher]

LGTM

[najlkin]

What is the idea with quadrature functions? Interpreting them as L2 elements would be very simple 😉 . I have done that in the past in my codes.

[tzanio]

What is the idea with quadrature functions? Interpreting them as L2 elements would be very simple 😉 . I have done that in the past in my codes.

This has been discussed multiple times, but there are essentially 3 options we have considered:

1. Plot p.w. const function on the LOR Gauss-Lobatto mesh, where each LOR element contains exactly one Gauss-Legendre quadrature point. (The p points of 1D GLL quadrature rule are interleaved by the p+1 points of the 1D GL rule)

2. Interpret quadrature values as the dofs of an L2 grid funcion

3. Draw a little sphere for each point

Option 3 is most general and will be required for "Support for visualization of particle data". I personally like option 1, because it does not introduce variations beyond the original data. That works only on quad/hex meshes though.

[najlkin]

I see, we must reach consensus then what way to go, which might take longer than actually implementing it 😁 .

[tzanio]

We can also support both options 😁

[najlkin]

I found out there is no smooth vizualization in 2D for vector fields atm. This could be also added, but it would require to differentiate the vector-to-scalar function 🤔