Johannes Terblanche
Johannes Terblanche
A rough benchmark of where we are. ```julia fg = initfg() fg = generateGraph_Beehive!(1000; dfg=fg, graphinit=false, locality=0.5); r=IIF.solveGraphParametric(fg; autodiff=:finite, computeCovariance=false); # Seconds run: 164 (vs limit 100) # Iterations: 1...
We still need the down solve only case.
PR #1002 provides the basic structure and is the start of downsolve only.
We need a better solution to keep the examples up to date. Ideas: 1. We can test it as a separate CI action to at least know when it's broken....
We should reevaluate this issue to fit with #1371.
I had 2 very distinct problems: - The way the lambdas are currently built only influences user (or Zygote) provided gradients. It will just have to be changed to support...
xref https://github.com/JuliaRobotics/IncrementalInference.jl/issues/1546
Also see package: https://github.com/SciML/PreallocationTools.jl
Going to defer this to later as the original purpose of solveGraphParametric was to solve a clique subgraph, but now it has to support large batch solving also. The return...
Yes, we spoke about it and the results looked correct. The test might need updating.