Updated SAAQ version correction and new research note directory.

Research notes/Graph_explaination.md

@@ -0,0 +1 @@
+My Y variable "walker" goal is find the best routing pathway.  Best to think as is it looking for the best physical route with least resistance!  In way acts as my pulse of electrical energy (a spike).

Research notes/Physics_SNN.md

@@ -0,0 +1,5 @@
+Since I am building an biological system, the walker energy cannot just flow everywhere at once.   The goal is have the network to test different paths.
+
+During Routing Collaspe (the bad graph), all the voltage energy was so high that a single walker (around ID 620) got permanently blasted with energy and became an blackhole.
+
+In the new Attractor Discovery(good graph), the L2 normalized the voltage, allows the walker to naturally explore the network.  Walker is physically looking for the path with least resistance.  Energy settled into comfortable state, through Walker 2000, with secondary echoes in Walkers 700 and 1450.

Research notes/notes.md

@@ -0,0 +1,2 @@
+Massive win: after running two seperate tests the graph don't look exact same.   First real failed due to misconfiguration in cuda.  The Cuda kernels were searching for F32 instead of F16.   Second test was successful by running prompt "Teaching OMLoE the language of SNN".   Third test I changed the vectors to represent rust syntax prompt "fn main () { println!(); }.   This is where the true victory of this project came so far because the L2 Normalization just forced the hardware to adopt.   Had it been the same as first result it would indicate that L2 Normalization created an lazy resting state, regardless of what I fed it.  It would have been an failure.
+

{SAAQ 2.0 → SAAQ 1.0}/README.md

@@ -1,4 +1,4 @@
-# Sparse Activity-Aware Quantization (SAAQ) 2.0
+# Sparse Activity-Aware Quantization (SAAQ) 1.0
 
 ## Foundational SAAQ Equation