nearFull, nearEmpty

README.md

@@ -44,14 +44,21 @@ La fullness di un Tub puo' essere cambiata in DEV. Dal punto di vista della comp
 
 L'assegnazione dei livelli di fullness, active e emptiness viene fatta una volta completata l'espansione come per RF. Infatti l'assegnazione deve essere collegata a RF.
 
-#### full, medium and empty
+#### full and empty
 
-Il controllo della condition in un context o episode viene fatto con 3 "stati".
+Il controllo di condition in un context o hpothesis in un episode viene fatto con:
 
 - full: se gli actual sono maggiori o uguali a fullness
-- medium: se gli actual sono minori di fullness e maggiori di emptiness
 - empty: se gli actual sono uguali o meno di emptiness
 
+#### nearFull, medium and nearEmpty
+
+Il controllo di condition in un context o hpothesis in un episode viene fatto con:
+
+- nearFull: se gli actual sono almeno 70% di fullness
+- nearEmpty: se gli actual sono almeno 30% di fullness
+- medium: se gli actual sono fra 30% e 70% di fullness
+
 ### RF
 
 Teorema di Shannon inverso. E' come se facessimo un campionamento su un ipotesi di continuita'.

neuron/BEH-AXO.md

@@ -415,8 +415,8 @@ context: SeContext
   in_context: Fixed
   rf: ( active: 600x )
 
-  condition: NOT (Rrp empty) 
+  condition: NOT (rp empty) 
-    out_context: RRPNotEmpty
+    out_context: RPNotEmpty
 
   condition: NOT (CaTrace empty) 
     out_context: CaTracesNotEmpty
@@ -461,21 +461,65 @@ episode: CaTracesClearance
 
 #### RP->RRP shuttling
 
-How RP is moved to RRP
+This happens in the seconds loop, once per second.
-This happens in the seconds loop, once per second. The recruitment step reads Tr_Ca and moves a fraction of the available RP into the RRP:
 
-current_recruitment_rate = map_trace_to_speed(Tr_Ca)   # /s
+![nt-release.png](.attachments/RRP-RP-shuttle.png)
-refill_amount = current_recruitment_rate *dt_slow_s* N_RP * (Max_RRP - N_RRP) / Max_RRP
-refill_amount = max(0.0, refill_amount)
-refill_amount = min(refill_amount, N_RP)
 
-N_RRP = min(N_RRP + refill_amount, Max_RRP)
+##### RP->RRPSlow: Episode
-N_RP  = max(0.0, N_RP - refill_amount)
+
-Three factors multiply together to determine how many vesicles move:
+Recruitment machinery at base speed:
-current_recruitment_rate — set by Tr_Ca history. High recent activity means faster recruitment. This is the only place Tr_Ca has any effect in the model.
+
-N_RP — how many vesicles are available in the reserve. As N_RP depletes, fewer vesicles are available to move regardless of how fast the rate is. This is the depletion bottleneck — a synapse that has been firing for minutes may have a high Tr_Ca driving fast recruitment, but if N_RP is nearly empty there is nothing left to recruit.
+- Synapse quiet: CaTrace empty
-(Max_RRP - N_RRP) / Max_RRP — the headroom in the RRP, normalised. When the RRP is nearly full, recruitment slows automatically because there is little room to fill. When the RRP is empty after a burst, headroom is maximal and recruitment runs at full speed. This prevents overfilling and makes the system self-regulating — recruitment is fastest precisely when it is most needed.
+- Supply available: RP medium
-The two guard clauses ensure the arithmetic stays physical: refill_amount cannot be negative, and cannot exceed what N_RP actually contains.
+- Headroom available: RRP medium
+
+```Gen
+episode: RP->RRPSlow
+  contained_by: BEH-PRE
+
+  in_context: RPNotEmpty
+  rf: ( active: 24x ) # Slow
+
+  hypothesis: (RRP medium) AND (RP Medium) AND (CaTrace empty)
+    action: [RP decrease, RRP increase]
+    trace: None
+```
+
+##### RP->RRPFast: Episode
+
+Recruitment high:
+
+- Recent activity has primed the machinery: CaTrace medium
+- Supply Available: RP medium or full
+- Headroom available: RRP medium or
+
+
+```Gen
+episode: RP->RRPFast
+  contained_by: BEH-PRE
+
+  in_context: RPNotEmpty
+  rf: ( active: 24x ) # Fast
+
+  hypothesis: 
+    action: [RP decrease, RRP increase]
+    trace: None
+```
+
+##### RP->RRPMaximal: Episode
+
+...
+
+```Gen
+episode: RP->RRPMaximal
+  contained_by: BEH-PRE
+
+  in_context: RPNotEmpty
+  rf: ( active: 24x ) # Maximal
+
+  hypothesis: 
+    action: [RP decrease, RRP increase]
+    trace: None
 
 ### min: behaviors