home *** CD-ROM | disk | FTP | other *** search
- csv2.1 ! Control Strategy Version 2.1
- !************************************************************************
- !* *
- !* Traveling Salesman Problem Control Strategy *
- !* *
- !************************************************************************
- !
- ! This control strategy assumes that the first layer above the input
- ! layer is the "hopfield" layer. The layers beyond it are used to
- ! generate inhibitions. The output layer is ONLY activated when needed.
-
- L___Risa trace aux3
- L___Ri_a cset recall,1
- L___Ri_a lset in,0
- L___Ri_a math fire|sum|rnoise|tran|output !input layer
- L___Ri_a lset in,1
- L___Ri_a math fire|set|tran|output !hopfield layer
- L___R__a br @InhLp !process forward
- !
- ! We are now positioned to the hopfield layer. Process this through
- ! the output layer -1 repeatedly.
- !
- L___R_sa @Loop lset in,1 !hopfield layer
- L___R_sa cadd recall,1
- L___R_sa math sum|rnoise|tran
- L___R_sa math output
- L___R_sa io wrstep
- L___R_sa fcmp lflag,36
- L___R_sa banz @Done !at end of LRS
- L___R_sa @InhLp lset cur,1 !compute inhibitions
- L___R_sa math fire|sum|tran|output !full inhibition computation
- L___R_sa lcmp out,-1 !see if output layer next
- L___R_sa blt @InhLp !finish computing inhibition
- L___R_s_ eos !check for end of step
- L___R_sa br @Loop !compute next hopfield iter
- L___R_sa @Done lset out,0
- L___R_sa math fire|sum|tran|output
- L___R_sa io write
- LisaRisa trace 0
-