On Tue, 28 May 2013, apoorv wrote: > On May 27, 10:37 am, Graham Cooper <grahamcoop...@gmail.com> wrote: > > On May 27, 2:32 pm, William Elliot <ma...@panix.com> wrote:
> > > > > > Is there some study on situation where the output from a TM at time T is > > > > > > fed back so that the total input at time T+1 is some function of output > > > > > > at time T and external input at time T+1.? > > > > > > > So imput is on the right and out on the left and when done at time > > > > > t, move the output to the right of the tape and start again. > > > > > > > Time t+1 begins the new output from the old output. You get to wait > > > > > until time t2 for the full second output. > > > Eliot's post ,which appears to be deleted, showed that a single TM , could > implement feedback. However, if we put the added Requirement that the input > at time t is processed with delay not exceeding A fixed time t0, then two TM > s appear necessary . -Apoorv
If t0 - t is larger than processing time, then you wouldn't.
However for a second machine to start at time t+1 or some time t0 when t0 - t is less than processing time, a second machine can't wait until the output is completed, it has to be processing the output as is it' coming out.