The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design Languages >> Verilog-AMS >> How to chose seed?
https://designers-guide.org/forum/YaBB.pl?num=1163658055

Message started by neoflash on Nov 15^th, 2006, 10:20pm

Title: How to chose seed?
Post by neoflash on Nov 15^th, 2006, 10:20pm

How we choose the value of seed, of random number generator and how it affects?
Does this number play any role, besides specify a determistic start point? What if I chose seed=111?

Title: Re: How to chose seed?
Post by neoflash on Nov 15^th, 2006, 10:21pm

attach the sample usage here: dT = jitter*$rdist_normal(seed,0,1);

Title: Re: How to chose seed?
Post by Geoffrey_Coram on Nov 16^th, 2006, 11:24am

As far as I know, it's just a start point in the sequence of pseudo-random numbers.

Title: Re: How to chose seed?
Post by neoflash on Nov 16^th, 2006, 5:03pm

I have attached one sample code from this website, why author chose seed=286, not other number?

module osc2 (out);

output out; voltage out; // output signal
parameter real freq=1 from (0:inf); // output frequency
parameter real vl=-1; // high output voltage
parameter real vh=1; // low output voltage
parameter real tt=0.01/freq from (0:inf); // transition time of output
parameter real jitter=0 from [0:0.1/freq); // white period jitter
integer n, seed;
real next, dT;

analog begin
@(initial_step) begin
seed = 286;
next = 0.5/freq + $abstime;
end
@(timer(next)) begin
n = !n;
dT = jitter*$rdist_normal(seed,0,1);
next = next + 0.5/freq + `M_SQRT1_2*dT;
end
V(out) <+ transition(n ? vh : vl, 0, tt);
end
endmodule

Title: Re: How to chose seed?
Post by Ken Kundert on Nov 16^th, 2006, 7:20pm

It was a completely arbitrary choice. The only criteria I used was to use a different seed for each rdist function present in the system. If you give the same seed to two different rdist functions they will produce an identical sequence of random numbers.

-Ken

Title: Re: How to chose seed?
Post by Geoffrey_Coram on Nov 17^th, 2006, 5:54am

Ken Kundert wrote on Nov 16^th, 2006, 7:20pm:

By "two different rdist functions" Ken does not mean to say that $rdist_normal() and $rdist_uniform will give the same sequence of random numbers. I am fairly sure he meant "give the same seed to two different calls of the same rdist function":

Modifying the example:

integer n, seed, seed2;
real next, dT, dT2;

analog begin
@(initial_step) begin
seed = 286;
seed2 = 286;
next = 0.5/freq + $abstime;
end
@(timer(next)) begin
n = !n;
dT = jitter*$rdist_normal(seed,0,1);
dT2 = jitter*$rdist_normal(seed2,0,1); // will have the same sequence of values as dT
next = next + 0.5/freq + `M_SQRT1_2*dT;
end
V(out) <+ transition(n ? vh : vl, 0, tt);
end

dT2 will have the exact same sequence of random values as dT. Also, seed and seed2 are changed by the $rdist function calls, and the sequence of their values will be the same.

I wonder: if you did
rn1 = $rdist_normal(seed,0,1);
rn2 = $rdist_uniform(seed2,-1,1);

will "seed" and "seed2" have the same sequence of (integer) values? I guess you might or might not, since the LRM doesn't say.

-Geoffrey