sivia.tar holds little toy nested-sampling programs
in 'C', as printed in the update of Devinder's book "Data
Analysis: a Bayesian Tutorial" (2nd edition) OUP 2006.

It should be fairly obvious.  The main program is "mininest.c",
and it requires a file "apply.c" to be #included.

Copy any of the
four application programs lighthouse.c (Steve's famous problem),
crystal.c (my toy order/disorder problem), onoff.c (a binary
switching problem), or quantify.c (binary switching with amplitudes)
to apply.c, compile mininest, and run.

Two of the applications also
have the service procedures "tree.c" #included.  Hence the six .c files.

-------------------------------------------------------------------

Alternatively, using Makefiles added by David MacKay

   "make directories; make all"

should do the trick.


Then you can run the following:

> ./lighthouse 
# iterates = 1000
Evidence: ln(Z) = -160.293 +- 0.163994
Information: H = 2.68939 nats = 3.87997 bits
mean(x) = 1.24208, stddev(x) = 0.182403
mean(y) = 1.00426, stddev(y) = 0.194502

> ./crystal 
# iterates = 800
Evidence: ln(Z) = 280.366 +- 26.7871
Information: H = 717.548 nats = 1035.2 bits
  -1.00   1.9980
  -2.00   2.0040
  -3.00   2.0160
  -4.00   2.0380
  -5.00   2.2440
  -6.00   2.3260
   . . . . . .
-792.00 999.0000
-793.00 999.0000
-794.00 999.0000
-795.00 999.0000
-796.00 999.0000
-797.00 999.0000
-798.00 999.0000
-799.00 999.0000
-800.00 999.0000

> ./onoff 
# iterates = 1000
Evidence: ln(Z) = -3.32457 +- 0.0782521
Information: H = 0.612339 nats = 0.883419 bits
0: Posterior(ON)=  6%
1: Posterior(ON)= 25%
2: Posterior(ON)= 14%
3: Posterior(ON)= 63%
4: Posterior(ON)= 14%

./quantify 
# iterates = 1000
Evidence: ln(Z) = -9.63624 +- 0.276217
Information: H = 7.62957 nats = 11.0071 bits
0: Posterior(ON) = 99%, q =  2.20 +-  0.43
1: Posterior(ON) = 17%, q =  0.63 +-  0.71
2: Posterior(ON) = 99%, q =  1.76 +-  0.37
3: Posterior(ON) = 11%, q =  0.54 +-  0.49
4: Posterior(ON) =  4%, q =  0.14 +-  0.15