Open improbable-22 opened 7 years ago
I don't think this feature is exposed by the C library in the same way it is in Mathematica. A possibility is to use verbosity, this prints to screen many additional information including the list of subregions (but this is lee handy than what you're asking for). For example:
julia> cuhre((x, f) -> f[1] = cos(x[1]), flags = 3)
Cuhre input parameters:
ndim 2
ncomp 1
nvec 1
epsrel 0.0001
epsabs 1e-12
flags 3
mineval 0
maxeval 1000000
key 0
statefile ""
Region (0.000000) - (1.000000)
(0.000000) - (1.000000)
[1] 0.841471 +- 2.22045e-16
Iteration 1: 65 integrand evaluations so far
[1] 0.841471 +- 2.22045e-16 chisq 0 (0 df)
Region (0.000000) - (0.500000)
(0.000000) - (1.000000)
[1] 0.479426 +- 1.56691e-12
Region (0.500000) - (1.000000)
(0.000000) - (1.000000)
[1] 0.362045 +- 1.1831e-12
Iteration 2: 195 integrand evaluations so far
[1] 0.841471 +- 2.22045e-16 chisq 1.86265e-09 (1 df)
Component:
1: 0.8414709848078966 ± 2.2204460420128823e-16 (prob.: 3.443539937576958e-5)
Integrand evaluations: 195
Fail: 0
Number of subregions: 2
Thanks, I would not have thought of that.
Will experiment a bit...
In addition, Cuba library comes with a tool called partview
which allows visualizing the tessellation. You can run a Julia script with verbosity = 3 from the command line and pipe the output to partview
.
In the Mathematica interface to Cuba, I can obtain a list of the regions used by option Regions -> True. Is there any way to get this information in Julia?
After skimming the web page, it's unclear to me whether the C++ library exports this data, so perhaps this cannot be done.