Error in 3D prism volume computation?

[arbenede]

I am trying to compute the volume of a simple 3D prism using vol (the .ine file is below). I know that the volume of the prism is approximately equal to 0.0118 (area of the polygonal base X height=1). I have run vol from the latest development branch with different values of the error tolerance:

$ ./vol --file1 prism.ine -e 0.1
Reading input from file...
Experiment 1 
3 8 1 -1 0.1 [-0.9,-1.1] 131833 10 0.02023386 [0.02023386,0.02023386] 0 1.020234 0 0.419022 0 
$ ./vol --file1 prism.ine -e 0.2
Reading input from file...
Experiment 1 
3 8 1 -1 0.2 [-0.8,-1.2] 32958 10 0.02025999 [0.02025999,0.02025999] 0 1.02026 0 0.129329 0 
$ ./vol --file1 prism.ine -e 0.01
Reading input from file...
Experiment 1 
3 8 1 -1 0.01 [-0.99,-1.01] 13183347 10 0.02038337 [0.02038337,0.02038337] 0 1.020383 0 44.884 0 

and from these runs it seems that the volume computed by vol might be off by a factor 2. Or is vol computing something different, like a signed volume?

H-representation
begin
8 4 real
0.008509755085097572 0.10187124277134102 0.08149699421707308 -0.0
0.008509755085097572 -0.10187124277134102 -0.08149699421707308 -0.0
0.013491075134910584 -0.05780634428435466 0.15375492457251516 -0.0
0.013491075134910584 0.05780634428435466 -0.15375492457251516 -0.0
0.015359070153590594 -0.15967758705569568 0.07225793035544206 -0.0
0.015359070153590596 0.15967758705569568 -0.07225793035544209 -0.0
0.5 -0.0 -0.0 -1.0
0.5 -0.0 -0.0 1.0
end

[arbenede]

I made a few more tests:

1. I run the same problem with the MATLAB code by Cousins and Vempala, obtaining a very similar value for the volume (0.0204) (???)
2. I ran vol only on the base of the prism, getting as expected the same value (height=1)
3. I ran vol in sampling mode on the base of the prism and verified that it is sampling correctly (at least in this 2D subproblem)
4. I have re-scaled all the coordinates (this is easy to do since the base of the prism is the Voronoi cell of a lattice centered at the origin and the prism extends symmetrically in the Z dimension). The estimated volumes scales expected but the result is still off by a factor of 2.

It looks like sampling is working, so maybe there is some problem with the rounding/sandwiching steps of the algorithms?

[vissarion]

You probably computed the exact volume incorrectly or you have some typo in you input file. Using vinci with your ine file I computed 2.037424855427e-02 with is pretty close to the numbers you computed above with our package. So if you agree I am closing the issue.