64 bit versions give incorrect results about view3d

[DrCrlsn]

Pinney and Bean report incorrect values for obstructed view factors F16 and F61. I calculated F16 by modifying an expression from Hamilton and Morgan (1952) for an unobstructed line source and plane ("Configuration L-2"), and my result agrees with View3D (see attached file PB.pdf). With this correction, some other discrepancies (F17 and F18) may be recalculated and, also, agree with View3D.

I built View3D 4.0 on ubuntu 18.04 LTS 64-bit.

[DrCrlsn]

Other discrepancies remain in Rows 7 and 8 between Pinney and Bean and View3D. I haven't yet done the independent calculation.

See excerpt of View3d.log below; entire file attached.

View factors after combining surfaces:
          #        name   SUMj Fij (encl err)
 Row:       1, Name: srf-1, sum of view factors:  0.999976 (0.000024)
.000000 .113154 .378093 .027434 .032957 .182357 .132990 .132990 
 Row:       2, Name: srf-2, sum of view factors:  1.000127 (0.000127)
.339463 .000000 .318996 .000000 .000000 .098872 .121398 .121398 
 Row:       3, Name: srf-3, sum of view factors:  0.999942 (0.000058)
.567140 .159498 .000000 .000000 .000000 .041152 .116076 .116076 
 Row:       4, Name: srf-4, sum of view factors:  0.999942 (0.000058)
.041152 .000000 .000000 .000000 .159498 .567140 .116076 .116076 
 Row:       5, Name: srf-5, sum of view factors:  1.000127 (0.000127)
.098872 .000000 .000000 .318996 .000000 .339463 .121398 .121398 
 Row:       6, Name: srf-6, sum of view factors:  0.999976 (0.000024)
.182357 .032957 .027434 .378093 .113154 .000000 .132990 .132990 
 Row:       7, Name: srf-7, sum of view factors:  0.999824 (0.000176)
.239382 .072839 .139291 .139291 .072839 .239382 .000000 .096800 
 Row:       8, Name: srf-8, sum of view factors:  0.999824 (0.000176)
.239382 .072839 .139291 .139291 .072839 .239382 .096800 .000000 
Summary:
Max row sumF error:  1.76e-04
RMS row sumF error:  1.13e-04

View3D.log

[jasondegraw]

@GlennEng Thanks for the information! I've edited your last comment so that information is all text, but that's all very helpful. I recall correctly (it's been a few years), this issue was that with GCC on Windows I was getting different answers from 32 bit and 64 bit executables. That's something I never saw on any other system, and is probably different than any discrepancies you're seeing.

With that said, differences between the program and a published source is an issue, so I'd very much welcome additional comments here or on an entirely new issue. Would you consider contributing your calculations/writeup to the manual?

[DrCrlsn]

I guess I should have created a new issue instead of piggybacking on yours.

Yes, feel free to use my calc/writeup in the View3D manual.