- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: <panov_at_dec1.sinp.msu.ru>

Date: Fri, 03 Mar 2017 23:43:27 +0300

Dear Luigi,

Thank you very much for your help. I understand what was my main error.

In fact I should expand the expression with parentheses by hands and

write it out explicitly without parentheses at all using the boolean

algebra. The usage of the same half-spaces to define many different

regions also is very good idea. Unfortunately I can't use meaningful

names for bodies and regions since I program very complicated geometry

with many thousands regions and I don't write input file for it by hand.

I write special program that produces the input file with all this

variety of things. Therefore I have to generate all names automatically.

Thank you again,

Alexander

On 2017-03-03 12:46, Luigi Salvatore Esposito wrote:

*> Dear Alexander,
*

*> your geometry definition could be improved a lot.
*

*> The problem is the subtraction of zones defined by parenthesis in the vacuum definition.
*

*> Parentheses require expansions and these implies a diverging number of terms and therefore crashesâ€¦
*

*>
*

*> In general, one can always avoid the use of parentheses.
*

*> Please find your input file with a fixed geometry that runs with no problem.
*

*>
*

*> Few additional comments:
*

*> - use meaningful names for bodies and regions: it would be easier to debug/maintain a geometry
*

*> - your hexagonal plates are identical and aligned: to define the hexagon you only need 6 planes that you can reuse for each plate
*

*> - prefer XYP, YZP, XZP to PLA: again it would be easier to maintain the geometry
*

*>
*

*> Best regards, luigi
*

*>
*

*>> On 2 Mar 2017, at 21:32, panov_at_dec1.sinp.msu.ru wrote:
*

*>>
*

*>> Dear All,
*

*>>
*

*>> I need a geometry represented by a pack of thin hexagonal silicon plates. There is no tool in FLUKA to construct a hexagonal prism, therefore I construct each hexagonal plate (actually a prism with a small height) by eight generic half-spaces (PLA). To define such a geometry I have to use name-based free-format notation together with parentheses to define vacuum region. My construction works very well with 1,2,3 and 4 plates in the pack, but it does not work with 5 and more plates (I actually need 60 plates as a minimum). A task with more than 4 plates produces segmentation violation error after 5-10 seconds of normal running. The error message about segmentation violation appears in the log-file, not in the console:
*

*>>
*

*>> --------------------------
*

*>> Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
*

*>>
*

*>> Backtrace for this error:
*

*>> #0 0x7FD05A82AE08
*

*>> #1 0x7FD05A829F90
*

*>> #2 0x7FD059F5C4AF
*

*>> #3 0x7FD05A073F45
*

*>> #4 0x7ACC0B in copytx_ at copytx.f:48
*

*>> #5 0x5E2A14 in rpnrul_ at rpnrul.f:157 (discriminator 2)
*

*>> #6 0x5E1125 in rpnorm_ at rpnorm.f:57
*

*>> #7 0x5D163D in flkcgi_ at flkcgi.f:1525
*

*>> #8 0x4DC867 in geoinp_ at geoinp.f:238
*

*>> #9 0x405E09 in flukam_ at flukam.f:2961
*

*>> #10 0x40223B in fluka at fluka.f:289
*

*>> #11 0x7FD059F4782F
*

*>> -----------------------------
*

*>>
*

*>> Can you help to overcome this problem? I send two collections of files, for normal work with four plates and for error with five plates.
*

*>>
*

*>> Thank you in advance,
*

*>> Alexander
*

*>> <Work4.zip><DoesNotWork5.zip>
*

__________________________________________________________________________

You can manage unsubscription from this mailing list at https://www.fluka.org/fluka.php?id=acc_info

Received on Sat Mar 04 2017 - 06:28:16 CET

Date: Fri, 03 Mar 2017 23:43:27 +0300

Dear Luigi,

Thank you very much for your help. I understand what was my main error.

In fact I should expand the expression with parentheses by hands and

write it out explicitly without parentheses at all using the boolean

algebra. The usage of the same half-spaces to define many different

regions also is very good idea. Unfortunately I can't use meaningful

names for bodies and regions since I program very complicated geometry

with many thousands regions and I don't write input file for it by hand.

I write special program that produces the input file with all this

variety of things. Therefore I have to generate all names automatically.

Thank you again,

Alexander

On 2017-03-03 12:46, Luigi Salvatore Esposito wrote:

__________________________________________________________________________

You can manage unsubscription from this mailing list at https://www.fluka.org/fluka.php?id=acc_info

- application/octet-stream attachment: HexagonTest5-.inp__size_7855

*
This archive was generated by hypermail 2.3.0
: Sat Mar 04 2017 - 06:28:17 CET
*