Issue with logarithm function in OpenFOAM

fluidflowsteel · February 15, 2017, 13:08

Hi ,

I am able to solve the following PDE successfully in OpenFOAM

fvm::ddt(X) == fvc::div(Y1*Y2*fvc::grad(X1/(1-X1))) where X,X1, Y1 and Y2 are scalar field variables

but when I modify the PDE in the following form

fvm::ddt(X) == fvc::div(Y1*Y2*fvc::grad(log(X1/(1-X1)))) where X1 is 0.09 (i.e less than 1)

I am getting the following error

#0 Foam::error:

rintStack(Foam::Ostream&) at ??:?
#1 Foam::sigFpe::sigHandler(int) at ??:?
#2 ? in "/lib/x86_64-linux-gnu/libc.so.6"
#3 log in "/lib/x86_64-linux-gnu/libm.so.6"
#4 Foam::log(Foam::Field<double>&, Foam::UList<double> const&) at ??:?
#5 ? at ??:?
#6 ? at ??:?
#7 __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6"
#8 ? at ??:?

Can anyone help me how can get rid of the error ?

With regards

alexeym · February 15, 2017, 16:51

Hi,

Is X1 constant? Can it be 0? Can it be negative? Just use X1.writeMinMax(Info) to see the range of X1.

fluidflowsteel · February 16, 2017, 05:05

Hi,

NO X1 is not a constant. X1, Y1 and Y2 are volume scalar field variables. The initial value of X1 is 0.09. I think because of this value (which is very close to 0), the program is showing the error.

With regards

alexeym · February 16, 2017, 05:21

Quote:

Originally Posted by fluidflowsteel

The initial value of X1 is 0.09. I think because of this value (which is very close to 0), the program is showing the error.

Well, this is nonsense-like sentence. If we assume that X1 = 0.09 EVERYWHERE, X1/(1 - X1) == 0.(098901), which is rather far from 0. But if we suppose, that SOMEWHERE X1 == 0 or X1 < 0, then it is a valid reason for FPE.

The error happens on the first solver iteration?

fluidflowsteel · February 16, 2017, 05:31

yes the error happens on first solver iteration. Look grad (logX1) = (1/X1)*grad(X1). since X1 initial value is 0.09 i think because of 1/0.09, the error is coming

alexeym · February 16, 2017, 06:14

And you really think, compiler use chain rule for gradient? Instead of GUESSING, you can:

1. Just CHECK values of X1.
2. Use debug build of OpenFOAM to see exact line number, where error is located.

Or if this thread is psychoanalytic one, yes, we can continue to discuss, your thoughts. Next step is to talk about your feelings about this error. Did you have particular problems with logarithms in childhood?

Also, if I did not make a mistake

$\vec\nabla\left(\log\frac{X1}{1 - X1}\right) = \vec\nabla\left(\log\left(X1\right) - \log\left(1 - X1\right)\right) =$

$= \frac{1}{X1}\vec\nabla{X1} - \frac{1}{1 - X1}\vec\nabla\left(1 - X1\right) = \frac{1}{X1}\vec\nabla{X1} + \frac{1}{1 - X1}\vec\nabla{X1} = \frac{1}{X1\left(1 - X1\right)}\vec\nabla{X1}$

nsf · February 17, 2017, 09:14

Another possible cause is that X1 is zero on one of the boundaries.

/N

February 15, 2017, 13:08	Issue with logarithm function in OpenFOAM	#1
fluidflowsteel New Member fluidflowsteel Join Date: Jun 2016 Posts: 21 Rep Power: 10	Hi , I am able to solve the following PDE successfully in OpenFOAM fvm::ddt(X) == fvc::div(Y1Y2fvc::grad(X1/(1-X1))) where X,X1, Y1 and Y2 are scalar field variables but when I modify the PDE in the following form fvm::ddt(X) == fvc::div(Y1Y2fvc::grad(log(X1/(1-X1)))) where X1 is 0.09 (i.e less than 1) I am getting the following error #0 Foam::error:rintStack(Foam::Ostream&) at ??:? #1 Foam::sigFpe::sigHandler(int) at ??:? #2 ? in "/lib/x86_64-linux-gnu/libc.so.6" #3 log in "/lib/x86_64-linux-gnu/libm.so.6" #4 Foam::log(Foam::Field<double>&, Foam::UList<double> const&) at ??:? #5 ? at ??:? #6 ? at ??:? #7 __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6" #8 ? at ??:? Can anyone help me how can get rid of the error ? With regards

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Map of the OpenFOAM Forum - Understanding where to post your questions!	wyldckat	OpenFOAM	10	September 2, 2021 06:29
issue compiling new turbulence model	perplexed user	OpenFOAM Programming & Development	1	January 13, 2012 04:40
New OpenFOAM Forum Structure	jola	OpenFOAM	2	October 19, 2011 07:55
[swak4Foam] wmake groovyBC in OpenFOAM 1.7 ?	randomid	OpenFOAM Community Contributions	1	August 27, 2010 06:15
Problem with compile the setParabolicInlet	ivanyao	OpenFOAM Running, Solving & CFD	6	September 5, 2008 21:50

February 15, 2017, 16:51		#2
alexeym Senior Member Alexey Matveichev Join Date: Aug 2011 Location: Nancy, France Posts: 1,938 Rep Power: 39	Hi, Is X1 constant? Can it be 0? Can it be negative? Just use X1.writeMinMax(Info) to see the range of X1.

February 16, 2017, 05:05		#3
fluidflowsteel New Member fluidflowsteel Join Date: Jun 2016 Posts: 21 Rep Power: 10	Hi, NO X1 is not a constant. X1, Y1 and Y2 are volume scalar field variables. The initial value of X1 is 0.09. I think because of this value (which is very close to 0), the program is showing the error. With regards Kummi likes this.

February 16, 2017, 05:31		#5
fluidflowsteel New Member fluidflowsteel Join Date: Jun 2016 Posts: 21 Rep Power: 10	yes the error happens on first solver iteration. Look grad (logX1) = (1/X1)*grad(X1). since X1 initial value is 0.09 i think because of 1/0.09, the error is coming

February 16, 2017, 06:14		#6
alexeym Senior Member Alexey Matveichev Join Date: Aug 2011 Location: Nancy, France Posts: 1,938 Rep Power: 39	And you really think, compiler use chain rule for gradient? Instead of GUESSING, you can: 1. Just CHECK values of X1. 2. Use debug build of OpenFOAM to see exact line number, where error is located. Or if this thread is psychoanalytic one, yes, we can continue to discuss, your thoughts. Next step is to talk about your feelings about this error. Did you have particular problems with logarithms in childhood? Also, if I did not make a mistake $\vec\nabla\left(\log\frac{X1}{1 - X1}\right) = \vec\nabla\left(\log\left(X1\right) - \log\left(1 - X1\right)\right) =$ $= \frac{1}{X1}\vec\nabla{X1} - \frac{1}{1 - X1}\vec\nabla\left(1 - X1\right) = \frac{1}{X1}\vec\nabla{X1} + \frac{1}{1 - X1}\vec\nabla{X1} = \frac{1}{X1\left(1 - X1\right)}\vec\nabla{X1}$ Tobi and Kummi like this.

February 17, 2017, 09:14		#7
nsf Senior Member Nicolas Edh Join Date: Mar 2010 Location: Uppsala, Sweden Posts: 123 Rep Power: 18	Another possible cause is that X1 is zero on one of the boundaries. /N