Floating Point Exception (SIGFPE) for iprobes =2 about hos-nwt HOT 3 CLOSED

There was indeed a problem in the current version of the code. The treatment of the case k=0 was done afterwards (without preventing a division over 0). It is actually only detected thanks to the compilation option '-ffpe=invalid' that I was not using.

By the way, to correct it, you should change lines 1510 to 1527 of velocities.f90 to the following:

DO i2=1,1
DO i1=1,n1
index1=i1+n1*(i2-1)
DO ii2=1,1
DO ii1=1,n1
index2=ii1+n1*(ii2-1)
IF(k(ii1,ii2).EQ.0.0_rp) THEN ! Specific treatment of k=0
M(index1,index2) = 1.0_rp
ELSEIF(k(ii1,ii2).LT.50.0_rp) THEN
M(index1,index2) = COS(kx(ii1)*x(i1,i2))*COS(ky(ii2)*y(i1,i2))SINH(k(ii1,ii2)(eta(i1,i2)+1.0_rp))/SINH(k(ii1,ii2))
ELSE
M(index1,index2) = COS(kx(ii1)*x(i1,i2))*COS(ky(ii2)*y(i1,i2))*EXP(k(ii1,ii2)*eta(i1,i2))
ENDIF
ENDDO
ENDDO
ENDDO
ENDDO

from hos-nwt.

Shouldn't it be k(ii1,ii2).LE.tiny

from hos-nwt.

Yes indeed, would be safer.

from hos-nwt.