[BUG] Segmentation fault in sfvfsacrsnh applied to multi layer model

[Dirack]

🐛 Bug report

Describe the bug

A segmentation faut ocurred when I try to use sfvfsa in the multi Layer experiment. In the Flow above:

creGatherStack/experiments/multiLayerModel/cre/SConstruct

Lines 62 to 66 in 28de03d

	# Very Fast Simulated Aneelling Global Optimization (VFSA)
	Flow(crsParameters,dataCube,
	'''
	vfsacrsnh m0=%g v0=%g t0=%g verb=y repeat=3
	''' % (m0,v0,t0))

To Reproduce
Steps to reproduce the behavior:

1. Run the multi layer experiment with 'scons'
2. See error when m0=4Km.

< multiLayerDataCube.rsf /root/madagascar/bin/sfvfsacrsnh m0=4 v0=1.5 t0=1 verb=y repeat=3 > crsParameters-m0-4-t0-0.rsf
sfvfsacrsnh: Active mode on!!!
sfvfsacrsnh: Command line parameters: 
sfvfsacrsnh: m0=4.000000 v0=1.500000 t0=1.000000 c0=0.500000 temp0=10.000000 repeat=3
sfvfsacrsnh: Input file parameters: 
sfvfsacrsnh: n1=1001 d1=0.004000 o1=0.000000
sfvfsacrsnh: n2=161 d2=0.012500 o2=0.000000
sfvfsacrsnh: n3=201 d3=0.025000 o3=0.000000
sh: line 1: 13576 Segmentation fault      (core dumped) /root/madagascar/bin/sfvfsacrsnh m0=4 v0=1.5 t0=1 verb=y repeat=3 < "multiLayerDataCube.rsf" > "crsParameters-m0-4-t0-0.rsf"
scons: *** [crsParameters-m0-4-t0-0.rsf] Error 139

Desktop (please complete the following information):

• OS: Linux CentOS
• Version 7.8kvm

Additional context

This problem probably occurs because standard aperture used in Mvfsacrsnh.c to define a square of the approximated surface where to calculate the semblance is 50 samples in CMP and Offset coordinates.

To check that, I run sfin program:

sfin multiLayerDataCube.rsf 
multiLayerDataCube.rsf:
    in="/root/rsfdata/experiments/multiLayerModel/cre/multiLayerDataCube.rsf@"
    esize=4 type=float form=native 
    n1=1001        d1=0.004       o1=0          label1="Time" unit1="s" 
    n2=161         d2=0.0125      o2=0          label2="Offset" unit2="Km" 
    n3=201         d3=0.025       o3=0          label3="CMP" unit3="Km" 
	32393361 elements 129573444 bytes

The data cube CMP axis dimension is:

~$ echo "201*0.025" | bc -l
5.025

If someone try to sum up vfsa standard aperture to 4Km (m0 position):

 ~$ echo "4+50*0.025" | bc -l
5.250

[Dirack]

Try to use 'interpolatedDataCube2' instead of original 'dataCube'. Because it's sampling is minor than original data, it will decrease the aperture despite the number of samples considered in vfsa aperture remains 50.

~$ echo "50*0.00625" | bc -l
.31250

# Very Fast Simulated Aneelling Global Optimization (VFSA) 
 		Flow(crsParameters,'interpolatedDataCube2', 
 			''' 
 			vfsacrsnh m0=%g v0=%g t0=%g verb=y repeat=3 
 			''' % (m0,v0,t0)) 

Please report this bug to vfsa package repository