停电，中断计算，续算时报错：mpirun Floating point exception，一次性计算没问题

[五好青年]

1. 最近经常停电，然后续算，总是报错，提示下面内容
2. 只要完整，一次性计算，就不会报错
3. 如果中断计算，再接着续算，大概30%概率会遇到以下报错：
   求大佬帮忙看下，为什么断电续算会报错呢？

Time = 14.626952879

Courant Number mean: 0.0162202738735 max: 0.547412638078 velocity magnitude: 0.972694272435
DILUPBiCG:  Solving for Ux, Initial residual = 5.51751945657e-05, Final residual = 1.21131845941e-07, No Iterations 1
DILUPBiCG:  Solving for Uy, Initial residual = 0.00192416415543, Final residual = 2.83092826269e-06, No Iterations 1
DILUPBiCG:  Solving for Uz, Initial residual = 0.00192162985727, Final residual = 2.27969090213e-06, No Iterations 1
GAMGPCG:  Solving for p, Initial residual = 0.0445458617273, Final residual = 0.000312778215492, No Iterations 3
IB time step continuity errors : sum local = 1.94733457266e-10, global = 1.30333893019e-13, cumulative = -2.85480937623e-12
GAMGPCG:  Solving for p, Initial residual = 0.00400247451546, Final residual = 2.29650918295e-05, No Iterations 4
IB time step continuity errors : sum local = 1.43803150508e-11, global = 2.57306076116e-13, cumulative = -2.59750330012e-12
GAMGPCG:  Solving for p, Initial residual = 0.000264903909441, Final residual = 2.47545703992e-06, No Iterations 6
IB time step continuity errors : sum local = 1.5492859545e-12, global = 2.31568779699e-13, cumulative = -2.36593452042e-12
Thrust from Body Force = 0.0974865447818	Thrust from Act. Line = 0.0977889499694	Ratio = 0.996907573017
Torque from Body Force = 0.00155780274379	Torque from Act. Line = 0.00156265463869	Ratio = 0.996895094552
ExecutionTime = 2028.52 s  ClockTime = 2095 s

fieldAverage fieldAverage1 output:
    Calculating averages

Time = 14.62804371

Courant Number mean: 0.0162202632119 max: 0.54826182015 velocity magnitude: 0.974203179026
DILUPBiCG:  Solving for Ux, Initial residual = 5.51651007481e-05, Final residual = 1.21918770605e-07, No Iterations 1
DILUPBiCG:  Solving for Uy, Initial residual = 0.00192414480744, Final residual = 2.62900058636e-06, No Iterations 1
DILUPBiCG:  Solving for Uz, Initial residual = 0.0019217484706, Final residual = 2.60889084187e-06, No Iterations 1
GAMGPCG:  Solving for p, Initial residual = 0.0444294506698, Final residual = 0.000310879570179, No Iterations 3
IB time step continuity errors : sum local = 1.93563121456e-10, global = 2.67585622386e-13, cumulative = -2.09834889803e-12
GAMGPCG:  Solving for p, Initial residual = 0.00397418566853, Final residual = 2.60870920638e-05, No Iterations 4
IB time step continuity errors : sum local = 1.63549690983e-11, global = 2.1607547894e-13, cumulative = -1.88227341909e-12
GAMGPCG:  Solving for p, Initial residual = 0.000265252820756, Final residual = 2.33165488111e-06, No Iterations 6
IB time step continuity errors : sum local = 1.45553081823e-12, global = 8.3257735207e-14, cumulative = -1.79901568389e-12
Thrust from Body Force = 0.0976809290426	Thrust from Act. Line = 0.097986534837	Ratio = 0.996881144997
Torque from Body Force = 0.00156809160673	Torque from Act. Line = 0.00157301669264	Ratio = 0.996869018662
triangles: 240 hit: 13039
[dyfluid:02855] *** Process received signal ***
[dyfluid:02855] Signal: Floating point exception (8)
[dyfluid:02855] Signal code:  (-6)
[dyfluid:02855] Failing at address: 0x3e800000b27
[dyfluid:02855] [ 0] /lib/x86_64-linux-gnu/libc.so.6(+0x354a0)[0x7ff3c6cfd4a0]
[dyfluid:02855] [ 1] /lib/x86_64-linux-gnu/libc.so.6(gsignal+0x38)[0x7ff3c6cfd418]
[dyfluid:02855] [ 2] /lib/x86_64-linux-gnu/libc.so.6(+0x354a0)[0x7ff3c6cfd4a0]
[dyfluid:02855] [ 3] /home/dyfluid/foam/foam-extend-4.0/lib/linux64GccDPOpt/libsampling.so(_ZN4Foam16vtkSurfaceWriterIdE9writeDataERNS_7OstreamERKNS_5FieldIdEE+0x7f)[0x7ff3c993148f]
[dyfluid:02855] [ 4] /home/dyfluid/foam/foam-extend-4.0/lib/linux64GccDPOpt/libsampling.so(_ZNK4Foam16vtkSurfaceWriterIdE5writeERKNS_8fileNameES4_RKNS_5FieldINS_6VectorIdEEEERKNS_4ListINS_4faceEEES4_RKNS5_IdEEb+0x3a1)[0x7ff3c99333c1]
[dyfluid:02855] [ 5] /home/dyfluid/foam/foam-extend-4.0/lib/linux64GccDPOpt/libimmersedBoundary.so(_ZNK4Foam28immersedBoundaryFvPatchFieldIdE5writeERNS_7OstreamE+0x48e)[0x7ff3c949416e]
[dyfluid:02855] [ 6] pisoIbFoam-HAT-ALMorADM(_ZNK4Foam14GeometricFieldIdNS_12fvPatchFieldENS_7volMeshEE22GeometricBoundaryField10writeEntryERKNS_4wordERNS_7OstreamE+0x114)[0x4595a4]
[dyfluid:02855] [ 7] pisoIbFoam-HAT-ALMorADM(_ZNK4Foam14GeometricFieldIdNS_12fvPatchFieldENS_7volMeshEE9writeDataERNS_7OstreamE+0xaf)[0x45cfaf]
[dyfluid:02855] [ 8] /home/dyfluid/foam/foam-extend-4.0/lib/linux64GccDPOpt/libfoam.so(_ZNK4Foam11regIOobject11writeObjectENS_8IOstream12streamFormatENS1_13versionNumberENS1_15compressionTypeE+0x41c)[0x7ff3c8194ffc]
[dyfluid:02855] [ 9] /home/dyfluid/foam/foam-extend-4.0/lib/linux64GccDPOpt/libfoam.so(_ZNK4Foam14objectRegistry11writeObjectENS_8IOstream12streamFormatENS1_13versionNumberENS1_15compressionTypeE+0x1db)[0x7ff3c8198cdb]
[dyfluid:02855] [10] /home/dyfluid/foam/foam-extend-4.0/lib/linux64GccDPOpt/libfoam.so(_ZNK4Foam14objectRegistry11writeObjectENS_8IOstream12streamFormatENS1_13versionNumberENS1_15compressionTypeE+0x1db)[0x7ff3c8198cdb]
[dyfluid:02855] [11] /home/dyfluid/foam/foam-extend-4.0/lib/linux64GccDPOpt/libfoam.so(_ZNK4Foam4Time11writeObjectENS_8IOstream12streamFormatENS1_13versionNumberENS1_15compressionTypeE+0x542)[0x7ff3c81b5212]
[dyfluid:02855] [12] pisoIbFoam-HAT-ALMorADM[0x421795]
[dyfluid:02855] [13] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf0)[0x7ff3c6ce8830]
[dyfluid:02855] [14] pisoIbFoam-HAT-ALMorADM[0x424969]
[dyfluid:02855] *** End of error message ***
--------------------------------------------------------------------------
mpirun noticed that process rank 7 with PID 2855 on node dyfluid exited on signal 8 (Floating point exception).
--------------------------------------------------------------------------

[李东岳]

可以看下controlDict里面的写入精度，提高一下。有可能是重新启动计算之后，最后一个时间步的写入结果精度不够高，导致后面发散了

[五好青年]

@李东岳
东岳老师，我的精度是12，请问这个一般够么？感谢

  writeFormat       binary;

  writePrecision    12;

  writeCompression  compressed;

  timeFormat        general;

  timePrecision     12;

[李东岳]

@五好青年 够。那其他的不清楚为啥了。

另外，为啥老断电，这个很虚啊

[五好青年]

@李东岳
东岳老师，我仔细核对了：成功续算的case+报错浮点溢出的case
发现报错均出现在第一次结果写出的位置

我是间隔20个时间步保存一次，就是在第20步写入数据的时候，提示上面的报错

[李东岳]

@五好青年 写文件就出错，不写文件呢？如果30步一写呢

[五好青年]

@李东岳

1. 时间精度，修改为6，直接无法读取上一步结果
2. 时间精度，修改为20，同样到首次写入结果的位置就报错
3. 20步写入结果，更改为1步写入结果，30步写入结果。续算时，都在首次写入结果时报错，而且报错内容相同，如提问中的浮点溢出

[Tens]

@五好青年 是自己改的程序吗，我之前好像遇到过。
可能是某个变量在计算过程中随着时间变化但没有保存，重新开始计算后这个变量被初始化，初始化的值和当前时刻的差很多，导致发散