Program received signal SIGSEGV, Segmentation fault

[lifeisgood]

@MikeyBoy: the call stack is empty. This is what I get from the Debugger tab:

Active debugger config: GDB/CDB debugger:Default
Building to ensure sources are up-to-date
Selecting target:
Debug
Adding source dir: E:\BC\Courses\Pablo_Methods\HW\Mendoza 2010 Replication\Lia Replication\
Adding source dir: E:\BC\Courses\Pablo_Methods\HW\Mendoza 2010 Replication\Lia Replication\
Adding file: E:\BC\Courses\Pablo_Methods\HW\Mendoza 2010 Replication\Lia Replication\bin\Debug\Lia_Replication.exe
Changing directory to: "E:/BC/Courses/Pablo_Methods/HW/Mendoza 2010 Replication/Lia Replication/."
Set variable: PATH=.;D:\Program Files (x86)\CodeBlocks\MinGW\bin;D:\Program Files (x86)\CodeBlocks\MinGW;C:\ProgramData\Oracle\Java\javapath;C:\Program Files (x86)\Common Files\Intel\Shared Libraries\redist\intel64\compiler;C:\Program Files (x86)\Common Files\Intel\Shared Libraries\redist\ia32\compiler;C:\Program Files (x86)\Intel\iCLS Client;C:\Program Files\Intel\iCLS Client;C:\Windows\System32;C:\Windows;C:\Windows\System32\wbem;C:\Windows\System32\WindowsPowerShell\v1.0;C:\Program Files\Intel\Intel(R) Management Engine Components\DAL;C:\Program Files\Intel\Intel(R) Management Engine Components\IPT;C:\Program Files (x86)\Intel\Intel(R) Management Engine Components\DAL;C:\Program Files (x86)\Intel\Intel(R) Management Engine Components\IPT;C:\Program Files (x86)\Intel\OpenCL SDK\2.0\bin\x86;C:\Program Files (x86)\Intel\OpenCL SDK\2.0\bin\x64;D:\Program Files\MATLAB\R2016b\runtime\win64;D:\Program Files\MATLAB\R2016b\bin;D:\Program Files\MATLAB\R2015a\runtime\win64;D:\Program Files\MATLAB\R2015a\bin;D:\Program Files (x86)\QuickTime\QTSystem;C:\Program Files (x86)\Windows Live\Shared;D:\Program Files (x86)\Calibre2;D:\Program Files (x86)\MiKTeX 2.9\miktex\bin;D:\Program Files (x86)\LyX 2.2\Perl\bin;D:\Program Files (x86)\Skype\Phone;C:\Program Files (x86)\Windows Kits\10\Windows Performance Toolkit;C:\ProgramData\chocolatey\bin;D:\MinGW\bin
Starting debugger: D:\Program Files (x86)\CodeBlocks\MINGW\bin\gdb.exe -nx -fullname -quiet -args "E:/BC/Courses/Pablo_Methods/HW/Mendoza 2010 Replication/Lia Replication/bin/Debug/Lia_Replication.exe"
done
Registered new type: wxString
Registered new type: STL String
Registered new type: STL Vector
Setting breakpoints
Debugger name and version: GNU gdb (GDB) 7.6.1
Child process PID: 12780
Program received signal SIGSEGV, Segmentation fault.

[Thomas1965]

Some problems:
line 183 if(i=j) // maybe you mean i == j
line 409 if(templb>=0 & tempvb>=0) maybe you mean &&
line 524 if(jjk+jjb>=1|skip==0) maybe you mean ||
line547 else if(jjk+jjb==0&skip==1) maybe you mean &&

In VS 2015 I get a stack overflow exception from chkstk.asm in the first line of main. I guess too many local arrays. Incresing the stack size to 10 MB fixed it.

[lifeisgood]

@Thomas1965: I corrected all the errors above and did settings -> compiler -> linker settings -> other linker options -> -WI,--stack,10000000

Still I got the same segmentation fault.

[lifeisgood]

@jlb: I will try to add output lines. Thank you for the suggestions!

[ne555]

upload your current code to github or similar.

[Thomas1965]

I changed it under Project->Build options and did a complete rebuild under Build->Rebuild or Crtl-F11
It works now.
https://pasteboard.co/1V0maKjMg.jpg

[jlb]

Instead of playing with increasing the stack space you really really should break the program into quite a few smaller functions instead. And as you're breaking the program into functions try to simplify the logic.

[lifeisgood]

@ne555: I have heard a lot about GitHub. Now I am creating a profile on that website. Thank you for the suggestion!

[lifeisgood]

@Thomas1965: omg it totally did! Thanks so much!

[lifeisgood]

@jlb: I will be fiddling with C++ this summer and this is first thing on the list <\determined>

[lifeisgood]

I put the code onto GitHub. https://github.com/flowerysleeves/replication/blob/master/main.cpp I am confused as to why the code would not stop running when the variable "con" hits 1 in the while loop.

[ne555]

$ g++ main.cpp -ggdb -Wall -Wextra -Wpedantic -fsanitize=undefined
main.cpp:378:33: warning: ‘binding’ may be used uninitialized in this function [-Wmaybe-uninitialized]

`binding' is a boolean, you never set it to anything.

(in bold is the output of the sanitizer)

$ ./a.out
omega=1.8461
sigma=2
delta=0.0880044
beta=0.305367
a=2.75
main.cpp:297:21: runtime error: index 8 out of bounds for type 'double [8]' (sa)
sa[m]=exp(Ashock[l])*abar; // state for TFP
main.cpp:298:21: runtime error: index 8 out of bounds for type 'double [8]' (sr)
sr[m]=exp(Rshock[i])*(1+rbar); // state for the interest rate
main.cpp:299:29: runtime error: index 8 out of bounds for type 'double [8]' (qb)
qb[m]=1/sr[m]; // state for bond prices
main.cpp:299:21: runtime error: index 8 out of bounds for type 'double [8]' (sr)
qb[m]=1/sr[m]; // state for bond prices
main.cpp:300:21: runtime error: index 8 out of bounds for type 'double [8]' (sp)
sp[m]=exp(Pshock[j])*pbar; // state for price of int'l goods
main.cpp:343:28: runtime error: index 8 out of bounds for type 'int [8]'
kzt[i][j][l]=-999999999;
main.cpp:344:28: runtime error: index 8 out of bounds for type 'int [8]'
kz2[i][j][l]=-999999999;
con5=0
con=1

Now look at the loops where those invalid access happen

    m=0;
    double sa[8];
    double sr[8];
    double qb[8];
    double sp[8];
    for(i=0;i<nr;i=i+1){ //¿why aren't i, j, l, local to the loops?
        for(j=0;j<np;j=j+1){
            for(l=0;l<na;l=l+1){
                m=m+1; //<-- m starts at 1
                sa[m]=exp(Ashock[l])*abar;
                sr[m]=exp(Rshock[i])*(1+rbar);
                qb[m]=1/sr[m];
                sp[m]=exp(Pshock[j])*pbar;
            }
        }
    }

also, it is not inmediate obvious that the size of the arrays is na*np*nr

    for(i=0;i<ne;i=i+1){
        for(j=0;j<nb;j=j+1){
            for(l=0;l<nk;l=l+1){
                kzt[i][j][l]=-999999999;
                kz2[i][j][l]=-999999999;
            }
        }
    } // initialize the values of kzt and kz2 to make sure that we don't stop at the first iteration 

if you see the declaration

int kzt[nk][nb][ne]; // counter for last period's capital
int kz2[nk][nb][ne]; // counter for capital from two periods ago 

you've got i and l backwards

[lifeisgood]

@ne555: I updated the code. Binding is now set to 1 under some coditions. errors with sa etc are fixed, and the i and l are switched. Now I need to fix some numerical problems. Some values are supposed to be positive but they are always negative. Thank you for looking it over!

[lifeisgood]

If it is okay, I would like to ask another question: with ofstream, I am able to output a lot of variables to .csv files except for a variable called "con".

    ofstream record_con;
    record_con.open("con.csv");

and then later:record_con << con << endl;

all of the code is on github in a previous reply.

[ne555]

> Binding is now set to 1 under some coditions
binding appears two times in all your code

bool binding; //line 90, declaration
if(binding==1){ //line 400, use 

you never initialise it.
¿what branch do you think your code will follow?


> Now I need to fix some numerical problems. Some values are supposed to be
> positive but they are always negative.
I still got some invalid access. Now for the `V' matrix, lines 468 and 475.
use the -fsanitize=undefined when compiling and it will pinpoint them for you when you run your program.

Your code is hard to follow: meaningless variable names, recycled iterators, lack of functions, lots of code...
Also, I know nothing about the problem you are trying to solve or the method that you're following.
Sorry, but you are on your own there.


> I am able to output a lot of variables to .csv files except for a variable
> called "con".
you need to describe the issue better.
when I run your code, it generates a `con.csv' file with twenty-six lines with a 0, and then some lines with 26 (I had killed the program)
¿you don't have the `con.csv' file? ¿is it empty?

[lifeisgood]

@ne555:

> Binding problem: now I realized that the variable was not useful, so I took it away

> using the Undefined Behavior Sanitizer is a little beyond me at the moment as I am using an IDE on Windows that has the compiler linked to it, but I am excited about it. A couple other people have mentioned that I needed to have functions, so I need to learn that as well.

> I don't have the file. All the other ones are there except for this one lol But it's okay. I have cout to go to in this case.

Thanks again for your help!

[lifeisgood]

@ne555: while attempting to create a function I have encountered the following error while compiling:

E:\BC\Courses\Pablo_Methods\HW\Mendoza 2010 Replication\Lia Replication\main.cpp|142|error: cannot convert 'double (*)[8][8]' to 'double***' in return|

Do you have any idea what is wrong, please? Thank you very much!

[ne555]

don't return arrays from functions. Those variables are local and will be destroyed when the function ends.
Instead pass it as an extra argument to the function or use std::array

void joint_stationary_distribution_k_v_e(double V[nk][nb][ne],double pi[ne][ne],double kkn[nk][nb][ne], double bbn[nk][nb][ne], double pka[nk][nb][ne]);
//double*** pkve=joint_stationary_distribution_k_v_e(V,pi,kkn,bbn);
double pkve[nk][nb][ne];
joint_stationary_distribution_k_v_e(V,pi,kkn,bbn,pkve);

std::array has some advantages, like knowing its own size and being copyable.
Also, take a look at std::valarray that works as a mathematical vector

[lifeisgood]

@ne555: I made changes but I wasn't able to get any output from pkve in the main program. :-/

[ne555]

> using the Undefined Behavior Sanitizer is a little beyond me at the moment as
> I am using an IDE on Windows that has the compiler linked to it
you may use http://coliru.stacked-crooked.com/ to compile and run your code

integer overflow on line 682
out of bounds in 659 and 666 (`V' matrix)


> I wasn't able to get any output from pkve in the main program. :-/
I've got -nan (not a number)
Check out your computations, pka2 keeps growing.