The stacksize might be too small. Increase the stacksize and rerun the code
```
ulimit -s unlimited
export OMP_STACKSIZE=5G
```
Note that the second line is only used if the hybrid parallelization was used. The value 5G is also chosen arbitratily to be *large enough*.
*Note for aiida users: One should make sure to have the stacksize large enough in the code setup. This means adding the two statements above to the `Prepend text` while setting up the aiida code.*
---
##### 4.2 Error loading shared library
**Symptoms:**
The KKRhost code cannot be executed because of:
```
bash-4.2$ ./kkr.x
./kkr.x: error while loading shared libraries: libmkl_intel_lp64.so: cannot open shared object file: No such file or directory
```
**Fix:**
The code was compiled with a different compiler/librabry environment. This happens for example if for compilation `source compiler-select intel` (PGI-1 in-house setup) was used while the code was executed in another shell where the envoronment is different (e.g. no or different `compiler-select` chosen).
To fix this one simply has to use the same environment for running as for compilation (e.g. in this example also run `source compiler-select intel` in the shell where the code is run).
*Note for aiida users: In the code setup one has to make sure to use the same environment for running as for compilation by adding the same `compiler-select` choice to the `Prepend text` in the code setup.*
---
##### 4.3 Voronoi does not find starting potential
**Symptoms:**
The Voronoi code exits with (or similar for other potential file than Cu)
```
$ ./voronoi.exe
...
**** JELLSTART POTENTIALS ****
From atom No. 1 to atom No. 1
Generating potential for atom 1 at site 1 with shape 1
Using database ....: ElementDataBase/Cu29.pot
Error in JELLSTART
Potential.............Cu29
Does not exist in the database
```
**Fix:**
The `ElementDataBase` directory is not seen by the voronoi code, it needs to be accessible from the directory in which the voronoi code is executed.
Before running voronoi do
```
ln -s <path-to-JuKKR-dir>/ElementDataBase .
```
then rerun voronoi which should work.
*Note for aiida users: Make sure you add the linking to the `ElementDataBase` to the `Prepend text` of the voronoi code at the code setup.*
---
#### 5. Fixing inconsistent inputs
These issues reflext an inconsistent choice in the input parameter but can be fixed if the `inputcard` is modified.
##### 5.1 `R_LOG` too small
**Symptoms:**
The KKRhost code stops with the error message `Error creating newmesh!`:
serial number for files: JuKKR_v3.1-1825-g5e0f83a2_gnu-debug_20200302120953
Calling MADELUNG3D
Exited MADELUNG3D
ERORR: non-spherical part of the potential needs
to be inside the log panel (i.e. R_LOG too small)
atom (I1): 1
R_LOG 0.10000000000000001
Rmesh(IRMIN(I1), I1) 0.41880251818625097
IRMIN(I1) 276
STOP Error creating newmesh!
```
**Fix:**
The `R_LOG` parameter needs to be bigger than `Rmesh(IRMIN(I1), I1)`. In the above example we can fix this by changing/adding the `R_LOG` input in the `inputcard` with a larger value:
```
R_LOG= 0.45
```
---
# FAQ
#### 1. How can I compile the code?
Take a look at the ["compiling the code" quick start guide](Compiling the code).
#### 2. What does keyword *X* in the `inputcard` mean?
Take a look at the [list of inputcard codewords](jumu/inputcard_codewords).
#### 3. What are possible run and test options?
See the [list of run options](jumu/runoption).
#### 4. What to can I do if the code terminates unexpectedly?
##### 4.1 Stacksize too small
**Symptoms:**
The KKRhost code crashes with a segmentation fault
The stacksize might be too small. Increase the stacksize and rerun the code
```
ulimit -s unlimited
export OMP_STACKSIZE=5G
```
Note that the second line is only used if the hybrid parallelization was used. The value 5G is also chosen arbitratily to be *large enough*.
*Note for aiida users: One should make sure to have the stacksize large enough in the code setup. This means adding the two statements above to the `Prepend text` while setting up the aiida code.*
---
##### 4.2 Error loading shared library
**Symptoms:**
The KKRhost code cannot be executed because of:
```
bash-4.2$ ./kkr.x
./kkr.x: error while loading shared libraries: libmkl_intel_lp64.so: cannot open shared object file: No such file or directory
```
**Fix:**
The code was compiled with a different compiler/librabry environment. This happens for example if for compilation `source compiler-select intel` (PGI-1 in-house setup) was used while the code was executed in another shell where the envoronment is different (e.g. no or different `compiler-select` chosen).
To fix this one simply has to use the same environment for running as for compilation (e.g. in this example also run `source compiler-select intel` in the shell where the code is run).
*Note for aiida users: In the code setup one has to make sure to use the same environment for running as for compilation by adding the same `compiler-select` choice to the `Prepend text` in the code setup.*
---
##### 4.3 Voronoi does not find starting potential
**Symptoms:**
The Voronoi code exits with (or similar for other potential file than Cu)
```
$ ./voronoi.exe
...
**** JELLSTART POTENTIALS ****
From atom No. 1 to atom No. 1
Generating potential for atom 1 at site 1 with shape 1
Using database ....: ElementDataBase/Cu29.pot
Error in JELLSTART
Potential.............Cu29
Does not exist in the database
```
**Fix:**
The `ElementDataBase` directory is not seen by the voronoi code, it needs to be accessible from the directory in which the voronoi code is executed.
Before running voronoi do
```
ln -s <path-to-JuKKR-dir>/ElementDataBase .
```
then rerun voronoi which should work.
*Note for aiida users: Make sure you add the linking to the `ElementDataBase` to the `Prepend text` of the voronoi code at the code setup.*
---
#### 5. Fixing inconsistent inputs
These issues reflects an inconsistent choice in the input parameter but can be fixed if the `inputcard` is modified.
##### 5.1 `R_LOG` too small
**Symptoms:**
The KKRhost code stops with the error message `Error creating newmesh!`:
serial number for files: JuKKR_v3.1-1825-g5e0f83a2_gnu-debug_20200302120953
Calling MADELUNG3D
Exited MADELUNG3D
ERORR: non-spherical part of the potential needs
to be inside the log panel (i.e. R_LOG too small)
atom (I1): 1
R_LOG 0.10000000000000001
Rmesh(IRMIN(I1), I1) 0.41880251818625097
IRMIN(I1) 276
STOP Error creating newmesh!
```
**Fix:**
The `R_LOG` parameter needs to be bigger than `Rmesh(IRMIN(I1), I1)`. In the above example we can fix this by changing/adding the `R_LOG` input in the `inputcard` with a larger value:
```
R_LOG= 0.45
```
##### 5.2 `KPOIBZ` too small
**Symptoms:**
The KKRhost code stops with the error message `<BZIRR3D> increase parameter KPOIBZ'`. The reason is that the k-mesh in the inputcard (`BZDIVIDE` input) can't fit into the default size of corresponding working arrays. In the output it also tells you what to do (here an example for `BZDIVIDE= 64 64 64`:
```
...
No. of k-points 262144 > KPOIBZ
STOP <BZIRR3D> increase parameter KPOIBZ'
```
**Fix:**
The `KPOIBZ` parameter needs to be bigger than `product(BZDIVIDE)`. In the above example we can fix this by changing/adding the `KPOIBZ` input in the `inputcard` with a larger value:
