(calibration_markov_chain_run)=
# Running the estimate, exporting and loading chains, and continuing the calculation

The computation can be performed using the standard `estimateParameters` method, which 
automatically saves the chains in their associated files. Although it is generally unnecessary 
(since saving is done automatically), it is still possible to manually export the results of a 
chain or reload them later using `export_chain_MCMC` and `read_chain_MCMC`, whose prototypes 
are the following:

````{only} cpp
```cpp
// Export one chain
void export_chain_MCMC(const char *fileName);
// Read one chain
void read_chain_MCMC(const char *fileName);
```
````

````{only} py
```python
# Export one chain
export_chain_MCMC(fileName)
# Read one chain
read_chain_MCMC(fileName)
```
````

Both methods take a single string argument specifying the file name used for exporting or 
loading the chain results.

In most cases, these methods are not required. They become useful when a computation finishes 
without reaching convergence and the user wishes to extend the run. This can be done with the 
`continueCalculation` method, which executes additional iterations of the MCMC algorithm in 
order to try to achieve convergence. Its prototype is:

````{only} cpp
```cpp
void continueCalculation(int new_Ns);
```
````

````{only} py
```python
continueCalculation(new_Ns)
```
````

This method takes a single integer argument that specifies the number of additional iterations 
to run. In this case, it may be useful to reload one of the previously computed chains with 
`read_chain_MCMC` before calling `continueCalculation`. Regardless of which chain is loaded, 
the calculation will continue for all initialized chains.