In this section, we show how to make a stateful computation from nGraph’s stateless operations. The basic idea is that any computation with side-effects can be factored into a stateless function that transforms the old state into the new state.

## An example from C++¶

Let’s start with a simple C++ example, a function `count`

that
returns how many times it has already been called:

```
int count()
{
static int counter = 0;
return counter++;
}
```

The static variable `counter`

provides state for this function. The
state is initialized to 0. Every time `count`

is called, the current
value of `counter`

is returned and `counter`

is incremented. To
convert this to use a stateless function, define a function that
takes the current value of `counter`

as an argument and returns the
updated value.

```
std::tuple<int, int> stateless_count(int counter)
{
return std::tuple<int, int>(counter, counter + 1);
}
```

To use this version of counting,

```
int counter = 0;
{
auto r(stateless_count(counter));
counter = std::get<1>(r);
std::cout << std::get<0>(r);
}
std::cout << ", ";
{
auto r(stateless_count(counter));
counter = std::get<1>(r);
std::cout << std::get<0>(r);
}
std::cout << std::endl;
```

## Update in nGraph¶

In working with nGraph-based construction of graphs, updating takes the same approach. During training, we include all the weights as arguments to the training function and return the updated weights along with any other results. For more complex forms of training, such as those using momentum, we would add the momentum tensors as additional arguments and include their updated values as additional results. A simple case is illustrated in the documentation for how to Derive a trainable model.