Zuzanna

Interconnected cycles

2

As in email by MK

Avoid transposing in: ``` def prepare_output_with_grouping(self, group_sums, output_columns, one_core): group_by_column = self.settings["GROUP_BY_COLUMN"] log_message("Preparing output...", show_time=True, print_and_save=one_core) lst_dfs = [] for group, data in group_sums.items(): group_df = pd.DataFrame(data=np.transpose(data), columns=output_columns) group_df.insert(0, group_by_column,...

Currently, we calculate batch_size (to avoid memory error) only once. If memory error varies, maybe we should recalculate it after each batch. ```python while batch_start < range_end: # Calculate the...

You can redirect prints from terminal to the external text file. Something like that: ``` import os import sys # Open a file file = open("output.txt", "w") # Get the...

Don't use explicitly "multiprocessing" package; instead try to use map from the dask package because it incorporates multiprocessing. Currently, we split model points into cpu_count parts and calculate each of...

 Replace this csv results with printed data frame (values under the header) so that it's easier to read

Add NUM_STOCHASTIC_SCENARIOS to docs on settings

1

Add `NUM_STOCHASTIC_SCENARIOS` to: https://cashflower.acturtle.com/settings.html

```python @variable(array=True) def my_var(stoch): return [*range(720)] * stoch ```

Firstly we need a very stable Python version of the package, before moving to Cython. So this one is in general a good way forward but once we have tested...

Model-point dependent projection horizon

1

Currently we set projection horizon as T_MAX_CALCULATION setting. It's worth considering projection horizon to be model point dependent. Then there could be less calculations and also no need to a...