cadCAD 0.4.xx - Using/Updating notebooks

cadCAD How-To's
1

Using cadCAD 0.4.17

(see below for updating older projects)

Use python3.6 through virtual environment (optional)

python3.6 -m venv venv
source venv/bin/activate

Install cadCAD

pip install cadCAD==0.4.17

Verify:

$ pip freeze
cadCAD==0.4.17
dill==0.3.2
fn==0.4.3
funcy==1.14
multiprocess==0.70.10
numpy==1.19.0
pandas==1.0.5
pathos==0.2.6
pox==0.2.8
ppft==1.6.6.2
python-dateutil==2.8.1
pytz==2020.1
six==1.15.0

When initially using python 3.8, these packages: pathos, funcy, fn, dill, multiprocess, pox, ppft; had - error: invalid command 'bdist_wheel'
But looked to have still worked.

Additional python dependencies needed:

  • pip install jupyterlab
  • pip install matplotlib
  • pip install networkx
  • pip install scipy

Updating older projects

In essense the core difference in upgrading to 0.4.17 is
in how Configuration objects are created and passed to the Executor. Also a few variations around use of the Executor.

Configurations

imports

from cadCAD.configuration.utils import config_sim
from cadCAD.configuration import append_configs
from cadCAD import configs

If no longer accessing a configuration directly, you can remove:

from cadCAD.configuration import Configuration

Adding configurations

Rather than creating individual Configuration objects with:
config = Configuration(initial_state, partial_state_update_blocks, sim_config)

Configurations are appended to the (imported) configs array via:
append_configs(initial_state, partial_state_update_blocks, sim_configs)

And as per docs each element (dictionary) should be preprocessed with config_sim(), and the result of this (dictionary or array) is passed as sim_configs.

Executor

The execution mode needs 3 changes:

  • ExecutionContext mode names have changed, eg can replace single_proc with local_mode.
  • Executor should be passed the configs array from above.
  • finally, sessions are additionally when calling execute()

That is, this:

exec_context = ExecutionContext(exec_mode.single_proc)
executor = Executor(exec_context, [config]) # Pass the configuration object inside an array
raw_result, tensor = executor.execute() # The `execute()` method returns a tuple; its first elements contains the raw results

Changes to:

exec_context = ExecutionContext(exec_mode.local_mode)\n",
executor = Executor(exec_context, configs) # Pass the configuration object inside an array\n",
raw_result, tensor, sessions = executor.execute() # The `execute()` method returns a tuple; its first elements contains the raw results"

Executor results

The results now contain an additional key, simulation, which is incremented for each sim_configs dictionary element passed to append_configs().

Note each config dictionary that has multiple runs (N > 1) will create multiple Configuration elements in the configs array, each with N = 1.

Per dictionary, the simulation_id will be shared in its corresponding Configurationobjects. The results of execute() return this under simulation.

4 Likes
2

See this latest guide for updating to 0.4.23

Also the demos repo for demos/tutorial examples

2 Likes