openbraininstitute / neurodamus / 14037603037

90%

Build Type

push

github

Committed by web-flow

Commit Message

unit test for loadbalance (#130)

## Context
In the ringtest circuit, we've changed to create cell components and
geometries from morphology files. The usual ring circuit cells read a
morphology file which creates soma + 2 dendrites. As discussed in #97,
to test the load balance, we create a new imbalanced `RingA` nodes files
which reads different morphologies, i.e. a big cell of 10 dendrites + 2
small cells of 2 dendrites, in order to test the loadbalance mode
`MultiSplit` and `WholeCell`.

## Scope
- Emodel template
`tests/simulations/ringtest/hoc/B_BallStickBigCell.hoc`->
`tests/simulations/ringtest/hoc/TestCell.hoc` which creates cell
components and geometries from a morphology file
- ASC morphologies `tests/simulations/ringtest/morphologies/asc`:
`cell_big.asc` for 10 dendrites and `cell_small.asc` for 2 dentries.
- `tests/simulations/ringtest/nodes_A.h5` : 3 cells all with
`"morphology": "cell_small"`
- `tests/simulations/ringtest/nodes_A_bigcell.h5` : 1 cells with
`"morphology": "cell_big"` and 2 cells with `"morphology": "cell_small"`



## Testing
- new test `tests/unit/test_loadbalance.py`:  
- move part of the tests from
`tests/integration-e2e/test_loadbalance.py`
- new tests with `lb_mode=MultiSplit` using RingA big cell circuit vs
RingA small cells.
- new tests with `lb_mode=WholeCell` using RingA big cell circuit vs
RingA small cells

- `tests/integration-e2e/test_loadbalance_integration.py`
- retrain the previous integration test in
`tests/integration-e2e/test_loadbalance.py` using `usecase3` circuit
with `lb_mode=WholeCell`

## Review
* [x] PR description is complete
* [x] Coding style (imports, function length, New functions, classes or
files) are good
* [x] Unit/Scientific test added
* [ ] Updated Readme, in-code, developer documentation

---------

Co-authored-by: Juan Jose Garcia Cantero <juanjose.garcia@openbraininstitute.org>

Run Details

0 of 4 new or added lines in 1 file covered. (0.0%)

1 existing line in 1 file now uncovered.

6194 of 9002 relevant lines covered (68.81%)

0.93 hits per line