change option char of wanted_size from "S" to "l" (for wanted sequence length) to avoid similar option char with seed option

Mr clean

See merge request aevol/aevol!27

Point was copied in the for loop, then the value inside the set
was never modified. This commit fix this problem.

In methode `check_result`:
+ remove warning about dangling else branch.
+ give alias to the individuals of version 6 and version 7 to add clarity
and context in the code

Toward simplifying ExpManager_7 to handle only the evolution.
The idea is to put all the IO outside the class.
This Commit simplify the run_a_step method.

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

They should not change anything but some of them could have been the cause of some bugs

Beware, this commit do not affect RAevol code

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

Some of these fixes will also improved Aevol 6 and 7

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

Nonetheless, we limit the test to 10,000 generations, rounding error with float (update_concentration+update_phenotype) can arise

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

[RAevol7] Fix all the issue with constant environment (up to 5000 generations validated against both aevol 6 and 7).
It also contains a fix that should apply to Aevol 7 causing an error on very small genome.

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

[Fuzzy] When searching for x for a given y between 2 points, it could happen that the x found can be out of the range between these two points due to rounding error
When such behavior happens, we just set x to the closest points to avoid adding invalidated points.

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

[RAevol7] Fix a read unitialized value within the RAevol part of the code. WARNINGS: This fix is critical for Aevol too (read comment).
The maximum number of codon within a protein is fixed to 3*64. But it was only 64 in Aevol 7. The maximum number is now 3*64.
Moreover, if this barrier is reached, a log error message is generated.

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

Nevertheless, with OpenMP, it seems that a non deterministic crash happens sometimes.

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

It shoud be working with constant env, dynamic env are not ready yet.

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

Naming has changed, so merge was not straighforward, could contain errors.

ROUZAUD-CORNABAS Jonathan authored 4 years ago and TURPIN Laurent committed 4 years ago

Contain a lot of debug message but RAevol constant env is working (dynamic with signal are still segfault)

When restarting aevol_run in a folder, the old statistics files are
copied before aevol appends the new statistics.
In this process, the first line of data (but not the header!)
disappeared, meaning that:
  - when running aevol_run for the 1st time, data for generation 0
    (written by aevol_create) disappeared.
  - when restarting aevol_run for the nth time, data for generation n-1
    (and all earlier generations) disappeared.

While not very important as the first generations are not so meaningful,
this bug was quite puzzling.