Timeline
04/05/18:
- 23:14 Changeset [773] by
- Improved styling implementation, fixes issue number 18 [refs #62]
- 20:03 Changeset [772] by
- Unified #define modifiers in f1 and f4, fixes issue number 16 [refs #62]
- 17:20 Changeset [771] by
- Added color and size modifiers in f4; avoided unnecessary conditions in styling; fixes issue 17 and concerns issue 18 [refs #62]
- 17:16 Changeset [770] by
- Avoided "min" and "max" as variable names, fixes issue 22 [refs #62]
03/30/18:
- 01:56 Changeset [769] by
- updated for Framsticks version 5.0rc13 (Model checkpoints, f4 improvements, doc corrections)
- 01:12 Changeset [768] by
- Reverted to 100x repeat-until-valid-mutation, required due to how mutations are implemented now (first, mutation sub-type is selected, and this sub-type may be impossible to apply to a given genotype, which may cause frequent failures)
03/29/18:
- 22:52 Changeset [767] by
- More strict parsing of the repetition counter; bug fixes; improved docs [refs #62]
- 22:51 Changeset [766] by
- Added a number of useful methods in 2D
- 20:36 Ticket #55 (f4: consider changing SString to std::string (except in SDK interface ...) closed by
- fixed
- 20:36 Ticket #56 (f4: support "modifiers" and excluded modifiers just as in f1) closed by
- fixed
- 20:35 Ticket #57 (f4: update parsing of numbers and make it more strict) closed by
- fixed
- 20:35 Ticket #60 (f4: consider unifying sources with f1 regarding "modifiers") closed by
- fixed
- 20:35 Ticket #61 (f4: after updating sources and adding new features, test compatibility ...) closed by
- fixed
03/28/18:
- 02:33 Changeset [765] by
- Wider ranges of allowed values for mutation when only partial information of min/max/default is provided
- 02:29 Changeset [764] by
- Less deterministic (less absorbing) mutation for very narrow allowed ranges (performs reflect+wrap, previously reflect+absorb)
- 02:28 Changeset [763] by
- Higher precision when storing variable values
- 02:27 Changeset [762] by
- Default sample values for a 2D problem, not 1D; added documentation and a sample 2D fitness function; crossover better protected against different lengths of parent vectors
- 02:26 Changeset [761] by
- Introduced a macro for the number of times GenMan? tries to repeat a mutation or crossover when the operator does not return acceptable genotype; decreased the number of repetitions from 100 to 10
03/15/18:
- 23:06 Ticket #62 (Multiple issues with f4 to fix each one independently, discuss when needed) created by
- 1. add f4 tests with invalid genotypes (to test problematic parsing, …
- 22:59 Ticket #59 (f4: effectors/neurons should not be hardcoded (N|@), but rather use ...) closed by
- fixed
- 22:58 Ticket #58 (f4: receptors/inputs should not be hardcoded (GTS*), but rather use ...) closed by
- fixed
- 22:55 Changeset [760] by
- - added support for new API for neuron types and their properties - added support for checkpoints
- 22:53 Changeset [759] by
- f1 and f4 stick properties unified and moved to a separate structure and file (geneprops.cpp/h)
- 22:50 Changeset [758] by
- Added a few useful functions providing lists of neurons with specific properties
- 00:00 Changeset [757] by
- Removed appending the original fn genotype as a comment in the converted f0 genotype (was only useful for debugging)
03/14/18:
- 23:58 Changeset [756] by
- Added frams/genetics/fn to ALL_DIRS so make clean cleans all dirs
- 23:52 Changeset [755] by
- Renamed ParamTree::Node -> ParamTree::ParamTreeNode? to avoid name conflicts with js
Note: See TracTimeline
for information about the timeline view.
