https://github.com/vitay/ANNarchy

sort by:
Revision Author Date Message Commit Date
73ca3c1 Finished version 4.6.8 08 June 2019, 15:54:20 UTC
0715623 Merge branch 'develop' 08 June 2019, 15:43:44 UTC
de6c123 doc: added a notebook on using hyperopt 08 June 2019, 15:42:46 UTC
e6f23fd Bug: proj.w = 1.0 not working when has_single_weight after compile. 08 June 2019, 13:58:14 UTC
c909e37 Fixed Python2.7 bug in previous commit. 06 June 2019, 14:08:53 UTC
512a62d Fixed potential bug in load/save method for newer numpy version >=1.16.3. If an user wants to store .npz we use the numpy.save/numpy.load methods. For those numpy v1.16.3 changes the default values for allow_pickle. We define now the optional allow_pickle=True for numpy.save / numpy.load calls to remain with the behavior of older versions. 06 June 2019, 13:52:17 UTC
d6fa1d7 Merge branch 'master' into develop 25 May 2019, 10:45:08 UTC
2b8e1a5 Fixed bug in load_parameters. Thanks @CarolinScholl. 25 May 2019, 10:43:39 UTC
a9c40b0 wrong setup argument in unittest. 23 May 2019, 09:43:31 UTC
cd00dbe Implemented CurrentInjection on CUDA devices. Added an unittest based on the example in the documentation. 23 May 2019, 09:35:13 UTC
5dd3956 Merge branch 'develop' of https://ai.informatik.tu-chemnitz.de/gogs/annarchy/annarchy into develop 22 May 2019, 18:31:25 UTC
60fedcb Implemented missing deallocators on GPUs. 22 May 2019, 18:16:47 UTC
af31557 Added a `CurrentInjection` specific population allowing to connect a rate-coded population to the conductance ofa spiking population. 22 May 2019, 18:11:27 UTC
72e0758 Bugfix to commit: b65cc445a5 The host defines of built-in functions are required for some parts of the CPU side implementation. 20 May 2019, 17:48:43 UTC
b65cc44 Added the built-in method power() on GPUs. The equation interface of ANNarchy provides a custom power() method to implement a power function with integer exponents as loop and instead of pow(). 20 May 2019, 14:53:41 UTC
8d90fae Bugfix in CUDA introduced by commit 90d33f. 20 May 2019, 08:02:39 UTC
6979f0a Disabled by default the openMP parallelization of event-driven synaptic transmission. By providing disable_omp=False to Projection() one can enable this if required. 16 May 2019, 06:47:42 UTC
0ca9a56 Merge branch 'master' into develop 15 May 2019, 18:32:31 UTC
127f88a Left some prints 15 May 2019, 18:31:35 UTC
f6bd0a7 Added ``clear_all_callbacks()`` method. 15 May 2019, 18:30:38 UTC
12e7552 Merge branch 'master' into develop 15 May 2019, 14:52:23 UTC
5610bfb Bug in simulate_with_callbacks (@every) which could not sort spike times correctly with different callbacks. 15 May 2019, 14:49:37 UTC
90d33f9 Separated intialization of the C++ RNG distribution objects from the other initializations. Otherwise, RNGs which depends on population variables, like: Neuron( parameters=""" mean = 0.0 : population sigma = 1.0 : population """, equations=""" r = Normal(mean, sigma) """ ) will not work. 15 May 2019, 08:41:27 UTC
7942450 Merge branch 'develop' of https://ai.informatik.tu-chemnitz.de/gogs/annarchy/annarchy into develop 14 May 2019, 09:12:12 UTC
38a125e Added another test for structural plasticity. A potential bug remains: the original connectivity is not reset by Projection.reset(). 14 May 2019, 09:11:11 UTC
445966b removed new convolution class, aybe later 07 May 2019, 21:00:15 UTC
89ac4ac Merge branch 'develop' of https://ai.informatik.tu-chemnitz.de/gogs/annarchy/annarchy into develop 06 May 2019, 15:01:34 UTC
9140020 Added flag to disable openMP parallelization for event-driven synaptic transmission. This can be useful for small sparsely connected neural networks. 04 May 2019, 06:24:16 UTC
fc6d00a Bugfix: profile_enabled=True provided to compile() was without effect. 02 May 2019, 05:36:50 UTC
9a9fff6 Extend of commit 2c25ec1 30 April 2019, 13:46:08 UTC
d23c6a5 Bugfix (CUDA) + Accessor Population.sum(target) was not working, as the psp results were not copied back at the end of simulate(). + code-error in mean-template. 30 April 2019, 13:05:46 UTC
d24a45f Global operations (min, max, mean, sum) in a Synapse definition are now implemented for CUDA. + updated docs. + added unittest for mean operation 30 April 2019, 12:20:20 UTC
e77f976 Merge branch 'develop' of https://ai.informatik.tu-chemnitz.de/gogs/annarchy/annarchy into develop 30 April 2019, 11:23:41 UTC
f88f4ea Extended unittests to check global operations used in Synapse definition. 30 April 2019, 11:22:38 UTC
2c25ec1 Specific populations/projections can set 'size_in_bytes' in the _specific_template to overwrite the code generation for the C++ memory measurement. 29 April 2019, 18:29:51 UTC
c35564f Bugfix(CUDA): global operations in Synapse. There was a wrong initialization of array values which could lead to wrong results for MIN/MAX. 29 April 2019, 14:03:18 UTC
feeab2d Bugfix(CUDA): fixed key-error encountered by clear method of pop/proj monitor. Started to implement global operations used in psp on CUDA. 25 April 2019, 15:25:15 UTC
527808f minor changes. + Global.clear() method calls _clear() methods of populations to destroy C++ objects. + Added error messages if global operations used on Synapse() descriptions if used on CUDA devices. 25 April 2019, 09:31:56 UTC
d9f7463 doc: refreshed parallel_run() to be less comfusing. 24 April 2019, 12:25:03 UTC
6240634 Merge branch 'develop' of https://ai.informatik.tu-chemnitz.de/gogs/annarchy/annarchy into develop 24 April 2019, 11:16:03 UTC
7f654aa temporary modification of the convolution extension 24 April 2019, 10:42:26 UTC
c8f7e43 Implemented a monitor _clear() method to destroy C++ monitor instances. 24 April 2019, 08:08:34 UTC
6fc5b58 Added a MemoryStats module. The module can be used to track the memory consumption of C++ objects within the simulation core. This can be especially useful to track the consumption of monitors. 23 April 2019, 17:12:59 UTC
0ecb970 docs: added new figure for the network structure 23 April 2019, 07:47:50 UTC
967fd9c attempt2 for GPU 17 April 2019, 12:27:04 UTC
cc6c634 attempt on GPU 17 April 2019, 12:21:22 UTC
f578bea SpikeSourceArray can now receive projections. 17 April 2019, 12:12:28 UTC
c7048b8 Shared projections need to override also _data() to disable the saving of projection data. 16 April 2019, 11:59:58 UTC
515da05 minor bugfix. 16 April 2019, 11:20:39 UTC
94a9e72 Bugfix: the same error as in commit 16860fa 16 April 2019, 11:06:23 UTC
f7f7b34 Splitted the implementation of weight sharing (SharedProjection) into three classes. This commit extends previous work (commit bd6b1f4d, 7329953) and copied the code for convolution / copy. We still need for the whole module unittests to verify ... 16 April 2019, 08:52:17 UTC
6801b30 Added connector description to codegen.log 16 April 2019, 05:56:49 UTC
16860fa Bugfix: flexible delays and SharedProjection (not fully tested yet) In the last release we allowed the change of synaptic delays outside of the connect method. This was forgotten for SharedProjections. Unittests are not existing yet ... 15 April 2019, 14:33:44 UTC
7efae41 Python3.x compatibility: changed dict.iteritems() to dict.items(). Python2.7 users will not dramatically impaired by this as this dictionary is not that large. 15 April 2019, 14:28:21 UTC
5f8b582 Started version 4.6.8 15 April 2019, 14:01:24 UTC
c52fc5d Bitbucket will only build master 29 March 2019, 11:42:35 UTC
e1604f3 Updated requirements 29 March 2019, 11:26:02 UTC
64df4a7 Give python config when installing 29 March 2019, 11:22:22 UTC
fe9dd03 language level should be an int 29 March 2019, 10:49:34 UTC
ea7adf9 cython should stop emitting numpy warnings when installing 29 March 2019, 10:35:40 UTC
3f48e68 Get rid of cython's futurewarning 29 March 2019, 10:25:50 UTC
2ae10ba forgot subprocess 29 March 2019, 09:55:18 UTC
fc73dd4 python_environment() is partially copied to setup.py, to avoid dependency on ANNarchy before installing ANNarchy. 29 March 2019, 09:53:28 UTC
c9440a6 Updated installation instructions 29 March 2019, 09:33:26 UTC
bf26532 Preparing release 29 March 2019, 08:33:35 UTC
6c21d71 Bugfix: update of delayed spiking container code should be only added if there are synaptic delays. 26 March 2019, 15:53:27 UTC
0a2fa63 small addition to commit 6f28a80 . + reset of the r-vector within reset(). 13 March 2019, 15:50:51 UTC
6f28a80 TimedArray improvement. With the help of Alex we added several new unittests for TimedArray (until now only successful compilation was tested): + Removed doubled dictionary entry. + Removed wrong reinitialization of the firing rate. + Added unittests to verify correctness. 13 March 2019, 15:29:05 UTC
eb50e3b CUDA: added initial guess for kernel config if CudaCheck returns wrong values. 06 March 2019, 14:07:16 UTC
6dc9d1d replaced from cuda_check import by dynamic load of the shared library. 06 March 2019, 13:51:20 UTC
88eed47 removed relative imports of cuda_check (shared library) in CudaCheck python class. 06 March 2019, 07:26:00 UTC
0b496be some work on unittests and refactoring. + replaced some occurences of dendrite.rank and further updated doc + excluded rate transmission with nonuniform delay test for GPUs + test_Connectivity: PEP8 compliance + test_NeuronUpdate: restructured, added boundary test, PEP8 compliance 28 February 2019, 14:18:17 UTC
991530a dendrite.pre_ranks property should be used in future. 27 February 2019, 14:14:13 UTC
9491cc7 doc: non-uniform delays are not slow anymore 27 February 2019, 13:50:22 UTC
cd9e573 Dendrite: the use of dendrite.rank is deprecated, use dendrite.pre_ranks instead. 27 February 2019, 13:48:01 UTC
8804ff4 Updated RateTransmission test to cover also synaptic delays. 27 February 2019, 13:02:15 UTC
8d1c51d Merge branch 'develop' of https://ai.informatik.tu-chemnitz.de/gogs/annarchy/annarchy into develop 26 February 2019, 13:25:07 UTC
7852236 Code improvements in Compiler module. + added check for CUDA CC < 30, as the recent SDKs (>9.0), does only support Kepler and above, we check for the version and hint possible problems. + code adaptations to increase PEP8 compliance (trailing spaces, misplaced spaces etc.) 26 February 2019, 07:42:34 UTC
1d57625 Merge branch 'fast_variable_delays' into develop + minor fix on CUDA side. 25 February 2019, 15:58:51 UTC
e2a1f77 extended unit tests for structural plasticity. 22 February 2019, 16:50:53 UTC
0cc66d2 CUDA at least compiles now, but not implemented 20 February 2019, 13:20:16 UTC
fe6580e Bugfix: parentheses error in openMP template within the last commit (found by unittests) 15 February 2019, 08:25:22 UTC
132f0ca Several bugfixes related to CSR storage format (not completed yet ...) The currently available CSR implementation seems to target only a pre_to_post matrix order, which is obviously problematic. I particularly fixed that, but the CSR implementation is still in an unstable state ... I also disabled the proj._dense mode as it is for LIL format (default) completely without effect and leads to errors in case of CSR format. 15 February 2019, 08:05:47 UTC
b5a65cd CodeGenerator: added helper function. During compile() a log file is created which stores a mapping between network object (popX, projX) to the generated source files. 14 February 2019, 17:02:21 UTC
0701308 Merge branch 'develop' into fast_variable_delays 13 February 2019, 09:51:44 UTC
e4b9c8c Population._load_pop_data: exception was not catched 13 February 2019, 09:50:50 UTC
26d7a32 Merge branch 'develop' of https://ai.informatik.tu-chemnitz.de/gogs/annarchy/annarchy into develop 13 February 2019, 09:49:44 UTC
31af92a Added comments to templates 13 February 2019, 09:47:49 UTC
10578f0 Global functions added by add_function() were not available on CUDA devices. 06 February 2019, 14:51:54 UTC
7ea6301 Bugfixes and modifications for save/load of projections. Bugfixes: + IO._load_data(): the data structure returned by np.load().files contained np.ndarray instead of dictionaries (py3, numpy 1.14.0) + Projection.save()/load(): in case of randomly drawn synaptic delays the delays are now stored. Enhancement: + added warning if connect_XXX is called twice. + added error message if create/pruning are used in Synapse() definition without enabling structural plasticity. + added save() method to Network class. 06 February 2019, 10:01:57 UTC
ac3ed1d Merge branch 'develop' into fast_variable_delays 05 February 2019, 13:01:16 UTC
eb55117 delays are now saved and loaded 05 February 2019, 13:00:43 UTC
06e573e Merge branch 'develop' of https://ai.informatik.tu-chemnitz.de/gogs/annarchy/annarchy into develop 05 February 2019, 11:08:47 UTC
39787e6 Bugfix: if structural plasticity is enabled and uniform delays are used the implementation of removeSynapse() was not accounting all variables. 17 January 2019, 16:25:17 UTC
8392758 implemented ring buffer for spiking networks using variable delays TODO: CUDA interface broken 24 December 2018, 15:13:00 UTC
080e9dd first attempt to speedup spiking networks with variable delays. Speedup up impressive on COBA (1s instead of 50s...), but have to check validness 22 December 2018, 12:39:15 UTC
125ba59 added the possibility to change the delays after compilation. 22 December 2018, 08:39:16 UTC
d704303 Merge branch 'late_instantiation' into develop 21 December 2018, 11:43:50 UTC
b46ff98 doc: small typos 20 December 2018, 07:25:13 UTC
7819589 Bugfix: attribute error in _load_proj_data() 08 October 2018, 07:40:36 UTC
back to top