Skip to content

Latest commit

 

History

History
400 lines (202 loc) · 13.9 KB

CHANGES.md

File metadata and controls

400 lines (202 loc) · 13.9 KB

Version 0.6.0

  • Added option to shape conn weights dynamically to create temporal patterns (issue #33)

  • Store all params of synMechs exhaustively instead of by reference (issue #139)

  • Added netParams.importCellParamsFromNet() to import parameters of multiple cells from existing network (issue #154)

  • Added modifySynMechs function

  • Added option to record from all synMechs of a type (eg. 'AMPA')

  • Fixed bugs and improved efficiency of modify, modifyConns and modifyStims function

  • Fixed bug plotting traces

Version 0.5.9

  • Improved NeuroML2 import functions (issue #12)

  • Added support for homeostatic scaling synapse (requires self NetCon) (issue #141)

  • Added method to convert NetParams and SimConfig objects to dicts (issue #141)

  • Fixed NetParams and SimConfig constructors so arguments are converted to Dict/ODict (issue #141)

  • Added sim.loadBalance() func to calculate and print load balance information (issue #77)

Version 0.5.8

  • Fixed bug (introduced during debugging) that prevented random seeds from working properly

Version 0.5.7

  • Improved NeuroML import/export functions

  • Added option for cache_efficient (optimized load balance when many cores) (issue #77)

  • Added sim.clearAll() function to clear all objects in memory and avoid leaks

  • Cleared variables after gathering from multiple nodes to avoid memory leak

  • Separated common runSim() commands into preRun() (issue #83)

  • Fixed bugs in net.modifyStims() method

Version 0.5.6

  • Only load neuroml libraries and functions if installed to avoid error when importing sim

  • Disable subcellular connectivity since wasn't modifying the synMechs correctly.

Version 0.5.5

  • New example using Allen Institute cells showing how to save large net and reload for simulation

  • Improved export/import from/to NeuroML2 (issue #12)

  • All structures (including Dict and ODict) can now be saved and loaded via json

  • Added option to use separate simConfig when loading data from file

  • saveData() calls gatherData() if required

  • saveData() returns string with full path of saved file (issue #127)

  • Fixed bugs so its possible to create and save Python structure independently of NEURON objects

  • Fixed bug that created 2 instead of 1 NetStims per cell

  • Fixed bug regarding mutable default args in Python funcs

  • Fixed bug so NetStim stim params are set to default if missing

Version 0.5.4

  • Access to nested dictionaries via dot notation (secs.soma.geom) using new inherited classes Dict() and ODict() (issue #123)

  • 'hSection' replaced with 'hSec'

  • Added sim.popAvgRates() to calculate and optionally print population avg rates

  • Functions to modify cells, conns or stims also update sim.allCells (issue #59)

  • Fixed bug so can modify funcs now accept lists of strings (issue #59)

  • Fixed bug when recording traces of relative cell indices of populations across multiple nodes

Version 0.5.3

  • Functions to modify parameters of cells, conns and stims in an instantiated network (issue #59)

  • Added label of param rule used to create cells, conns and stims, so can modify later (issue #59)

  • Ordered sim.net.allCells by gid

  • Option to save figures with same name as data

  • Added support for VClamp list params (issue #32)

  • Fixed bug when creating pops consisiting of lists of cells

Version 0.5.2

  • Fixed bug that gave error since trying to import removed module default.py

Version 0.5.1

  • Easier and more consistent format for high-level specifications (issue #103):

    • replaced default.py with specs.py which defines classes NetParams and SimConfig

    • netParams and simConfig are now objects of class NetParams and SimConfig, respectively

    • Can use object methods to add params, eg. netParams.addPopParams(label, params)

    • All structures within netParams and simConfig are now OrderedDicts so can be referenced by label/key

    • If no label is assigned, then an increasing number is automatically used as label

    • Split netParams.stimParams into netParams.stimSourceParams and netParams.stimTargetParams (both OrderedDicts)

    • Simplified/clarified keys: 'conditions' -> 'conds', 'sections' -> 'secs', 'preTags' -> 'preConds', 'postTags' -> 'postConds'

    • Added netParams.importCellParams() to simplify

  • Added sim wrappers and modified names of existing ones to make consistent

  • Selectively save data to file: netParams, net, simConfig and/or simData (issue #30)

  • Added functions to load netParams, net, simConfig and/or simData from file (issue #30)

  • Added 'rerun' option to plotTraces()

  • Added function to get cell object by gid, sim.cellByGid(gid)

  • Fixed bug that was slowing down STDP by creating new h.Section() -- now uses existing.

  • Fixed bug so vinit gets set for all segments in a section.

  • Fixed bug related to readArgs() when running from IPYnotebook.

  • Fixed bug related to plasticity in cells with V not in section (eg. Izhi2007a)

Version 0.5.0

  • Added option 'dataSaveInclude' to select what data to save to file (issue #30)

  • Added sim.net.allPops which contains all cellGids of each when running on >1 nodes (issue #30)

  • Connectivity matrix can now be grouped by numeric tags in intervals (eg. cortical depth y in 50 um intervals) (issue #92)

  • Added support for all stims (VClamp, SEClamp, AlphaSynapse, etc) and can specify any of the stim params (amp, dur, delay,etc) as a function (eg. 'uniform(a,b)' or '2*post_ynorm') (issue #32)

  • Fixed bugs so plotRaster() is identical when running on >1 nodes, and is ordered by gid correctly

  • Fixed bug so can have multiple plot function calls even when one of them is False

Version 0.4.9

  • Modified format of simConfig analysis options to make it equivalent to calling analysis functions (issue #69)

  • Improved plotRaster() function to add customizable options (select cells, time range, order, overlay histogram, and others) (issue #39)

  • Improved plotTraces() function to add customizable options (select cells, time range, arrage by cell or trace, and others) (issue #39)

  • Improved plot2Dnet() function to add customizable options (select cells, show connections, and others) (issue #39)

  • Added spikeHist() function to plot spike histogram with customizable options (select cells, time range, bin size, and others) (issue #63)

  • Added plotConn() function to plot spike histogram with multiple customizable options (select cells, feature, order, and others) (issue #39)

  • Added option to save figure, save figure data, specify figure size, and show figure to all analysis/plotting functions (issue #39)

  • Fixed bug when generating function-based random probability values -- made random stream independent

  • Fixed bug when generating density-based random cell locations -- made random stream independent (issue #93)

Version 0.4.8

  • Removed framework and init modules, and consolidated in sim module (so just need: 'from netpyne import sim')

  • Added option to specify weight scale factor separately for each cell model, and for NetStims (issue #69)

  • Conn rules can now have list of synMechs (eg. [AMPA, NMDA]) and synMechWeightFactor (eg. [1.0, 0.1]) (issue #69)

  • Conn rules allow list of weights, delays and/or locs for each synMech in list (issue #69)

  • Conn rules allow synsPerConn and loc to be described functionally (at the cell connection level) (eg. 'uniform(5,2)') (issue #69)

  • Conn rules allow list of weights, delays, and/or locs when synsPerConn > 1 (issue #69)

  • Conn rules allow 2D list of weights, delays, and/or locs when have list of synMechs and synsPerConn > 1 (issue #69)

  • Conn rules allow list of sections or sectionList when synsPerConn > 1; synMechs distributed uniformly (loc list not allowed) (issue #69)

  • Extended fromList connectivity function so can also provide synapse locs (issue #69)

  • Added separate wrapper sim functions to create and simulate the network

  • Sim functions use simConfig and netParams from main as default (if not specified)

  • Connections between NetStims and cells are included in conn list (issue #69)

  • Fixed bug so can create Python and NEURON objects for connections independently (issue #69)

Version 0.4.7

  • Added option for random seeds for connectivity, stimulation, and cell locations (issue #49)

  • Added return pointers when create cells, pops, conns, stims, and recording (issue #48)

  • Renamed point process param labels: '_type' -> 'mod', '_loc' -> 'loc' (issue #24)

  • Added option to set h global variables (eg. celsius) (issue #31)

  • Fixed importCell() so h global variables reset after importing (issue #31)

  • Fixed importCell() so synapses stored in synMechParams (issue #25)

  • Fixed bug when pop names contained same subset of characters (issue #40)

Version 0.4.6

  • Preliminary version of exporter to NeuroML2

  • Added fromList connectivity function

  • Added new dict 'stimParams' with support for IClamps

  • Added 'start' param to NetStim populations

  • Modified izhi2007 'u' starting value to be 0 instead of 0.2

  • Modified izhi2007 'C' value and the section properties so that synaptic weights match HH

  • Renamed NMDA synapses to AMPA

  • Fixed bugs in tuts

Version 0.4.5

  • Fixed bug saving to JSON on single node

  • Fixed plotting traces from using 'all' cells option

Version 0.4.4

  • Fixed bug when plottinc synMech traces

  • Renamed 'pos' with 'loc' in recordTraces list

Version 0.4.3

  • Fixed bug in runSimWithIntervalFunc

  • Variable Netstims (NSLOCs) can now have noise > 0

Version 0.4.2

  • Placed show(block=False) in try except block since not supported by some graphic backends

  • Create parallel context before re-creating net to avoid seg fault

  • Removed unnecessary modules in analysis.py

  • Only call show() in analysis if some figure to show

Version 0.4.1

  • Fixed bug when distributing cells spatially based on xRange,yRange,zRange pop parameters

  • Added axis labels to 2D visualization, and now works when running on multiple nodes too

  • Reset Netstim random generators within runSim() so have reproducible results if working interactively

Version 0.4.0

  • Fixed bug when distributing cells spatially based on normRange pop parameter

  • Fixed bug in functional connectivity variables post_xnorm, post_ynorm, post_znorm

  • Reversed ynorm based raster so higher y values are shown at the bottom (cortical-like)

  • Added option to plot sync lines in raster and show sync measure

  • Added 2D visualization of network cells and conns

  • Fixed randomization of cell positions by adding lastGid

Version 0.3.9

  • Made synMech params independent of cell and referenced by labels (similar to NeuroML)

  • Keep Neuron objects after sim so can explore, modify and/or rerun sim.

  • Only gather data via py_alltoall if running on more than 1 node

Version 0.3.8

  • Fix recording of single cell of population when using MPI

  • Fixed raster plotting based on NCD when using MPI

  • Replaced save as .txt format with save as .csv

  • Fixed bug when importing distributed mechanisms of cells

  • importCell can use either *args or **kwargs

  • Fixed bug in secLists implementation

  • Removed mpl_toolkits.mplot3d import (unused and produced error in some Mac OS versions)

Version 0.3.7

  • Made conn functions more efficient using gid2lid and lid2gid

  • Replaced 'syn' (synapse) with 'synMech' (synaptic mechanism) to avoid confusion with synaptic connections

Version 0.3.6

  • Fixed bug: STDP objects need to be stored so it works.

  • Added support for SectionLists (modified format of importCell so also works)

  • Fixed bugs: function-based connectivity

Version 0.3.5

  • Fixed bugs: not checking connectivity rule conditions properly

  • Fixed bug: number of connections depended on number of nodes

Version 0.3.4

  • Added option to add STDP plasticity and RL to connections

  • Added option to run function at intervals during simulation, e.g. to interface with external program (such as virtual arm)

Version 0.3.3

  • Moved plot legend outside of plot area

  • Changed order of raster population colors to make separation clearer

  • Added option to select what cells to record/plot from separately; using new format with cells/pops in single list.

Version 0.3.2

  • Fixed bug: convergence connectivity produced error if used numeric value

Version 0.3.1

  • Added option to show and/or save to file the timing of initialization, cell creation, connection creation, setup recording, simulation run, data gathering, plotting, and saving.

  • Fixed bug: h.dt now set to value of simConfig.dt

Version 0.3

First version that was uploaded to pypi. Includes following features:

  • Clear separation (modularization) of parameter specifications, network instantiation and NEURON simulation code.

  • Easy-to-use, standardized, flexible, extensible and NEURON-independent format to specify parameters:

    • Populations
    • Cell property rules
    • Connectivity rules
    • Simulation configuration

  • Support for cell location (eg. cortical depth) dependence of cell density and connectivity.

  • Easy specification, importing and swapping of cell models (eg. point neuron vs multicompartment)

  • Support for hybrid networks eg. combining point and multicompartment neurons.

  • Multiple connectivity functions (eg. full, convergent, probabilistic) with optional parameters (eg. delay range)

  • Support for user-defined connectivity functions.

  • Populations, cell properties and connectivity rules can include reference to annotations (eg. for provenance).

  • NEURON-independent instantiation of network (all cells, connections, ...) using Python objects and containers.

  • NEURON-specific instantiation of network ready for simulation.

  • Enables sharing of Python-based network objects, which can then be instantiated and simulated in NEURON.

  • Easy MPI parallel simulation of network, including cell distribution across nodes an gathering of data from all nodes.

  • Analysis and visualization of network (eg. connecitivity matrix) and simulation output (eg. voltage traces, raster plot)

  • Data exporting/sharing to several formats (pickle, Matlab, JSON, HDF5) of the following:

    • Parameters/specifications
    • Instantiated networks
    • Simulation results