def run_script(cell_range): # path_to_data = "/Users/stevecharczynski/workspace/data/warden/recall_trials/" # save_dir = "/Users/stevecharczynski/workspace/data/warden/recall_trials/" save_dir = "/projectnb/ecog-eeg/stevechar/ml_runs/warden/recall_trials_second_main_time/" path_to_data = "/projectnb/ecog-eeg/stevechar/data/warden/recall_trials/" # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) #second_pres data_processor = analysis.DataProcessor( path_to_data, cell_range, window=[1250, 3000]) solver_params = { "niter": 1500, "stepsize": 100, "interval": 20, "method": "TNC", "use_jac": True, "T" : 100, "disp":False } bounds_stim = { "sigma": [1e-4, 1000.], "mu": [1250, 3000.], "a_1": [1e-10, 1/5.], "a_2": [1e-10, 1/5.], "a_3": [1e-10, 1/5.], "a_4": [1e-10, 1/5.], "a_0": [1e-10, 1/5.] } bounds_time = { "sigma": [1e-4, 1000.], "mu": [1250, 3000.], "a_1": [1e-10, 1/2.], "a_0": [1e-10, 1/2.] } pipeline = analysis.Pipeline(cell_range, data_processor, [ "Const","Gaussian", "GaussianStim"], save_dir=save_dir) pipeline.set_model_bounds("Gaussian", bounds_time) pipeline.set_model_bounds("GaussianStim", bounds_stim) pipeline.set_model_bounds("Const", {"a_0":[10e-10, 1]}) # with open("/Users/stevecharczynski/workspace/data/warden/recall_trials/info.json") as f: with open("/projectnb/ecog-eeg/stevechar/data/warden/recall_trials/info.json") as f: stims = json.load(f) stims = {int(k):v for k,v in stims.items()} pipeline.set_model_info("GaussianStim", "stim_identity", stims, per_cell=True) pipeline.set_model_x0("GaussianStim", [10, 1600, 1e-1, 1e-1,1e-1, 1e-1, 1e-1]) pipeline.set_model_x0("Gaussian", [10, 1600, 1e-1, 1e-1]) pipeline.set_model_x0("Const", [1e-1]) pipeline.fit_all_models(solver_params=solver_params) pipeline.fit_even_odd(solver_params=solver_params) # pipeline.compare_even_odd("Const", "SigmaMuTau", 0.01) # pipeline.compare_even_odd("SigmaMuTau", "SigmaMuTauStimWarden", 0.01) pipeline.compare_models("Const", "Gaussian", 0.01, smoother_value=100) pipeline.compare_models("Const", "GaussianStim", 0.01, smoother_value=100) pipeline.compare_models("Gaussian", "GaussianStim", 0.01, smoother_value=100) pipeline.compare_even_odd("Const", "Gaussian", 0.01) pipeline.compare_even_odd("Const", "GaussianStim", 0.01) pipeline.compare_even_odd("Gaussian", "GaussianStim", 0.01)
def run_script(cell_range): path_to_data = "/Users/stevecharczynski/workspace/data/sheehan/lin_pos_set/s11" data_processor = analysis.DataProcessor(path_to_data, cell_range) n_t = 2. solver_params = { "niter": 300, "stepsize": 500, "interval": 10, "method": "TNC", "use_jac": True, "T": 1, "disp": False } bounds_pos = { "a_1": [10**-10, 1 / n_t], "ut": [0., 200.], "st": [10., 2000.], "a_0": [10**-10, 1 / n_t] } pipeline = analysis.Pipeline( cell_range, data_processor, ["ConstVariable", "RelPosVariable", "AbsPosVariable"]) # pipeline.set_model_bounds("TimeVariableLength", bounds_t) pipeline.set_model_bounds("AbsPosVariable", bounds_pos) pipeline.set_model_bounds("RelPosVariable", bounds_pos) pipeline.set_model_bounds("ConstVariable", {"a_0": [10**-10, 1]}) pipeline.set_model_x0("AbsPosVariable", [1e-5, 20, 10, 1e-5]) pipeline.set_model_x0("RelPosVariable", [1e-5, 20, 10, 1e-5]) pipeline.set_model_x0("ConstVariable", [1e-5]) # pipeline.show_rasters() import numpy as np import json with open(path_to_data + "/abs_pos.json", 'r') as f: abs_pos = np.array(json.load(f)) with open(path_to_data + "/rel_pos.json", 'r') as f: rel_pos = np.array(json.load(f)) with open(path_to_data + "/velocity.json", 'r') as f: velocity = np.array(json.load(f)) pipeline.set_model_info("AbsPosVariable", "abs_pos", abs_pos, True) pipeline.set_model_info("RelPosVariable", "rel_pos", rel_pos, True) pipeline.set_model_info("AbsPosVariable", "velocity", velocity, True) pipeline.set_model_info("RelPosVariable", "velocity", velocity, True) pipeline.fit_all_models(solver_params=solver_params) pipeline.compare_models("ConstVariable", "RelPosVariable", 0.01, smoother_value=100) pipeline.compare_models("ConstVariable", "AbsPosVariable", 0.01, smoother_value=100)
import maxlikespy.analysis as analysis path_to_data = '/Users/stevecharczynski/workspace/data/jay/2nd_file' cell_range = range(5, 10) data_processor = analysis.DataProcessor(path_to_data, cell_range) solver_params = { "niter": 2, "stepsize": 10000, "interval": 10, "method": "TNC", "use_jac": True, "T": 1, "disp": False } bounds_t = { "a_1": [0, 1 / 2], "ut": [-1000, 10000], "st": [100, 10000], "a_0": [10**-10, 1 / 2] } bounds_c = {"a_0": [10**-10, 1 / 2]} pipeline = analysis.Pipeline(cell_range, data_processor, ["Const", "Time"], 0) pipeline.set_model_bounds("Time", bounds_t) pipeline.set_model_bounds("Const", bounds_c) pipeline.set_model_x0(["Const", "Time"], [[1e-5], [1e-5, 100, 100, 1e-5]]) pipeline.fit_all_models(solver_params=solver_params) pipeline.compare_models("Const", "Time", 0.01)
def run_script(cell_range): # path_to_data = "/Users/stevecharczynski/workspace/data/rossi_pool/a2/" # save_dir = "/Users/stevecharczynski/workspace/data/rossi_pool/a2/" save_dir = "/projectnb/ecog-eeg/stevechar/ml_runs/rossi_pool/a2/" path_to_data = "/projectnb/ecog-eeg/stevechar/data/rossi_pool/a2/" # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) data_processor = analysis.DataProcessor(path_to_data, cell_range, window=[0, 3000]) solver_params = { "niter": 500, "stepsize": 100, "interval": 10, "method": "TNC", "use_jac": True, "T": 1, "disp": False } bounds_smtstim = { "sigma": [0, 1000.], "mu": [0, 1000.], "tau": [20, 10000.], "a_1": [10**-10, 1 / 3.], "a_2": [10**-10, 1 / 3.], "a_0": [10**-10, 1 / 3.] } bounds_smtclass = { "sigma": [0, 1000.], "mu": [0, 1000.], "tau": [20, 10000.], "a_1": [10**-10, 1 / 5.], "a_2": [10**-10, 1 / 5.], "a_3": [10**-10, 1 / 5.], "a_4": [10**-10, 1 / 5.], "a_0": [10**-10, 1 / 5.] } bounds_smt = { "sigma": [0, 1000.], "mu": [0, 1000.], "tau": [20, 10000.], "a_1": [10**-10, 1 / 2.], "a_0": [10**-10, 1 / 2.] } pipeline = analysis.Pipeline( cell_range, data_processor, ["Const", "SigmaMuTau", "SigmaMuTauStimRP", "SigmaMuTauStimClassRP"], save_dir=save_dir) pipeline.set_model_bounds("SigmaMuTau", bounds_smt) pipeline.set_model_bounds("SigmaMuTauStimRP", bounds_smtstim) pipeline.set_model_bounds("SigmaMuTauStimClassRP", bounds_smtclass) pipeline.set_model_bounds("Const", {"a_0": [10e-10, 1]}) # with open("/Users/stevecharczynski/workspace/data/rossi_pool/a2/info.json") as f: with open( "/projectnb/ecog-eeg/stevechar/data/rossi_pool/a2/info.json") as f: stims = json.load(f) stims = {int(k): v for k, v in stims.items()} pipeline.set_model_info("SigmaMuTauStimRP", "stim_identity", stims, per_cell=True) pipeline.set_model_info("SigmaMuTauStimClassRP", "stim_identity", stims, per_cell=True) pipeline.set_model_x0("SigmaMuTauStimRP", [0.01, 300, 10, 1e-1, 1e-1, 1e-1]) pipeline.set_model_x0("SigmaMuTauStimClassRP", [0.01, 300, 10, 1e-1, 1e-1, 1e-1, 1e-1, 1e-1]) pipeline.set_model_x0("SigmaMuTau", [0.01, 300, 10, 1e-1, 1e-1]) pipeline.set_model_x0("Const", [1e-1]) # pipeline.show_rasters() pipeline.fit_all_models(solver_params=solver_params) pipeline.fit_even_odd(solver_params=solver_params) # pipeline.compare_even_odd("Const", "SigmaMuTau", 0.01) # pipeline.compare_even_odd("SigmaMuTau", "SigmaMuTauStimRP", 0.01) pipeline.compare_models("Const", "SigmaMuTau", 0.01, smoother_value=100) pipeline.compare_models("SigmaMuTau", "SigmaMuTauStimRP", 0.01, smoother_value=100) pipeline.compare_models("SigmaMuTau", "SigmaMuTauStimClassRP", 0.01, smoother_value=100) pipeline.compare_models("SigmaMuTauStimRP", "SigmaMuTauStimClassRP", 0.01, smoother_value=100) pipeline.compare_even_odd("Const", "SigmaMuTau", 0.01) pipeline.compare_even_odd("SigmaMuTau", "SigmaMuTauStimRP", 0.05) pipeline.compare_even_odd("SigmaMuTau", "SigmaMuTauStimClassRP", 0.05) pipeline.compare_even_odd("SigmaMuTauStimRP", "SigmaMuTauStimClassRP", 0.05)
def run_script(cell_range): # save_dir = "/projectnb/ecog-eeg/stevechar/sheehan_runs/stable_lights_on_only/inbound/" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/stable_lights_on_only/inbound/" # data_processor = analysis.DataProcessor( # path_to_data, cell_range) # n_t = 2. # solver_params = { # "niter": 500, # "stepsize": 500, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # bounds_dual = { # "ut_a": [-10., 100.], # "st_a": [0.1, 90.], # "a_0a": [10**-10, 1 / 3], # "a_1a": [10**-10, 1 / 3], # "ut_b": [-10., 100.], # "st_b": [0.1, 90.], # "a_0b": [10**-10, 1 / 3], # "a_1b": [10**-10, 1 / 3], # } # bounds_norm = { # "a_1": [10**-10, 1 / n_t], # "ut": [-10., 100.], # "st": [0.1, 90.], # "a_0": [10**-10, 1 / n_t] # } # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # # "ConstVariable", "RelPosVariable","DualPeakedRel", "AbsPosVariable"]) # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "ConstVariable", "RelPosVariable", "AbsPosVariable", "DualPeakedRel", "DualPeakedAbs"], save_dir=save_dir) # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable", "AbsPosVariable"], save_dir=save_dir) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("AbsPosVariable", bounds_norm) # pipeline.set_model_bounds("RelPosVariable", bounds_norm) # pipeline.set_model_bounds("ConstVariable", {"a_0":[10**-10, 1]}) # pipeline.set_model_bounds("DualPeakedRel", bounds_dual) # pipeline.set_model_bounds("DualPeakedAbs", bounds_dual) # pipeline.set_model_x0("DualPeakedRel", [20, 1, 1e-5, 1e-5, 20, 1, 1e-5, 1e-5]) # pipeline.set_model_x0("DualPeakedAbs", [20, 1, 1e-5, 1e-5, 20, 1, 1e-5, 1e-5]) # pipeline.set_model_x0("AbsPosVariable", [1e-5, 20, 1, 1e-5]) # pipeline.set_model_x0("RelPosVariable", [1e-5, 20, 1, 1e-5]) # pipeline.set_model_x0("ConstVariable", [1e-5]) # # pipeline.show_rasters() # with open(path_to_data+"/abs_pos.json", 'r') as f: # abs_pos = np.array(json.load(f)) # with open(path_to_data+"/rel_pos.json", 'r') as f: # rel_pos = np.array(json.load(f)) # pipeline.set_model_info("AbsPosVariable", "abs_pos", abs_pos, True) # pipeline.set_model_info("RelPosVariable", "rel_pos", rel_pos, True) # pipeline.set_model_info("DualPeakedRel", "rel_pos", rel_pos, True) # pipeline.set_model_info("DualPeakedAbs", "abs_pos", abs_pos, True) # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_even_odd("RelPosVariable", "DualPeakedRel", 0.01) # pipeline.compare_even_odd("AbsPosVariable", "DualPeakedAbs", 0.01) # pipeline.compare_even_odd("ConstVariable", "RelPosVariable", 0.01) # pipeline.compare_even_odd("ConstVariable", "AbsPosVariable", 0.01) # pipeline.compare_models("AbsPosVariable", "DualPeakedAbs", 0.01, smoother_value=100) # pipeline.compare_models("RelPosVariable", "DualPeakedRel", 0.01, smoother_value=100) # pipeline.compare_models("ConstVariable", "RelPosVariable", 0.01, smoother_value=100) # pipeline.compare_models("ConstVariable", "AbsPosVariable", 0.01, smoother_value=100) save_dir = "/projectnb/ecog-eeg/stevechar/sheehan_runs/stable_lights_on_only/inbound/" path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/stable_lights_on_only/inbound/" data_processor = analysis.DataProcessor( path_to_data, cell_range) n_t = 2. solver_params = { "niter": 1000, "stepsize": 50, "interval": 10, "method": "TNC", "use_jac": True, "T" : 1, "disp":False } bounds_dual = { "ut_a": [-10., 100.], "st_a": [0.1, 90.], "a_0a": [10**-10, 1 / 3], "a_1a": [10**-10, 1 / 3], "ut_b": [-10., 100.], "st_b": [0.1, 90.], "a_0b": [10**-10, 1 / 3], "a_1b": [10**-10, 1 / 3], } bounds_norm = { "a_1": [10**-10, 1 / n_t], "ut": [-10., 100.], "st": [0.1, 90.], "a_0": [10**-10, 1 / n_t] } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable","DualPeakedRel", "AbsPosVariable"]) pipeline = analysis.Pipeline(cell_range, data_processor, [ "ConstVariable", "RelPosVariable", "AbsPosVariable"], save_dir=save_dir) # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "ConstVariable", "RelPosVariable", "AbsPosVariable"], save_dir=save_dir) # pipeline.set_model_bounds("TimeVariableLength", bounds_t) pipeline.set_model_bounds("AbsPosVariable", bounds_norm) pipeline.set_model_bounds("RelPosVariable", bounds_norm) pipeline.set_model_bounds("ConstVariable", {"a_0":[10**-10, 1]}) pipeline.set_model_x0("AbsPosVariable", [1e-5, 20, 1, 1e-5]) pipeline.set_model_x0("RelPosVariable", [1e-5, 20, 1, 1e-5]) pipeline.set_model_x0("ConstVariable", [1e-5]) # pipeline.show_rasters() with open(path_to_data+"/abs_pos.json", 'r') as f: abs_pos = np.array(json.load(f)) with open(path_to_data+"/rel_pos.json", 'r') as f: rel_pos = np.array(json.load(f)) pipeline.set_model_info("AbsPosVariable", "abs_pos", abs_pos, True) pipeline.set_model_info("RelPosVariable", "rel_pos", rel_pos, True) pipeline.fit_even_odd(solver_params=solver_params) pipeline.fit_all_models(solver_params=solver_params) pipeline.compare_even_odd("ConstVariable", "RelPosVariable", 0.01) pipeline.compare_even_odd("ConstVariable", "AbsPosVariable", 0.01) pipeline.compare_models("ConstVariable", "RelPosVariable", 0.01, smoother_value=100) pipeline.compare_models("ConstVariable", "AbsPosVariable", 0.01, smoother_value=100)
def run_script(cell_range): # path_to_data = "/Users/stevecharczynski/workspace/data/warden/recall_trials/" # save_dir = "/Users/stevecharczynski/workspace/data/warden/recall_trials_both/" save_dir = "/projectnb/ecog-eeg/stevechar/ml_runs/warden/recall_trials_both/" path_to_data = "/projectnb/ecog-eeg/stevechar/data/warden/recall_trials/" # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) data_processor = analysis.DataProcessor(path_to_data, cell_range, window=[0, 3000]) solver_params = { "niter": 1000, "stepsize": 100, "interval": 5, "method": "TNC", "use_jac": True, "T": 1, "disp": False } bounds_gaussianstim = { "sigma1": [1e-5, 1000.], "mu1": [0, 1700.], "sigma2": [1e-5, 1000.], "mu2": [1500, 3200.], "a_1": [1e-10, 1 / 5.], "a_2": [1e-10, 1 / 5.], "a_3": [1e-10, 1 / 5.], "a_4": [1e-10, 1 / 5.], "a_0": [1e-10, 1 / 5.] } bounds_gaussianstim_pos = { "sigma1": [1e-5, 1000.], "mu1": [0, 1700.], "sigma2": [1e-5, 1000.], "mu2": [1500, 3200.], "a_1": [1e-10, 1 / 5.], "a_2": [1e-10, 1 / 5.], "a_3": [1e-10, 1 / 5.], "a_4": [1e-10, 1 / 5.], "a_5": [1e-10, 1 / 5.], "a_6": [1e-10, 1 / 5.], "a_7": [1e-10, 1 / 5.], "a_8": [1e-10, 1 / 5.], "a_0": [1e-10, 1 / 5.] } bounds_gaussian = { "sigma1": [1e-5, 1000.], "mu1": [0, 1700.], "sigma2": [1e-5, 1000.], "mu2": [1500, 3200.], "a_1": [1e-10, 1 / 2.], "a_0": [1e-10, 1 / 2.] } bounds_gaussian_pos = { "sigma1": [1e-5, 1000.], "mu1": [0, 1700.], "sigma2": [1e-5, 1000.], "mu2": [1500, 3200.], "a_1": [1e-10, 1 / 2.], "a_2": [1e-10, 1 / 2.], "a_0": [1e-10, 1 / 2.] } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Const","GaussianBoth", "GaussianStimBoth", "GaussianStimBoth1", \ # "GaussianStimBoth2", "GaussianStimBoth3", "GaussianStimBoth4"], save_dir=save_dir) pipeline = analysis.Pipeline( cell_range, data_processor, [ "Const", "GaussianBoth", "GaussianStimBoth", "GaussianBothPos", "GaussianStimBothPos" ], save_dir=save_dir) pipeline.set_model_bounds("GaussianBoth", bounds_gaussian) # pipeline.set_model_bounds("GaussianStimBoth1", bounds_gaussian) # pipeline.set_model_bounds("GaussianStimBoth2", bounds_gaussian) # pipeline.set_model_bounds("GaussianStimBoth3", bounds_gaussian) # pipeline.set_model_bounds("GaussianStimBoth4", bounds_gaussian) pipeline.set_model_bounds("GaussianBothPos", bounds_gaussian_pos) pipeline.set_model_bounds("GaussianStimBothPos", bounds_gaussianstim_pos) pipeline.set_model_bounds("GaussianStimBoth", bounds_gaussianstim) pipeline.set_model_bounds("Const", {"a_0": [10e-10, 1]}) # with open("/Users/stevecharczynski/workspace/data/warden/recall_trials/info.json") as f: with open( "/projectnb/ecog-eeg/stevechar/data/warden/recall_trials/info.json" ) as f: stims = json.load(f) stims = {int(k): v for k, v in stims.items()} pipeline.set_model_info("GaussianStimBoth", "stim_identity", stims, per_cell=True) pipeline.set_model_info("GaussianStimBothPos", "stim_identity", stims, per_cell=True) # pipeline.set_model_info("GaussianStimBoth1", "stim_identity", stims, per_cell=True) # pipeline.set_model_info("GaussianStimBoth2", "stim_identity", stims, per_cell=True) # pipeline.set_model_info("GaussianStimBoth3", "stim_identity", stims, per_cell=True) # pipeline.set_model_info("GaussianStimBoth4", "stim_identity", stims, per_cell=True) pipeline.set_model_x0("GaussianStimBoth", [10, 1000, 10, 2000, 1e-1, 1e-1, 1e-1, 1e-1, 1e-1]) pipeline.set_model_x0("GaussianStimBothPos", [ 10, 1000, 10, 2000, 1e-1, 1e-1, 1e-1, 1e-1, 1e-1, 1e-1, 1e-1, 1e-1, 1e-1 ]) pipeline.set_model_x0("GaussianBothPos", [10, 1000, 10, 2000, 1e-1, 1e-1, 1e-1]) pipeline.set_model_x0("GaussianBoth", [10, 1000, 10, 2000, 1e-1, 1e-1]) # pipeline.set_model_x0("GaussianStimBoth1", [10, 1000,10, 2000, 1e-1, 1e-1]) # pipeline.set_model_x0("GaussianStimBoth2", [10, 1000,10, 2000, 1e-1, 1e-1]) # pipeline.set_model_x0("GaussianStimBoth3", [10, 1000,10, 2000, 1e-1, 1e-1]) # pipeline.set_model_x0("GaussianStimBoth4", [10, 1000,10, 2000, 1e-1, 1e-1]) pipeline.set_model_x0("Const", [1e-1]) pipeline.fit_all_models(solver_params=solver_params) pipeline.fit_even_odd(solver_params=solver_params) # pipeline.compare_even_odd("Const", "GaussianBoth", 0.01) # pipeline.compare_even_odd("GaussianBoth", "GaussianStimBoth", 0.01) pipeline.compare_models("Const", "GaussianBoth", 0.01, smoother_value=100) pipeline.compare_models("Const", "GaussianStimBoth", 0.01, smoother_value=100) pipeline.compare_models("GaussianBoth", "GaussianStimBoth", 0.01, smoother_value=100) pipeline.compare_models("Const", "GaussianBothPos", 0.01, smoother_value=100) pipeline.compare_models("Const", "GaussianStimBothPos", 0.01, smoother_value=100) pipeline.compare_models("GaussianBoth", "GaussianBothPos", 0.01, smoother_value=100) pipeline.compare_models("GaussianBoth", "GaussianStimBothPos", 0.01, smoother_value=100) pipeline.compare_models("GaussianBothPos", "GaussianStimBoth", 0.01, smoother_value=100) pipeline.compare_models("GaussianBothPos", "GaussianStimBothPos", 0.01, smoother_value=100) pipeline.compare_models("GaussianStimBoth", "GaussianStimBothPos", 0.01, smoother_value=100) # pipeline.compare_models("Const", "GaussianStimBoth1", 0.01, smoother_value=100) # pipeline.compare_models("Const", "GaussianStimBoth2", 0.01, smoother_value=100) # pipeline.compare_models("Const", "GaussianStimBoth3", 0.01, smoother_value=100) # pipeline.compare_models("Const", "GaussianStimBoth4", 0.01, smoother_value=100) # pipeline.compare_models("GaussianStimBoth1", "GaussianStimBoth", 0.01, smoother_value=100) # pipeline.compare_models("GaussianStimBoth2", "GaussianStimBoth", 0.01, smoother_value=100) # pipeline.compare_models("GaussianStimBoth3", "GaussianStimBoth", 0.01, smoother_value=100) # pipeline.compare_models("GaussianStimBoth4", "GaussianStimBoth", 0.01, smoother_value=100) pipeline.compare_even_odd("Const", "GaussianBoth", 0.01) pipeline.compare_even_odd("Const", "GaussianStimBoth", 0.01) pipeline.compare_even_odd("GaussianBoth", "GaussianStimBoth", 0.01) pipeline.compare_even_odd("Const", "GaussianBothPos", 0.01) pipeline.compare_even_odd("Const", "GaussianStimBothPos", 0.01) pipeline.compare_even_odd("GaussianBoth", "GaussianBothPos", 0.01) pipeline.compare_even_odd("GaussianBoth", "GaussianStimBothPos", 0.01) pipeline.compare_even_odd("GaussianBothPos", "GaussianStimBoth", 0.01) pipeline.compare_even_odd("GaussianBothPos", "GaussianStimBothPos", 0.01) pipeline.compare_even_odd("GaussianStimBoth", "GaussianStimBothPos", 0.01)
def run_script(cell_range): # path_to_data = "/Users/stevecharczynski/workspace/data/jay/ca1_time_cell_data/" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/jay/ca1_time_cells" # save_dir = "/Users/stevecharczynski/Desktop/test/" # # save_dir = "/projectnb/ecog-eeg/stevechar/ml_output/jay/ca1_time_cells/" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/all_sessions/{0}".format(session) # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range) # solver_params = { # "niter": 1, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # n = 2 # bounds_t = { # "a_1": [0, 1 / n], # "ut": [0, 9000], # "st": [10, 6000], # "a_0": [10**-10, 1 / n] # } # bounds_dual = { # "a_1": [0, 1/3], # "a_2": [0, 1/3], # "ut_1": [0, 9000], # "ut_2": [0, 9000], # "st_1": [10, 6000], # "st_2": [10, 6000], # "a_0": [10**-10, 1/3] # } # bounds_c = { # "a_0": [10**-10, 1 / n] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "ConstVariable", "TimeVariableLength", "DualVariableLength"], save_dir=save_dir) # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable"], save_dir=save_dir) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("ConstVariable", bounds_c) # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("DualVariableLength", bounds_dual) # pipeline.set_model_x0("TimeVariableLength", [1e-5, 1000, 300, 1e-5]) # pipeline.set_model_x0("ConstVariable", [1e-5]) # pipeline.set_model_x0("DualVariableLength", [1e-5, 1e-5, 1000, 3000, 300, 300, 1e-5]) # # pipeline.show_rasters() # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_even_odd("ConstVariable", "TimeVariableLength", 0.01) # pipeline.compare_even_odd("TimeVariableLength", "DualVariableLength", 0.01) # # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.compare_models("ConstVariable", "TimeVariableLength", 0.01, smoother_value=100) # # pipeline.compare_models("TimeVariableLength", "DualVariableLength", 0.01, smoother_value=100) # # path_to_data = "/Users/stevecharczynski/workspace/data/jay/lec/" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/jay/lec/" # # save_dir = "/Users/stevecharczynski/workspace/data/jay/lec/" # save_dir = "/projectnb/ecog-eeg/stevechar/ml_runs/jay/lec/" # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range) # solver_params = { # "niter": 500, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # n = 2 # bounds_t = { # "a_1": [0, 1 / n], # "ut": [0, 9000], # "st": [10, 6000], # "a_0": [10**-10, 1 / n] # } # bounds_dual = { # "a_1": [0, 1/3], # "a_2": [0, 1/3], # "ut_1": [0, 9000], # "ut_2": [0, 9000], # "st_1": [10, 6000], # "st_2": [10, 6000], # "a_0": [10**-10, 1/3] # } # bounds_c = { # "a_0": [10**-10, 1 / n] # } # bounds_smt = { # "sigma": [0, 6000.], # "mu": [0, 9000.], # "tau": [20, 20000.], # "a_1": [10**-10, 1/2.], # "a_0": [10**-10, 1/2.] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "ConstVariable", "TimeVariableLength", "DualVariableLength", "SigmaMuTauVariableLength"], save_dir=save_dir) # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable"], save_dir=save_dir) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("ConstVariable", bounds_c) # pipeline.set_model_bounds("SigmaMuTauVariableLength", bounds_smt) # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("DualVariableLength", bounds_dual) # # pipeline.set_model_bounds("LogNormalVariableLength", bounds_t) # pipeline.set_model_x0("SigmaMuTauVariableLength", [100, 1000, 300, 1e-5, 1e-5]) # pipeline.set_model_x0("ConstVariable", [1e-5]) # # pipeline.set_model_x0("LogNormalVariableLength", [1e-5, 1000, 100, 1e-5]) # pipeline.set_model_x0("TimeVariableLength", [1e-5, 1000, 100, 1e-5]) # pipeline.set_model_x0("DualVariableLength", [1e-5, 1e-5, 1000, 3000, 300, 300, 1e-5]) # # pipeline.show_rasters() # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_even_odd("ConstVariable", "SigmaMuTauVariableLength", 0.01) # pipeline.compare_even_odd("ConstVariable", "TimeVariableLength", 0.01) # pipeline.compare_even_odd("TimeVariableLength", "SigmaMuTauVariableLength", 0.01) # pipeline.compare_even_odd("TimeVariableLength", "DualVariableLength", 0.01) # # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.compare_models("ConstVariable", "TimeVariableLength", 0.01, smoother_value=100) # pipeline.compare_models("TimeVariableLength", "SigmaMuTauVariableLength", 0.01, smoother_value=100) # pipeline.compare_models("ConstVariable", "SigmaMuTauVariableLength", 0.01, smoother_value=100) # pipeline.compare_models("TimeVariableLength", "DualVariableLength", 0.01, smoother_value=100) # pipeline.compare_models("TimeVariableLength", "DualVariableLength", 0.01, smoother_value=100) # path_to_data = "/Users/stevecharczynski/workspace/data/jay/lec/" path_to_data = "/projectnb/ecog-eeg/stevechar/data/jay/lec_pyr" # save_dir = "/Users/stevecharczynski/workspace/data/jay/lec/" save_dir = "/projectnb/ecog-eeg/stevechar/ml_runs/jay/lec/" # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) data_processor = analysis.DataProcessor(path_to_data, cell_range) solver_params = { "niter": 500, "stepsize": 1000, "interval": 10, "method": "TNC", "use_jac": True, "T": 1, "disp": False } n = 2 bounds_t = { "a_1": [0, 1 / n], "ut": [0, 9000], "st": [10, 6000], "a_0": [10**-10, 1 / n] } bounds_inhib = { "a_1": [10e-10, 1 / n], "ut": [0, 9000], "st": [10, 6000], "a_0": [10**-10, 1 / n] } bounds_dual = { "a_1": [0, 1 / 3], "a_2": [0, 1 / 3], "ut_1": [0, 9000], "ut_2": [0, 9000], "st_1": [10, 6000], "st_2": [10, 6000], "a_0": [10**-10, 1 / 3] } bounds_c = {"a_0": [10**-10, 1 / n]} bounds_smt = { "sigma": [0, 6000.], "mu": [0, 9000.], "tau": [20, 20000.], "a_1": [10**-10, 1 / 2.], "a_0": [10**-10, 1 / 2.] } # bounds_smt_inhib = { # "sigma": [0, 6000.], # "mu": [0, 9000.], # "tau": [20, 20000.], # "a_1": [10**-10, 1/2.], # "a_0": [10**-10, 1/2.] # } pipeline = analysis.Pipeline( cell_range, data_processor, [ "ConstVariable", "SigmaMuTauVariableLength", "InhibitSigmaMuTauJayLEC", "TimeVariableLength", "TimeInhibitVariableLength" ], save_dir=save_dir) # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "ConstVariable", "RelPosVariable"], save_dir=save_dir) # pipeline.set_model_bounds("TimeVariableLength", bounds_t) pipeline.set_model_bounds("ConstVariable", bounds_c) pipeline.set_model_bounds("TimeVariableLength", bounds_t) pipeline.set_model_bounds("TimeInhibitVariableLength", bounds_inhib) pipeline.set_model_bounds("InhibitSigmaMuTauJayLEC", bounds_smt) pipeline.set_model_bounds("SigmaMuTauVariableLength", bounds_smt) # pipeline.set_model_bounds("LogNormalVariableLength", bounds_t) pipeline.set_model_x0("ConstVariable", [1e-5]) pipeline.set_model_x0("SigmaMuTauVariableLength", [100, 1000, 300, 1e-5, 1e-5]) pipeline.set_model_x0("InhibitSigmaMuTauJayLEC", [100, 1000, 300, 1e-5, 1e-4]) # pipeline.set_model_x0("LogNormalVariableLength", [1e-5, 1000, 100, 1e-5]) pipeline.set_model_x0("TimeVariableLength", [1e-5, 1000, 100, 1e-5]) pipeline.set_model_x0("TimeInhibitVariableLength", [1e-5, 1000, 100, 1e-4]) # pipeline.show_rasters() pipeline.fit_even_odd(solver_params=solver_params) pipeline.fit_all_models(solver_params=solver_params) pipeline.compare_even_odd("ConstVariable", "TimeInhibitVariableLength", 0.01) pipeline.compare_even_odd("ConstVariable", "SigmaMuTauVariableLength", 0.01) pipeline.compare_even_odd("ConstVariable", "InhibitSigmaMuTauJayLEC", 0.01) pipeline.compare_even_odd("ConstVariable", "TimeVariableLength", 0.01) # pipeline.compare_even_odd("Const", "Time", 0.01) pipeline.compare_models("ConstVariable", "TimeVariableLength", 0.01, smoother_value=100) pipeline.compare_models("ConstVariable", "TimeInhibitVariableLength", 0.01, smoother_value=100) pipeline.compare_models("ConstVariable", "SigmaMuTauVariableLength", 0.01, smoother_value=100) pipeline.compare_models("ConstVariable", "InhibitSigmaMuTauJayLEC", 0.01, smoother_value=100)
def run_script(cell_range): # path_to_data = "/Users/stevecharczynski/workspace/data/warden/recog_trials/" # save_dir = "/Users/stevecharczynski/workspace/data/warden/recog_trials_first_stim_time/" save_dir = "/projectnb/ecog-eeg/stevechar/ml_runs/warden/recog_trials_first_stim_time/" path_to_data = "/projectnb/ecog-eeg/stevechar/data/warden/recog_trials/" data_processor = analysis.DataProcessor(path_to_data, cell_range, window=[-500, 1500]) solver_params = { "niter": 1500, "stepsize": 100, "interval": 20, "method": "TNC", "use_jac": True, "T": 100, "disp": False } bounds_stim = { "sigma": [1e-4, 1000.], "mu": [0, 1500.], "a_1": [1e-10, 1 / 2.], "a_0": [1e-10, 1 / 2.] } pipeline = analysis.Pipeline(cell_range, data_processor, [ "Const","GaussianStim1", \ "GaussianStim2", "GaussianStim3", "GaussianStim4"], save_dir=save_dir) pipeline.set_model_bounds("GaussianStim1", bounds_stim) pipeline.set_model_bounds("GaussianStim2", bounds_stim) pipeline.set_model_bounds("GaussianStim3", bounds_stim) pipeline.set_model_bounds("GaussianStim4", bounds_stim) pipeline.set_model_bounds("Const", {"a_0": [10e-10, 1]}) # with open("/Users/stevecharczynski/workspace/data/warden/recog_trials/info.json") as f: with open( "/projectnb/ecog-eeg/stevechar/data/warden/recog_trials/info.json" ) as f: stims = json.load(f) stims = {int(k): v for k, v in stims.items()} pipeline.set_model_info("GaussianStim1", "stim_identity", stims, per_cell=True) pipeline.set_model_info("GaussianStim2", "stim_identity", stims, per_cell=True) pipeline.set_model_info("GaussianStim3", "stim_identity", stims, per_cell=True) pipeline.set_model_info("GaussianStim4", "stim_identity", stims, per_cell=True) pipeline.set_model_x0("GaussianStim1", [10, 10, 1e-1, 1e-1]) pipeline.set_model_x0("GaussianStim2", [10, 10, 1e-1, 1e-1]) pipeline.set_model_x0("GaussianStim3", [10, 10, 1e-1, 1e-1]) pipeline.set_model_x0("GaussianStim4", [10, 10, 1e-1, 1e-1]) pipeline.set_model_x0("Const", [1e-1]) pipeline.fit_all_models(solver_params=solver_params) pipeline.fit_even_odd(solver_params=solver_params) pipeline.compare_models("Const", "GaussianStim1", 0.01, smoother_value=100) pipeline.compare_models("Const", "GaussianStim2", 0.01, smoother_value=100) pipeline.compare_models("Const", "GaussianStim3", 0.01, smoother_value=100) pipeline.compare_models("Const", "GaussianStim4", 0.01, smoother_value=100) pipeline.compare_even_odd("Const", "GaussianStim1", 0.01) pipeline.compare_even_odd("Const", "GaussianStim2", 0.01) pipeline.compare_even_odd("Const", "GaussianStim3", 0.01) pipeline.compare_even_odd("Const", "GaussianStim4", 0.01)
def run_script(cell_range, session): # path_to_data = '/Users/stevecharczynski/workspace/data/jay/2nd_file' # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/jay" # # x = np.full(95, 0) # # y = np.full(95, 8000) # # trial_lengths = np.array(list(zip(x,y))) # # with open('/Users/stevecharczynski/workspace/data/jay/2nd_file/trial_lengths.json', 'w') as f: # # json.dump(trial_lengths.tolist(), f) # data_processor = DataProcessor( # path_to_data, cell_range) # n = 2 # solver_params = { # "niter": 2, # "stepsize": 10000, # "interval": 10, # "method": "TNC", # "use_jac": True, # } # bounds_smt = { # "sigma": [10, 10000], # "mu": [-1000, 10000], # "tau": [0.0002, 0.10], # "a_1": [10**-7, 0.5], # "a_0": [10**-7, 0.5] # } # bounds_t = { # "a_1": [0, 1 / n], # "ut": [-1000,10000], # "st": [100, 10000], # "a_0": [10**-10, 1 / n] # } # bounds_c = { # "a_0": [10**-10, 1 / n] # } # pipeline = AnalysisPipeline(cell_range, data_processor, [ # "Const", "Time"], 0) # # pipeline.set_model_bounds("Time", bounds_t) # # pipeline.set_model_bounds("Const", bounds_c) # pipeline.set_model_bounds(["Time", "Const"], [bounds_t, bounds_c]) # pipeline.set_model_x0(["Const", "Time"], [[1e-5], [1e-5, 100, 100, 1e-5]]) # # pipeline.set_model_bounds("SigmaMuTau", bounds_smt) # pipeline.fit_even_odd(solver_params) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_models("Const", "Time", 0.01) # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.compare_models("Time", "SigmaMuTau", 0.01) # path_to_data = '/Users/stevecharczynski/workspace/data/brincat_miller' # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/brincat_miller/" # time_info = [500, 1750] # data_processor = DataProcessor( # path_to_data, cell_range, time_info=time_info) # n = 2 # mean_delta = 0.10 * (time_info[1] - time_info[0]) # mean_bounds = ( # (time_info[0] - mean_delta), # (time_info[1] + mean_delta)) # solver_params = { # "niter": 100, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # } # bounds_smt = { # "sigma": [10, 1000], # "mu": [100, 2300], # "tau": [10, 5000], # "a_1": [10**-7, 0.5], # "a_0": [10**-7, 0.5] # } # bounds = { # "a_1": [0, 1 / n], # "ut": [mean_bounds[0], mean_bounds[1]], # "st": [10, 1000], # "a_0": [10**-10, 1 / n] # } # pipeline = AnalysisPipeline(cell_range, data_processor, [ # "SigmaMuTau", "Time"], 0) # # pipeline = AnalysisPipeline(cell_range, data_processor, [ # # "Time","SigmaMuTau"], 0) # pipeline.set_model_bounds("Time", bounds) # pipeline.set_model_bounds("SigmaMuTau", bounds_smt) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_models("Time", "SigmaMuTau", 0.01) # path_to_data = '/Users/stevecharczynski/workspace/data/brincat_miller' # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/brincat_miller/" # time_info = [500, 1750] # data_processor = DataProcessor( # path_to_data, cell_range, time_info=time_info) # n = 2 # mean_delta = 0.10 * (time_info[1] - time_info[0]) # mean_bounds = ( # (time_info[0] - mean_delta), # (time_info[1] + mean_delta)) # solver_params = { # "niter": 5, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # } # bounds_t = { # "a_1": [0, 1 / n], # "ut": [mean_bounds[0], mean_bounds[1]], # "st": [10, 1000], # "a_0": [10**-10, 1 / n] # } # bounds_c = { # "a_0": [10**-10, 1 / n] # } # pipeline = AnalysisPipeline(cell_range, data_processor, [ # "Const", "Time"], 0) # # pipeline = AnalysisPipeline(cell_range, data_processor, [ # # "Time","SigmaMuTau"], 0) # pipeline.set_model_bounds("Time", bounds_t) # pipeline.set_model_bounds("Const", bounds_c) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_models("Const", "Time", 0.01) # path_to_data = "/Users/stevecharczynski/workspace/data/salz" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/salz" # time_info = [1000, 21000] # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=time_info) # solver_params = { # "niter": 2, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # } # bounds = { # "a_1": [0, 1 / 2], # "ut": [0, 42000,], # "st": [10, 50000], # "a_0": [10**-10, 1 / 2] # } # bounds_c = {"a_0": [10**-10, 0.999]} # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Const", "Time"], 0) # pipeline.set_model_bounds("Time", bounds) # pipeline.set_model_bounds("Const", bounds_c) # pipeline.set_model_x0("Time", [1e-5, 500, 500, 1e-5]) # pipeline.set_model_x0("Const", [1e-5]) # pipeline.fit_even_odd(solver_params) # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.fit_all_models(solver_params) # pipeline.compare_models("Const", "Time", 0.01) # path_to_data = "/Users/stevecharczynski/workspace/data/cromer" # # path_to_data = '/projectnb/ecog-eeg/stevechar/data/cromer' # # path_to_data = "/usr3/bustaff/scharcz/workspace/cromer/" # time_info = [400, 2000] # data_processor = DataProcessor( # path_to_data, cell_range, time_info=time_info) # n_c = 5 # solver_params = { # "niter": 1, # "stepsize": 100, # "interval": 10, # "method": "TNC", # "use_jac": False, # } # bounds_cat = { # "ut": [0, 2400], # "st": [10, 5000], # "a_0": [10**-10, 1 / n_c], # "a_1": [10**-10, 1 / n_c], # "a_2": [10**-10, 1 / n_c], # "a_3": [10**-10, 1 / n_c], # "a_4": [10**-10, 1 / n_c] # } # n_cs = 3 # bounds_cs = { # "ut": [0, 2400], # "st": [10, 5000], # "a_0": [10**-10, 1 / n_cs], # "a_1": [10**-10, 1 / n_cs], # "a_2": [10**-10, 1 / n_cs], # } # pipeline = AnalysisPipeline(cell_range, data_processor, [ # "CatSetTime","CatTime"], 0.01) # pipeline.set_model_bounds("CatSetTime", bounds_cs) # pipeline.set_model_bounds("CatTime", bounds_cat) # pipeline.set_model_info("CatSetTime", "pairs", [(1,2), (3,4)]) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_models("CatSetTime", "CatTime", 0.01) # pipeline.show_condition_fit("CatTime") # pipeline.show_condition_fit("CatSetTime") # path_to_data = "/Users/stevecharczynski/workspace/data/kim" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/kim" # time_info = [0, 4784] # data_processor = DataProcessor( # path_to_data, cell_range, time_info=time_info) # n = 2 # swarm_params = { # "phip": 0.5, # "phig": 0.5, # "omega": 0.5, # "minstep": 1e-10, # "minfunc": 1e-10, # "maxiter": 1000 # } # bounds = { # "a_1": [0, 1 / n], # "ut": [-200, 5200], # "st": [10, 10000], # "a_0": [10**-10, 1 / n] # } # bounds_c = {"a_0": [10**-10, 0.999]} # pipeline = AnalysisPipeline(cell_range, data_processor, [ # "Const", "Time"], 0, swarm_params) # pipeline.set_model_bounds("Time", bounds) # pipeline.set_model_bounds("Const", bounds_c) # pipeline.fit_all_models(1) # pipeline.compare_models("Const", "Time", 0.01) # path_to_data = "/Users/stevecharczynski/workspace/data/cromer" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/cromer" # # with open(path_to_data+'/number_of_trials.json', 'r') as f: # # num = json.load(f) # # x = np.full(max(num), 400) # # y = np.full(max(num), 2000) # # trial_lengths = np.array(list(zip(x,y))) # # with open(path_to_data+'/trial_lengths.json', 'w') as f: # # json.dump(trial_lengths.tolist(), f) # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=[400,2000]) # n_t = 2. # solver_params = { # "niter": 1, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp": False, # } # bounds = { # "a_1": [10**-10, 1 / n_t], # "ut": [0., 6000.], # "st": [10., 5000.], # "a_0": [10**-10, 1 / n_t] # } # bounds_c = {"a_0": [10**-10, 0.999]} # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Time", "Const"], 0) # # pipeline.show_rasters() # pipeline.set_model_bounds("Time", bounds) # pipeline.set_model_bounds("Const", bounds_c) # pipeline.set_model_x0("Time", [1e-5, 400, 100, 1e-5]) # pipeline.set_model_x0("Const", [1e-5]) # # pipeline.fit_even_odd(solver_params) # # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_models("Const", "Time", p_value=0.01, smoother_value=1) # path_to_data = '/Users/stevecharczynski/workspace/data/cromer' # # path_to_data = "/usr3/bustaff/scharcz/workspace/cromer/" # time_info = [400, 2000] # data_processor = DataProcessor( # path_to_data, cell_range, time_info=time_info) # n_c = 3 # swarm_params = { # "phip": 0.5, # "phig": 0.7, # "omega": 0.7, # "minstep": 1e-10, # "minfunc": 1e-10, # "maxiter": 1000 # } # bounds_cat = { # "ut": [0, 2400], # "st": [10, 5000], # "a_0": [10**-10, 1 / n_c], # "a_1": [10**-10, 1 / n_c], # "a_2": [10**-10, 1 / n_c], # } # n_t = 2 # bounds_t = { # "a_1": [0, 1 / n_t], # "ut": [0, 2400], # "st": [10, 5000], # "a_0": [10**-10, 1 / n_t] # } # # bounds_cat = ((0,2400), (10, 5000), (10**-10, 1 / n), (0, 1 / n),(0, 1 / n), (0, 1 / n), (0, 1 / n)) # # bounds= ((0, 1 / n), (0,2400), (10, 5000), (10**-10, 1 / n)) # pipeline = AnalysisPipeline(cell_range, data_processor, [ # "CatSetTime", "Time"], 0, swarm_params) # pipeline.set_model_bounds("Time", bounds_t) # pipeline.set_model_bounds("CatSetTime", bounds_cat) # pipeline.set_model_info("CatSetTime", "pairs", [(1,2), (3,4)]) # pipeline.fit_all_models(1) # pipeline.compare_models("Time", "CatSetTime", 0.01) # pipeline.show_condition_fit("CatSetTime") # path_to_data = '/Users/stevecharczynski/workspace/rui_fake_cells/mixed_firing' # time_info = RegionInfo(0, 2000) # data_processor = DataProcessor( # path_to_data, cell_range, time_info=time_info) # n = 2 # bounds = { # "a_1": [0, 1 / n], # "ut": [-500, 2500], # "st": [0.01, 5000], # "a_0": [10**-10, 1 / n] # } # bounds_c = {"a_0": [10**-10, 0.99]} # swarm_params = { # "phip": 0.5, # "phig": 0.5, # "omega": 0.5, # "minstep": 1e-10, # "minfunc": 1e-10, # "maxiter": 1000 # } # pipeline = AnalysisPipeline(cell_range, data_processor, [ # "Const", "Time"], 0, swarm_params) # pipeline.set_model_bounds("Time", bounds) # pipeline.set_model_bounds("Const", bounds_c) # pipeline.fit_all_models(3) # pipeline.compare_models("Const", "Time", 0.01) # util.collect_data(cell_range, "log_likelihoods") # util.collect_data(cell_range, "model_comparisons") # util.collect_data(cell_range, "cell_fits") # path_to_data = "/Users/stevecharczynski/workspace/data/sheehan/lin_track_s1" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/cromer" # time_info = [0, 3993] # data_processor = DataProcessor( # path_to_data, cell_range) # n_t = 2. # solver_params = { # "niter": 10, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # } # bounds_c = { # "a_0": [10**-10, 1 / n_t] # } # bounds_t = { # "a_1": [10**-10, 1 / n_t], # "ut": [0., 2400.], # "st": [10., 5000.], # "a_0": [10**-10, 1 / n_t] # } # pipeline = AnalysisPipeline(cell_range, data_processor, [ # "Const", "TimeVariableLength"], 0) # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("Const", {"a_0":[10**-10, 1]}) # # with open("/Users/stevecharczynski/workspace/data/sheehan/lin_track_s1/trial_length.json", 'rb') as f: # # trial_length = json.load(f) # # pipeline.set_model_info("TimeVariableLength", "trial_length", trial_length) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_models("Const", "TimeVariableLength", 0.01) # path_to_data = "/Users/stevecharczynski/workspace/data/crcns/hc-2/hc2/ec013.527" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/s25" # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=[0, 1076686]) # n_t = 2. # solver_params = { # "niter": 20, # "stepsize": 50, # "interval": 4, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":True # } # bounds_pos = { # "ut_x": [20, 200], # "st_x": [1., 30.], # "ut_y": [50., 250.], # "st_y": [1., 30.], # "a_0": [10**-10, 1 / n_t], # "a_1": [10**-10, 1 / n_t] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "PlaceField", "Const"]) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("PlaceField", bounds_pos) # pipeline.set_model_bounds("Const", {"a_0":[10**-10, 0.5]}) # pipeline.set_model_x0("PlaceField", [100,10,100,10,1e-5, 1e-5]) # pipeline.set_model_x0("Const", [1e-5]) # import numpy as np # import json # with open(path_to_data+"/pos_minus_removed.json", 'rb') as f: # pos = np.array(json.load(f)) # pipeline.set_model_info("PlaceField", "pos", pos, False) # model_dict = pipeline.fit_all_models(solver_params=solver_params) # plotting.plot_summed_2d(data_processor.spikes_binned[8], [300,300], pos, model_dict["PlaceField"][8].fit) # pipeline.compare_models("Const", "PlaceField", 0.01, smoother_value=1000) # path_to_data = "/Users/stevecharczynski/workspace/data/bulkin/" # path_to_data = "/projectnb/ecog-eeg/stevechar/bolkan/clusters_files/" # data_processor = analysis.DataProcessor(path_to_data, cell_range, [0, 60000]) # solver_params = { # "niter": 300, # "stepsize": 5000, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # bounds_vel = { # "a_1": [10**-10, 1 / 2], # "ut": [0., 70000.], # "st": [100, 80000.], # "a_0": [10**-10, 1 / 2] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Const", "Time"]) # pipeline.set_model_bounds("Time", bounds_vel) # pipeline.set_model_bounds("Const", {"a_0":[10**-10, 1]}) # pipeline.set_model_x0("Time", [1e-5, 20000, 1000, 1e-5]) # pipeline.set_model_x0("Const", [1e-5]) # # pipeline.show_rasters() # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_models("Const", "Time", 0.001, smoother_value=1000) # pipeline.compare_even_odd("Const", "Time", 0.001) # path_to_data = '/Users/stevecharczynski/workspace/data/brincat_miller' # path_to_data = "/projectnb/ecog-eeg/stevechar/data/brincat_miller/" # time_info = [500, 1750] # save_dir = "/Users/stevecharczynski/Desktop/test/" # save_dir = "/projectnb/ecog-eeg/stevechar/ml_output/brincat_miller/" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/all_sessions/{0}".format(session) # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=time_info) # n_t = 2. # solver_params = { # "niter": 5, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # mean_delta = 0.10 * (time_info[1] - time_info[0]) # mean_bounds = ( # (time_info[0] - mean_delta), # (time_info[1] + mean_delta)) # n = 2 # bounds_t = { # "a_1": [0, 1 / n], # "ut": [mean_bounds[0], mean_bounds[1]], # "st": [10, 1000], # "a_0": [10**-10, 1 / n] # } # bounds_c = { # "a_0": [10**-10, 1 / n] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Const", "Time"], save_dir=save_dir) # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable"], save_dir=save_dir) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("Const", bounds_c) # pipeline.set_model_bounds("Time", bounds_t) # pipeline.set_model_x0("Time", [1e-5, 700, 300, 1e-5]) # pipeline.set_model_x0("Const", [1e-5]) # # pipeline.show_rasters() # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_even_odd("Const", "Time", 0.01) # # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.compare_models("Const", "Time", 0.01, smoother_value=100) # # path_to_data = "/Users/stevecharczynski/workspace/data/jay/ca1_time_cell_data/" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/jay/ca1_time_cells" # # save_dir = "/Users/stevecharczynski/Desktop/test/" # save_dir = "/projectnb/ecog-eeg/stevechar/ml_output/jay/ca1_time_cells/" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/all_sessions/{0}".format(session) # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range) # solver_params = { # "niter": 500, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # n = 2 # bounds_t = { # "a_1": [0, 1 / n], # "ut": [0, 9000], # "st": [10, 6000], # "a_0": [10**-10, 1 / n] # } # bounds_dual = { # "a_1": [0, 1/3], # "a_2": [0, 1/3], # "ut_1": [0, 9000], # "ut_2": [0, 9000], # "st_1": [10, 6000], # "st_2": [10, 6000], # "a_0": [10**-10, 1/3] # } # bounds_c = { # "a_0": [10**-10, 1 / n] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "ConstVariable", "TimeVariableLength", "DualVariableLength"], save_dir=save_dir) # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable"], save_dir=save_dir) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("ConstVariable", bounds_c) # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("DualVariableLength", bounds_dual) # pipeline.set_model_x0("TimeVariableLength", [1e-5, 1000, 300, 1e-5]) # pipeline.set_model_x0("ConstVariable", [1e-5]) # pipeline.set_model_x0("DualVariableLength", [1e-5, 1e-5, 1000, 3000, 300, 300, 1e-5]) # # pipeline.show_rasters() # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_even_odd("ConstVariable", "TimeVariableLength", 0.01) # pipeline.compare_even_odd("TimeVariableLength", "DualVariableLength", 0.01) # # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.compare_models("ConstVariable", "TimeVariableLength", 0.01, smoother_value=100) # pipeline.compare_models("TimeVariableLength", "DualVariableLength", 0.01, smoother_value=100) # # path_to_data = "/Users/stevecharczynski/workspace/data/warden/recog_trials/" # # save_dir = "/Users/stevecharczynski/workspace/data/warden/recog_trials/" # save_dir = "/projectnb/ecog-eeg/stevechar/ml_runs/warden/recog_trials/" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/warden/recog_trials/" # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=[0, 1500]) # solver_params = { # "niter": 1000, # "stepsize": 100, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # bounds_smtstim = { # "sigma": [0, 1000.], # "mu": [0, 1500.], # "tau": [20, 20000.], # "a_1": [10**-10, 1/5.], # "a_2": [10**-10, 1/5.], # "a_3": [10**-10, 1/5.], # "a_4": [10**-10, 1/5.], # "a_0": [10**-10, 1/5.] # } # bounds_smt = { # "sigma": [0, 1000.], # "mu": [0, 1500.], # "tau": [20, 20000.], # "a_1": [10**-10, 1/2.], # "a_0": [10**-10, 1/2.] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Const","SigmaMuTau", "SigmaMuTauStim"], save_dir=save_dir) # pipeline.set_model_bounds("SigmaMuTau", bounds_smt) # pipeline.set_model_bounds("SigmaMuTauStim", bounds_smtstim) # pipeline.set_model_bounds("Const", {"a_0":[10e-10, 1]}) # # with open("/Users/stevecharczynski/workspace/data/warden/recog_trials/info.json") as f: # with open("/projectnb/ecog-eeg/stevechar/data/warden/recog_trials/info.json") as f: # stims = json.load(f) # stims = {int(k):v for k,v in stims.items()} # pipeline.set_model_info("SigmaMuTauStim", "stim_identity", stims, per_cell=True) # pipeline.set_model_x0("SigmaMuTauStim", [0.01, 1000, 100, 1e-1, 1e-1,1e-1, 1e-1, 1e-1]) # pipeline.set_model_x0("SigmaMuTau", [0.01, 1000, 100, 1e-1, 1e-1]) # pipeline.set_model_x0("Const", [1e-1]) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.compare_even_odd("Const", "SigmaMuTau", 0.01) # pipeline.compare_even_odd("SigmaMuTau", "SigmaMuTauStim", 0.01) # pipeline.compare_models("Const", "SigmaMuTau", 0.01, smoother_value=100) # pipeline.compare_models("SigmaMuTau", "SigmaMuTauStim", 0.01, smoother_value=100) # # path_to_data = "/Users/stevecharczynski/workspace/data/rossi_pool/a2/" # # save_dir = "/Users/stevecharczynski/workspace/data/rossi_pool/a2/" # save_dir = "/projectnb/ecog-eeg/stevechar/ml_runs/rossi_pool/a2/" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/rossi_pool/a2/" # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=[0, 3000]) # solver_params = { # "niter": 1000, # "stepsize": 100, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # bounds_smtstim = { # "sigma": [0, 1000.], # "mu": [0, 3000.], # "tau": [20, 20000.], # "a_1": [10**-10, 1/5.], # "a_2": [10**-10, 1/5.], # "a_0": [10**-10, 1/5.] # } # bounds_smt = { # "sigma": [0, 1000.], # "mu": [0, 3000.], # "tau": [20, 20000.], # "a_1": [10**-10, 1/2.], # "a_0": [10**-10, 1/2.] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Const","SigmaMuTau", "SigmaMuTauStimRP"], save_dir=save_dir) # pipeline.set_model_bounds("SigmaMuTau", bounds_smt) # pipeline.set_model_bounds("SigmaMuTauStimRP", bounds_smtstim) # pipeline.set_model_bounds("Const", {"a_0":[10e-10, 1]}) # # with open("/Users/stevecharczynski/workspace/data/rossi_pool/a2/info.json") as f: # with open("/projectnb/ecog-eeg/stevechar/data/rossi_pool/a2/info.json") as f: # stims = json.load(f) # stims = {int(k):v for k,v in stims.items()} # pipeline.set_model_info("SigmaMuTauStimRP", "stim_identity", stims, per_cell=True) # pipeline.set_model_x0("SigmaMuTauStimRP", [0.01, 1000, 100, 1e-1, 1e-1, 1e-1]) # pipeline.set_model_x0("SigmaMuTau", [0.01, 1000, 100, 1e-1, 1e-1]) # pipeline.set_model_x0("Const", [1e-1]) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.compare_even_odd("Const", "SigmaMuTau", 0.01) # pipeline.compare_even_odd("SigmaMuTau", "SigmaMuTauStimRP", 0.01) # pipeline.compare_models("Const", "SigmaMuTau", 0.01, smoother_value=100) # pipeline.compare_models("SigmaMuTau", "SigmaMuTauStimRP", 0.01, smoother_value=100) # # path_to_data = "/Users/stevecharczynski/workspace/data/warden/recall_trials/" # # save_dir = "/Users/stevecharczynski/workspace/data/warden/recall_trials/" # save_dir = "/projectnb/ecog-eeg/stevechar/ml_runs/warden/recall_trials" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/warden/recall_trials/" # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=[0, 3000]) # n_t = 2. # solver_params = { # "niter": 100, # "stepsize": 100, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # n = 2 # bounds_smtstim = { # "sigma1": [0, 1000.], # "sigma2": [0, 1000.], # # "sigma": [0.1, 0.15], # "mu1": [0, 3000.], # "mu2": [0, 3000.], # "tau1": [20, 20000.], # "tau2": [20, 20000.], # "a_1": [10**-10, 1/5.], # "a_2": [10**-10, 1/5.], # "a_3": [10**-10, 1/5.], # "a_4": [10**-10, 1/5.], # "a_0": [10**-10, 1/5.] # } # bounds_smt = { # "sigma1": [0, 1000.], # "sigma2": [0, 1000.], # # "sigma": [0.1, 0.15], # "mu1": [0, 3000.], # "mu2": [0, 3000.], # "tau1": [20, 20000.], # "tau2": [20, 20000.], # "a_1": [10**-10, 1/5.], # "a_0": [10**-10, 1/5.] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "SigmaMuTauDual", "SigmaMuTauDualStim"], save_dir=save_dir) # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable"], save_dir=save_dir) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("SigmaMuTauDual", bounds_smt) # pipeline.set_model_bounds("SigmaMuTauDualStim", bounds_smtstim) # # with open("/Users/stevecharczynski/workspace/data/warden/recall_trials/info.json") as f: # with open("/projectnb/ecog-eeg/stevechar/data/warden/recall_trials/info.json") as f: # stims = json.load(f) # stims = {int(k):v for k,v in stims.items()} # pipeline.set_model_info("SigmaMuTauDualStim", "stim_identity", stims, per_cell=True) # pipeline.set_model_x0("SigmaMuTauDualStim", [0.01,0.01, 1000,2000, 100, 100, 1e-1, 1e-1,1e-1, 1e-1, 1e-1]) # pipeline.set_model_x0("SigmaMuTauDual", [0.01,0.01, 1000,2000, 100, 100, 1e-1, 1e-1]) # pipeline.fit_all_models(solver_params=solver_params) # # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.compare_models("SigmaMuTauDual", "SigmaMuTauDualStim", 0.01, smoother_value=100) # # path_to_data = "/Users/stevecharczynski/workspace/data/sheehan/iti/sep_pos/1/" # # save_dir = "/Users/stevecharczynski/workspace/data/sheehan/iti/sep_pos/1/" # save_dir = "/projectnb/ecog-eeg/stevechar/sheehan_runs/iti/" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/iti/" # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=[0, 10500]) # n_t = 2. # solver_params = { # "niter": 300, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # n = 2 # bounds_t = { # "a_1": [0, 1 / n], # "ut": [-1000, 11000], # "st": [10, 6000], # "a_0": [10**-10, 1 / n] # } # bounds_c = { # "a_0": [10**-10, 1] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Const", "Time"], save_dir=save_dir) # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable"], save_dir=save_dir) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("Const", bounds_c) # pipeline.set_model_bounds("Time", bounds_t) # pipeline.set_model_x0("Time", [1e-5, 1000, 300, 1e-5]) # pipeline.set_model_x0("Const", [1e-5]) # # pipeline.show_rasters(show=False) # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_even_odd("Const", "Time", 0.01) # # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.compare_models("Const", "Time", 0.01, smoother_value=100) # path_to_data = "/Users/stevecharczynski/workspace/data/rossi_pool/a1/" # save_dir = "/Users/stevecharczynski/workspace/data/rossi_pool/a1/results/" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/jay/ca1_time_cells" # # save_dir = "/Users/stevecharczynski/Desktop/test/" # # save_dir = "/projectnb/ecog-eeg/stevechar/ml_output/jay/ca1_time_cells/" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/all_sessions/{0}".format(session) # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=[0, 5000]) # n_t = 2. # solver_params = { # "niter": 10, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": False, # "T" : 1, # "disp":False # } # n = 2 # bounds_smt = { # "sigma": [1e-4, 1.], # "mu1": [0, 5000.], # "mu2": [0, 5000.], # "tau": [10, 5000.], # "a_1": [10**-2, 1/3.], # "a_2": [10**-2, 1/3.], # "a_0": [10**-4, 1/3.] # } # bounds_c = { # "a_0": [10**-10, 1] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Const", "SigmaMuTauDual"], save_dir=save_dir) # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable"], save_dir=save_dir) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("SigmaMuTauDual", bounds_smt) # pipeline.set_model_bounds("Const", bounds_c) # with open("/Users/stevecharczynski/workspace/data/rossi_pool/a1/stim_identity.json") as f: # stims = json.load(f) # pipeline.set_model_info("SigmaMuTauDual", "stim_identity", stims) # pipeline.set_model_x0("SigmaMuTauDual", [0.01, 1000, 4000, 300, 1e-1, 1e-1, 1e-1]) # pipeline.set_model_x0("Const", [1e-5]) # # pipeline.show_rasters(save=True) # pipeline.fit_all_models(solver_params=solver_params) # # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.compare_models("Const", "SigmaMuTauDual", 0.01, smoother_value=100) # path_to_data = "/Users/stevecharczynski/workspace/data/rossi_pool/a1/" # save_dir = "/Users/stevecharczynski/workspace/data/rossi_pool/a1/results/" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/jay/ca1_time_cells" # # save_dir = "/Users/stevecharczynski/Desktop/test/" # # save_dir = "/projectnb/ecog-eeg/stevechar/ml_output/jay/ca1_time_cells/" # # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/all_sessions/{0}".format(session) # # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) # data_processor = analysis.DataProcessor( # path_to_data, cell_range, window=[0, 5000]) # n_t = 2. # solver_params = { # "niter": 5, # "stepsize": 1000, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # n = 2 # bounds_smtstim = { # "sigma1": [0, 1000.], # "sigma2": [0, 1000.], # # "sigma": [0.1, 0.15], # "mu1": [0, 5000.], # "mu2": [0, 5000.], # "tau1": [20, 20000.], # "tau2": [20, 20000.], # "a_1": [10**-5, 1/3.], # "a_2": [10**-5, 1/3.], # "a_0": [10**-5, 1/3.] # } # bounds_smt = { # "sigma1": [0, 1000.], # "sigma2": [0, 1000.], # # "sigma": [0.1, 0.15], # "mu1": [0, 5000.], # "mu2": [0, 5000.], # "tau1": [20, 20000.], # "tau2": [20, 20000.], # "a_1": [10**-5, 1/3.], # "a_0": [10**-5, 1/3.] # } # bounds_c = { # "a_0": [10**-10, 1] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "SigmaMuTauDual", "SigmaMuTauDualStim"], save_dir=save_dir) # # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable"], save_dir=save_dir) # # pipeline.set_model_bounds("TimeVariableLength", bounds_t) # pipeline.set_model_bounds("SigmaMuTauDualStim", bounds_smtstim) # pipeline.set_model_bounds("SigmaMuTauDual", bounds_smt) # with open("/Users/stevecharczynski/workspace/data/rossi_pool/a1/stim_identity.json") as f: # stims = json.load(f) # pipeline.set_model_info("SigmaMuTauDualStim", "stim_identity", stims) # pipeline.set_model_x0("SigmaMuTauDualStim", [0.01,0.01, 1000,4000, 100, 100, 1e-1, 1e-1, 1e-1]) # pipeline.set_model_x0("SigmaMuTauDual", [0.01,0.01, 1000,4000, 100, 100, 1e-1, 1e-1]) # # pipeline.set_model_x0("Const", [1e-5]) # # pipeline.show_rasters(save=True) # pipeline.fit_all_models(solver_params=solver_params) # # pipeline.compare_even_odd("Const", "Time", 0.01) # pipeline.compare_models("SigmaMuTauDual", "SigmaMuTauDualStim", 0.01, smoother_value=100) # path_to_data = "/Users/stevecharczynski/workspace/maxlikespy/examples/input_data/" # save_dir = "/Users/stevecharczynski/Desktop/test/" # data_processor = analysis.DataProcessor(path_to_data, cell_range, [0, 10000]) # solver_params = { # "niter": 5, # "stepsize": 200, # "interval": 10, # "method": "TNC", # "use_jac": True, # "T" : 1, # "disp":False # } # bounds_vel = { # "a_1": [10e-10, 1 / 2], # "ut": [-1000, 12000.], # "st": [100, 20000.], # "a_0": [10e-10, 1 / 2] # } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "Const", "Time"], save_dir=save_dir) # pipeline.set_model_bounds("Time", bounds_vel) # pipeline.set_model_bounds("Const", {"a_0":[10**-10, 1]}) # pipeline.set_model_x0("Time", [1e-5, 2000, 200, 1e-5]) # pipeline.set_model_x0("Const", [1e-5]) # # pipeline.show_rasters() # pipeline.fit_even_odd(solver_params=solver_params) # pipeline.fit_all_models(solver_params=solver_params) # pipeline.compare_models("Const", "Time", 0.001, smoother_value=100) # pipeline.compare_even_odd("Const", "Time", 0.001) path_to_data = "/Users/stevecharczynski/workspace/data/sheehan/random_box_lights_on/lights1_move1/inbound/" save_dir = "/Users/stevecharczynski/workspace/data/sheehan/random_box_lights_on/lights1_move1/inbound/" # save_dir = "/projectnb/ecog-eeg/stevechar/sheehan_runs/random_lights_on_only/inbound/" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/random_lights_on_only/inbound/" # save_dir = "/projectnb/ecog-eeg/stevechar/sheehan_runs/stable_lights_on_only/outbound/" # path_to_data = "/projectnb/ecog-eeg/stevechar/data/sheehan/stable_lights_on_only/outbound/" # time_info = list(zip(np.zeros(len(trial_length), dtype=int), trial_length)) data_processor = analysis.DataProcessor(path_to_data, cell_range) n_t = 2. solver_params = { "niter": 1, "stepsize": 500, "interval": 10, "method": "TNC", "use_jac": True, "T": 1, "disp": False } bounds_dual = { "ut_a": [0., 100.], "st_a": [0.1, 100.], "a_0a": [10**-10, 1 / 3], "a_1a": [10**-10, 1 / 3], "ut_b": [0., 100.], "st_b": [0.1, 100.], "a_0b": [10**-10, 1 / 3], "a_1b": [10**-10, 1 / 3], } bounds_norm = { "a_1": [10**-10, 1 / n_t], "ut": [0., 100.], "st": [0.1, 100.], "a_0": [10**-10, 1 / n_t] } bounds_t = { "a_1": [10**-10, 1 / n_t], "ut": [0., 5000.], "st": [10., 5000.], "a_0": [10**-10, 1 / n_t] } # pipeline = analysis.Pipeline(cell_range, data_processor, [ # # "ConstVariable", "RelPosVariable","DualPeakedRel", "AbsPosVariable"]) pipeline = analysis.Pipeline( cell_range, data_processor, [ "ConstVariable", "RelPosVariable", "AbsPosVariable", "DualPeakedRel", "DualPeakedAbs" ], save_dir=save_dir) # pipeline = analysis.Pipeline(cell_range, data_processor, [ # "ConstVariable", "RelPosVariable", "AbsPosVariable"], save_dir=save_dir) # pipeline.set_model_bounds("TimeVariableLength", bounds_t) pipeline.set_model_bounds("AbsPosVariable", bounds_norm) pipeline.set_model_bounds("RelPosVariable", bounds_norm) pipeline.set_model_bounds("ConstVariable", {"a_0": [10**-10, 1]}) pipeline.set_model_bounds("DualPeakedRel", bounds_dual) pipeline.set_model_bounds("DualPeakedAbs", bounds_dual) pipeline.set_model_x0("DualPeakedRel", [20, 1, 1e-5, 1e-5, 20, 1, 1e-5, 1e-5]) pipeline.set_model_x0("DualPeakedAbs", [20, 1, 1e-5, 1e-5, 20, 1, 1e-5, 1e-5]) pipeline.set_model_x0("AbsPosVariable", [1e-5, 20, 1, 1e-5]) pipeline.set_model_x0("RelPosVariable", [1e-5, 20, 1, 1e-5]) pipeline.set_model_x0("ConstVariable", [1e-5]) # pipeline.show_rasters() import numpy as np import json with open(path_to_data + "/abs_pos.json", 'r') as f: abs_pos = np.array(json.load(f)) with open(path_to_data + "/rel_pos.json", 'r') as f: rel_pos = np.array(json.load(f)) pipeline.set_model_info("AbsPosVariable", "abs_pos", abs_pos, True) pipeline.set_model_info("RelPosVariable", "rel_pos", rel_pos, True) pipeline.set_model_info("DualPeakedRel", "rel_pos", rel_pos, True) pipeline.set_model_info("DualPeakedAbs", "abs_pos", abs_pos, True) pipeline.fit_even_odd(solver_params=solver_params) pipeline.fit_all_models(solver_params=solver_params) pipeline.compare_even_odd("RelPosVariable", "DualPeakedRel", 0.01) pipeline.compare_even_odd("AbsPosVariable", "DualPeakedAbs", 0.01) pipeline.compare_even_odd("ConstVariable", "RelPosVariable", 0.01) pipeline.compare_even_odd("ConstVariable", "AbsPosVariable", 0.01) pipeline.compare_models("AbsPosVariable", "DualPeakedAbs", 0.01, smoother_value=100) pipeline.compare_models("RelPosVariable", "DualPeakedRel", 0.01, smoother_value=100) pipeline.compare_models("ConstVariable", "RelPosVariable", 0.01, smoother_value=100) pipeline.compare_models("ConstVariable", "AbsPosVariable", 0.01, smoother_value=100)