from file_io import get_lines, put_lines from io import StringIO import os import sys import neuron as nrn import numpy as np from cell import cell_numel import subprocess import json def get_comp_index(types, compt_name): ind = [] for i in range(1, len(types) + 1): if(compt_name in types[i - 1] and types[i - 1][len(compt_name)] == '('): ind.append(i) return ind def write_fn_topo(out_fn,hoc_base_fn): out = StringIO.StringIO() nrn.h.hoc_stdout() sys.stdout = out nrn.h.MyTopology() my_topology = out.getvalue().split() sys.stdout = sys.__stdout__ def modify_param_set(param_set): param_fn = nrn.h.paramsFile new_param_fn = param_fn[:-4] + 'tmp.csv' if param_set is not False: pram_fn = param_set param_list = np.loadtxt(param_fn) for i in range(len(param_list)): prefix = 12300 param_list[i] = prefix + i np.savetxt(new_param_fn,param_list,newline = ' ') return new_param_fn def proc_add_param_to_hoc_for_opt(all_parameters_non_global_c, hoc_base_fn, base_p, available_mechanisms, neuron_sc, reversals, comp_names, comp_mechanisms, g_globals, n_globals, neuron_types, ftypestr, p_size_set, param_set,data_dir,all_states = None,kin_mod_inds = None,map_flag=False): if(map_flag): param_set = modify_param_set(param_set) nrn.h("access root.sec") print ("in procadd") KINFLG = True if kin_mod_inds == None or len(kin_mod_inds)==0: KINFLG = False if KINFLG: kin_mod_inds = [x+1 for x in kin_mod_inds] available_mechanisms = sorted(available_mechanisms, key=lambda s: s.lower()) all_parameters_non_global_c_d = {} tmp = [] for i in neuron_sc: #if len(i)>0: tmp.append(list(i)) #neuron_sc = [list(i) for i in neuron_sc] neuron_sc = tmp reversals = [] for i in neuron_sc: for j in i: if j not in reversals: reversals.append(j) comp_names = ['#' + i for i in comp_names] neuron_types = [i + '(0-1)' for i in neuron_types] comp_mechanisms = [list(i) for i in comp_mechanisms] param_start_i = [0] if len(all_parameters_non_global_c) == 1: param_start_i += [len(all_parameters_non_global_c[0])] else: param_start_i += list(np.cumsum(cell_numel(np.array(all_parameters_non_global_c)))) if len(all_parameters_non_global_c) == 0: param_start_i = [0] param_start_i = np.array(param_start_i) state_start_i = [0] if KINFLG: if len(kin_mod_inds) == 1: state_start_i += [len(all_states[kin_mod_inds[0]-1])] else: all_kin_states = [all_states[x] for x in kin_mod_inds ] state_start_i += list(np.cumsum(cell_numel(np.array(all_kin_states)))) if len(kin_mod_inds) == 0: state_start_i = [0] state_start_i = np.array(state_start_i) fn = hoc_base_fn fn_with_topo = fn[:-4] + '_topo.hoc' #write_fn_topo(fn_with_topo,hoc_base_fn) fn_with_param = fn[:-4] + '_param.hoc' file_sep = "/" fn_param_m ='ParamM.dat' fn_kinetic_states_init = 'KinStates.dat' fn_mat = 'Mat.dat' lines = get_lines(fn_with_topo) add_line_i = np.where(np.array(['End point processess mechanisms output' in i for i in lines]))[0] if add_line_i.size != 1: raise RuntimeError('Problem with finding place to add code: End point processes mechanisms output') p_size_line = np.where(np.array(['psize =' in i for i in lines])) if p_size_set: #p_size_line = np.where(np.array(['psize =' in i for i in lines])) if p_size_line[0].size != 0: for i in p_size_line[0]: lines[i] = lines[i][0:7] + str(p_size_set) time_steps_line = np.where(np.array(['ntimestep =' in i for i in lines])) n_sets = [] for i in p_size_line[0]: n_sets.append(int(lines[i][7:])) nt = [] for i in time_steps_line[0]: nt.append(int(lines[i][11:])) params_set_line = np.where(np.array(['paramsFile =' in i for i in lines])) if param_set: if params_set_line[0].size != 0: for i in params_set_line[0]: lines[i] = lines[i][:16] + param_set + '"' param_name = lines[params_set_line[0][0]][16:-5] reversals_c = [None for i in range(len(available_mechanisms))] if neuron_sc: for cur_mod_i in range(1, len(available_mechanisms) + 1): print(cur_mod_i) reversals_c[cur_mod_i - 1] = neuron_sc[cur_mod_i - 1] reversals_c = np.array(reversals_c) rep_comp = [None for i in range(len(reversals))] for i in range(0, len(reversals)): found_rep_comp = False while not found_rep_comp: for c in range(0, len(comp_names)): cur_mech_f = [] for k in available_mechanisms: cur_mech_f.append(k in comp_mechanisms[c]) cur_mech_f = np.where(np.array(cur_mech_f))[0] tmp = reversals_c[cur_mech_f] cur_reversals = [] for j in range(0, len(tmp)): curr_temp = list(tmp[j]) for jj in range(0, len(curr_temp)): cur_reversals.append(curr_temp[jj]) cur_reversals = np.array(cur_reversals) if reversals[i] in cur_reversals: rep_comp[i] = c found_rep_comp = True break added_lines = [] added_lines.append('// Start params output') added_lines.append('objref fn2') added_lines.append('fn2 = new File()') added_lines.append('proc writeReversals(){') for i in range(1, len(reversals) + 1): added_lines.append('access ' + comp_names[rep_comp[i - 1] ][1:]) added_lines.append('a=' + str(reversals[i - 1])) added_lines.append('fn.vwrite(&a)') added_lines.append('}') added_lines.append('proc writeGGlobals(){') for i in range(1, len(g_globals) + 1): added_lines.append('a=' + str(g_globals[i - 1])) added_lines.append('fn.vwrite(&a)') added_lines.append('}') added_lines.append('proc writeNGlobals(){') for i in range(1, len(n_globals) + 1): added_lines.append('a=' + str(n_globals[i - 1])) added_lines.append('fn.vwrite(&a)') added_lines.append('}') funcs_index = len(added_lines) added_lines.append('proc printParams(){') added_lines.append('fn.wopen("' + fn_param_m + '")') added_lines.append('fn2.wopen("' + fn_kinetic_states_init + '")') added_lines.append('writeReversals()') added_lines.append('writeGGlobals()') added_lines.append('writeNGlobals()') added_lines.append('for (ii=0;ii 50: funcs.append('}') proc_counter += 1 counter = 0 func_name = 'proc' + str(proc_counter) + '()' added_lines.append(func_name) funcs.append('proc ' + func_name + '{') F = [] for i in available_mechanisms: F.append(i in comp_mechanisms[c - 1]) F = np.where(np.array(F)) funcs.append('access ' + comp_names[c - 1][1:]) counter += 1 for m in range(1, len(F[0]) + 1): cur_mech_params = all_parameters_non_global_c[F[0][m - 1]] curr_mech_states = all_states[F[0][m-1]] for p in range(1, len(cur_mech_params) + 1): funcs.append('a=' + cur_mech_params[p - 1]) funcs.append('fn.vwrite(&a)') counter += 2 if KINFLG: if m in kin_mod_inds: for p in range(1, len(curr_mech_states) + 1): funcs.append('a=' + curr_mech_states[p - 1]) funcs.append('fn2.vwrite(&a)') counter += 2 if counter > 0: funcs.append('}') added_lines_temp = [] added_lines_temp.extend([added_lines[i] for i in range(funcs_index)]) added_lines_temp.extend(funcs) added_lines_temp.extend([added_lines[i] for i in range(funcs_index, len(added_lines))]) added_lines = added_lines_temp added_lines.append('}') added_lines.append('fn.close()') added_lines.append('}') added_lines.append('//printParams()') added_lines.append('// End params output') added_lines.append('// Start Mat Output') added_lines.append('fcurrent()') #added_lines.append('hoc_stdout("' + fn_mat + '")') added_lines.append('MyPrintMatrix3()') #added_lines.append('hoc_stdout()') added_lines.append('// endMat Output') out_lines = [] out_lines.extend([lines[i] for i in range(int(add_line_i[0]) + 1)]) out_lines.extend(added_lines) out_lines.extend([lines[i] for i in range(int(add_line_i[0]) + 1, len(lines))]) # put_lines(fn_with_param, out_lines) runmodelparam_file = open('runModel_param.hoc', 'w') runmodelparam_file.write("%s\n" % '\n'.join(out_lines)) runmodelparam_file.close() runModel_hoc_object = nrn.hoc.HocObject() subprocess.call(["nrniv", "runModel_param.hoc"]) print ("running runmodel_param") #runModel_hoc_object.execute('~' + '\n'.join(out_lines)) #runModel_hoc_object.load_file(1, fn_with_param) # subprocess.call(["nrniv", "runModel_param.hoc"]) f = open(fn_param_m, 'rb') f2 = open(fn_kinetic_states_init, 'rb') reversals_v, g_globals_v, n_globals_v = [0 for i in range(len(reversals))], [0 for i in range(len(g_globals))], [0 for i in range(len(n_globals))] for i in range(len(reversals)): reversals_v[i] = np.fromfile(f, dtype=np.float64, count=1)[0] for i in range(len(g_globals)): g_globals_v[i] = np.fromfile(f, dtype=np.float64, count=1)[0] for i in range(len(n_globals)): n_globals_v[i] = np.fromfile(f, dtype=np.float64, count=1)[0] comp_topology_map = [None for i in range(len(comp_names))] all_params = np.zeros((n_sets[0], len(comp_names) * int(param_start_i[-1]))) all_states_vals = np.zeros((n_sets[0], len(comp_names) * int(state_start_i[-1]))) first_param_m = None for kk in range(1, n_sets[0] + 1): param_m = np.zeros((len(comp_names), int(param_start_i[-1]))) states_vals = np.zeros((len(comp_names), int(state_start_i[-1]))) for c in range(1, len(comp_names) + 1): comp_name = comp_names[c - 1] comp_ind = get_comp_index(neuron_types, comp_name[1:]) comp_topology_map[c - 1] = comp_ind kin_ind = -1 F = [] for i in available_mechanisms: F.append(i in comp_mechanisms[c - 1]) F = np.where(np.array(F)) for m in range(1, F[0].size + 1): cur_mech_params = all_parameters_non_global_c[F[0][m - 1]] curr_mech_states = all_states[F[0][m - 1]] for p in range(1, len(cur_mech_params) + 1): try: Tmp = np.fromfile(f, dtype=np.float64, count=1) param_m[[i - 1 for i in comp_ind], int(param_start_i[F[0][m - 1]] + p - 1)] = Tmp except: continue if KINFLG: if m in kin_mod_inds: kin_ind += 1 curr_mech_states = all_states[F[0][m - 1]] for p in range(1, len(curr_mech_states) + 1): Tmp = np.fromfile(f2, dtype=np.float64, count=1) states_vals[[i - 1 for i in comp_ind],state_start_i[kin_ind] + p-1] = Tmp tmp = param_m.flatten(order='F') all_params[kk - 1, :] = tmp tmp2 = states_vals.flatten(order='F') all_states_vals[kk - 1, :] = tmp2 if kk == 1: first_param_m = param_m # Store only the first param_m for templating purposes param_m = None # print param_m # all_params = all_params.reshape((all_params.shape[0] * all_params.shape[1],)) f.close() f2.close() f = open(data_dir + '/AllParams.csv', 'w') n_sets_s = StringIO() np.savetxt(n_sets_s, np.array(n_sets), fmt='%5.d', newline=',') n_sets_st = n_sets_s.getvalue() all_params_s = StringIO() np.savetxt(all_params_s, all_params, fmt='%.5e,', newline='\n') all_params_st = all_params_s.getvalue() f.write('%s\n%s\n' % (n_sets_st, all_params_st)) f.close() f2 = open(data_dir + '/AllStates.csv', 'w') all_states_s = StringIO() np.savetxt(all_states_s, all_states_vals, fmt='%.5e,', newline='\n') all_states_st = all_states_s.getvalue() f2.write('%s\n%s\n' % (n_sets_st, all_states_st)) f = open(data_dir + '/AllParams.dat', 'wb') f.write(np.array(n_sets).astype(np.uint16)) f.write(all_params.astype(np.float32)) f.close() f = open(data_dir + '/ParamsM.csv', 'w') first_param_m_s = StringIO() np.savetxt(first_param_m_s, first_param_m.reshape((first_param_m.shape[0] * first_param_m.shape[1],)), fmt='%f', newline=',') first_param_m_st = first_param_m_s.getvalue() f.write('%s\n' % first_param_m_st) f.close() return first_param_m, runModel_hoc_object def proc_add_param_to_hoc_for_map(all_parameters_non_global_c, hoc_base_fn, base_p, available_mechanisms, neuron_sc, reversals, comp_names, comp_mechanisms, g_globals, n_globals, neuron_types, ftypestr, p_size_set, param_set,data_dir,all_states = None,kin_mod_inds = None): nrn.h("access root.sec") # access the root section (starting point) print ("in procadd") KINFLG = True if kin_mod_inds == None or len(kin_mod_inds)==0: # ??? KINFLG = False if KINFLG: kin_mod_inds = [x+1 for x in kin_mod_inds] available_mechanisms = sorted(available_mechanisms, key=lambda s: s.lower()) # sort by lowercase alphanumeric all_parameters_non_global_c_d = {} tmp = [] for i in neuron_sc: # iterate over lists of [e___], coresponding to ion current values (?) #if len(i)>0: tmp.append(list(i)) #neuron_sc = [list(i) for i in neuron_sc] neuron_sc = tmp reversals = [] for i in neuron_sc: # reversals contain all unique ion current values for j in i: if j not in reversals: reversals.append(j) comp_names = ['#' + i for i in comp_names] # list of component names neuron_types = [i + '(0-1)' for i in neuron_types] # list of neuron types comp_mechanisms = [list(i) for i in comp_mechanisms] # list of component mechanisms # print(comp_names) # print('\n') # print(neuron_types) # print('\n') # print(comp_mechanisms) param_start_i = [0] if len(all_parameters_non_global_c) == 1: param_start_i += [len(all_parameters_non_global_c[0])] else: param_start_i += list(np.cumsum(cell_numel(np.array(all_parameters_non_global_c)))) if len(all_parameters_non_global_c) == 0: param_start_i = [0] param_start_i = np.array(param_start_i) state_start_i = [0] if KINFLG: if len(kin_mod_inds) == 1: state_start_i += [len(all_states[kin_mod_inds[0]-1])] else: all_kin_states = [all_states[x] for x in kin_mod_inds ] state_start_i += list(np.cumsum(cell_numel(np.array(all_kin_states)))) if len(kin_mod_inds) == 0: state_start_i = [0] state_start_i = np.array(state_start_i) fn = hoc_base_fn # fn = ./runModel.hoc fn_with_topo = fn[:-4] + '_topo.hoc' # [:-4] removes .hoc extension so we can append new ending #write_fn_topo(fn_with_topo,hoc_base_fn) fn_with_param = fn[:-4] + '_param.hoc' # [:-4] removes .hoc extension so we can append new endin file_sep = "/" fn_param_m ='ParamM.dat' fn_kinetic_states_init = 'KinStates.dat' fn_mat = 'Mat.dat' lines = get_lines(fn_with_topo) # dump all lines in "./runModel_topo.hoc" into lines (a list) # print("lines:") # print(len(lines)) # 354 lines of code # print(lines) add_line_i = np.where(np.array(['End point processess mechanisms output' in i for i in lines]))[0] if add_line_i.size != 1: raise RuntimeError('Problem with finding place to add code: End point processes mechanisms output') p_size_line = np.where(np.array(['psize =' in i for i in lines])) if p_size_set: #p_size_line = np.where(np.array(['psize =' in i for i in lines])) if p_size_line[0].size != 0: for i in p_size_line[0]: lines[i] = lines[i][0:7] + str(p_size_set) time_steps_line = np.where(np.array(['ntimestep =' in i for i in lines])) n_sets = [] for i in p_size_line[0]: # p_size_line = subset containing keyword "psize =" n_sets.append(int(lines[i][7:])) nt = [] for i in time_steps_line[0]: nt.append(int(lines[i][11:])) params_set_line = np.where(np.array(['paramsFile =' in i for i in lines])) if param_set: if params_set_line[0].size != 0: for i in params_set_line[0]: lines[i] = lines[i][:16] + param_set + '"' param_name = lines[params_set_line[0][0]][16:-5] reversals_c = [None for i in range(len(available_mechanisms))] if neuron_sc: for cur_mod_i in range(1, len(available_mechanisms) + 1): print(cur_mod_i) reversals_c[cur_mod_i - 1] = neuron_sc[cur_mod_i - 1] reversals_c = np.array(reversals_c) rep_comp = [None for i in range(len(reversals))] for i in range(0, len(reversals)): found_rep_comp = False while not found_rep_comp: for c in range(0, len(comp_names)): cur_mech_f = [] for k in available_mechanisms: cur_mech_f.append(k in comp_mechanisms[c]) cur_mech_f = np.where(np.array(cur_mech_f))[0] tmp = reversals_c[cur_mech_f] cur_reversals = [] for j in range(0, len(tmp)): curr_temp = list(tmp[j]) for jj in range(0, len(curr_temp)): cur_reversals.append(curr_temp[jj]) cur_reversals = np.array(cur_reversals) if reversals[i] in cur_reversals: rep_comp[i] = c found_rep_comp = True break added_lines = [] added_lines.append('// Start params output') # Line 345 of runModel_param.hoc; create all procedures to write values added_lines.append('objref fn2') added_lines.append('fn2 = new File()') added_lines.append('proc writeReversals(){') for i in range(1, len(reversals) + 1): added_lines.append('access ' + comp_names[rep_comp[i - 1] ][1:]) added_lines.append('a=' + str(reversals[i - 1])) added_lines.append('fn.vwrite(&a)') added_lines.append('}') added_lines.append('proc writeGGlobals(){') for i in range(1, len(g_globals) + 1): added_lines.append('a=' + str(g_globals[i - 1])) added_lines.append('fn.vwrite(&a)') added_lines.append('}') added_lines.append('proc writeNGlobals(){') for i in range(1, len(n_globals) + 1): added_lines.append('a=' + str(n_globals[i - 1])) added_lines.append('fn.vwrite(&a)') added_lines.append('}') funcs_index = len(added_lines) added_lines.append('proc printParams(){') added_lines.append('fn.wopen("' + fn_param_m + '")') # fn_param_m = ParamM.dat added_lines.append('fn2.wopen("' + fn_kinetic_states_init + '")') # fn = hoc_base_fn = "./runModel.hoc" added_lines.append('writeReversals()') added_lines.append('writeGGlobals()') added_lines.append('writeNGlobals()') added_lines.append('for (ii=0;ii 50: # comp_names = all component names i.e. create proc for each section in model funcs.append('}') proc_counter += 1 counter = 0 func_name = 'proc' + str(proc_counter) + '()' added_lines.append(func_name) funcs.append('proc ' + func_name + '{') F = [] for i in available_mechanisms: F.append(i in comp_mechanisms[c - 1]) F = np.where(np.array(F)) funcs.append('access ' + comp_names[c - 1][1:]) # in Neuron, access specific section # print(proc_counter) # print(comp_names[c-1][1:]) counter += 1 # Iterate over components stored in F for m in range(1, len(F[0]) + 1): cur_mech_params = all_parameters_non_global_c[F[0][m - 1]] curr_mech_states = all_states[F[0][m-1]] # print(cur_mech_params) for p in range(1, len(cur_mech_params) + 1): # if there is a new component name, create a new dictionary if (comp_names[c-1][1:] not in param_mappings): param_mappings[comp_names[c-1][1:]] = {} # if parameter 'cur_mech_params[p - 1]' is not already in the dictionary 'param_mappings[comp_names[c-1][1:]]', enumerate it if (cur_mech_params[p - 1] not in param_mappings[comp_names[c-1][1:]].keys()): param_mappings[comp_names[c-1][1:]][cur_mech_params[p - 1]] = mapping_counter mapping_counter += 1 # write the value for each parameter to 'runModel_param.hoc (e.g. a=1 fn.vwrite(&a))' funcs.append('a=' + str(param_mappings.get(comp_names[c-1][1:]).get(cur_mech_params[p - 1]))) funcs.append('fn.vwrite(&a)') counter += 2 if KINFLG: if m in kin_mod_inds: for p in range(1, len(curr_mech_states) + 1): funcs.append('a=' + curr_mech_states[p - 1]) funcs.append('fn2.vwrite(&a)') counter += 2 # print(param_mappings) if counter > 0: funcs.append('}') added_lines_temp = [] added_lines_temp.extend([added_lines[i] for i in range(funcs_index)]) added_lines_temp.extend(funcs) added_lines_temp.extend([added_lines[i] for i in range(funcs_index, len(added_lines))]) added_lines = added_lines_temp added_lines.append('}') added_lines.append('fn.close()') added_lines.append('}') added_lines.append('printParams()') added_lines.append('// End params output') added_lines.append('// Start Mat Output') added_lines.append('fcurrent()') #added_lines.append('hoc_stdout("' + fn_mat + '")') added_lines.append('MyPrintMatrix3()') #added_lines.append('hoc_stdout()') added_lines.append('// endMat Output') # Finish appending to .hoc file out_lines = [] out_lines.extend([lines[i] for i in range(int(add_line_i[0]) + 1)]) out_lines.extend(added_lines) out_lines.extend([lines[i] for i in range(int(add_line_i[0]) + 1, len(lines))]) # put_lines(fn_with_param, out_lines) ################################################################################ # open runModel_param.hoc and overwrite it with 'out_lines' runmodelparam_file = open('runModel_param.hoc', 'w') runmodelparam_file.write("%s\n" % '\n'.join(out_lines)) runmodelparam_file.close() runModel_hoc_object = nrn.hoc.HocObject() print("running runModel_param.hoc...") subprocess.call(["nrniv", "runModel_param.hoc"]) # (Run runModel_param.hoc in NEURON) f = open(fn_param_m, 'rb') #fn_param_m = ParamM.dat; contains binary values (?) f2 = open(fn_kinetic_states_init, 'rb') #fn_kinetic_states_init = KinStates.dat reversals_v, g_globals_v, n_globals_v = [0 for i in range(len(reversals))], [0 for i in range(len(g_globals))], [0 for i in range(len(n_globals))] # Create 1-D arrays of 0s, length = length of respective array for i in range(len(reversals)): reversals_v[i] = np.fromfile(f, dtype=np.float64, count=1)[0] #get all reversal values from ParamM.dat for i in range(len(g_globals)): g_globals_v[i] = np.fromfile(f, dtype=np.float64, count=1)[0] #get all global values from ParamM.dat for i in range(len(n_globals)): n_globals_v[i] = np.fromfile(f, dtype=np.float64, count=1)[0] #get all global values from ParamM.dat (g vs n?) comp_topology_map = [None for i in range(len(comp_names))] all_params = np.zeros((n_sets[0], len(comp_names) * int(param_start_i[-1]))) all_states_vals = np.zeros((n_sets[0], len(comp_names) * int(state_start_i[-1]))) first_param_m = None # for kk in range(1, n_sets[0] + 1): for kk in range(1, 2): # if kk % 8 == 0: # print("%d / %d" % (kk, n_sets[0] + 1)) param_m = np.zeros((len(comp_names), int(param_start_i[-1]))) states_vals = np.zeros((len(comp_names), int(state_start_i[-1]))) for c in range(1, len(comp_names) + 1): comp_name = comp_names[c - 1] comp_ind = get_comp_index(neuron_types, comp_name[1:]) comp_topology_map[c - 1] = comp_ind kin_ind = -1 F = [] for i in available_mechanisms: F.append(i in comp_mechanisms[c - 1]) F = np.where(np.array(F)) for m in range(1, F[0].size + 1): cur_mech_params = all_parameters_non_global_c[F[0][m - 1]] curr_mech_states = all_states[F[0][m - 1]] for p in range(1, len(cur_mech_params) + 1): Tmp = np.fromfile(f, dtype=np.float64, count=1) param_m[[i - 1 for i in comp_ind], int(param_start_i[F[0][m - 1]] + p - 1)] = Tmp if KINFLG: if m in kin_mod_inds: kin_ind += 1 curr_mech_states = all_states[F[0][m - 1]] for p in range(1, len(curr_mech_states) + 1): Tmp = np.fromfile(f2, dtype=np.float64, count=1) states_vals[[i - 1 for i in comp_ind],state_start_i[kin_ind] + p-1] = Tmp tmp = param_m.flatten(order='F') all_params[kk - 1, :] = tmp tmp2 = states_vals.flatten(order='F') all_states_vals[kk - 1, :] = tmp2 if kk == 1: first_param_m = param_m # Store only the first param_m for templating purposes param_m = None # all_params = all_params.reshape((all_params.shape[0] * all_params.shape[1],)) f.close() f2.close() # f = open(data_dir + '/AllParams.csv', 'w') # Move all important data to .CSV file f = open(data_dir + 'ParamTemplate.csv', 'w') n_sets_s = StringIO() # Read/write strings as files # np.savetxt(n_sets_s, np.array(n_sets), fmt='%5.d', newline=',') # State at top of .csv file the number of models np.savetxt(n_sets_s, [1], fmt='%5.d', newline=',') n_sets_st = n_sets_s.getvalue() # retrieve contents of file all_params_s = StringIO() np.savetxt(all_params_s, all_params, fmt='%.5d,', newline='\n') all_params_st = all_params_s.getvalue() # print("n_sets:", len(n_sets)) # print("n_sets_st:", len(n_sets_st)) # print("all_params:", len(all_params)) # print("all_params_st:", len(all_params_st)) num_models = len(all_params); total_vals = len(all_params_st); num_params = (int) (total_vals/num_models) # f.write('%s\n%s\n' % (n_sets_st, all_params_st)) f.write('%s\n%s\n' % (n_sets_st, all_params_st[:num_params])) f.close() f2 = open(data_dir + '/AllStates.csv', 'w') all_states_s = StringIO() np.savetxt(all_states_s, all_states_vals, fmt='%.5e,', newline='\n') all_states_st = all_states_s.getvalue() f2.write('%s\n%s\n' % (n_sets_st, all_states_st)) f = open(data_dir + '/AllParams.dat', 'wb') f.write(np.array(n_sets).astype(np.uint16)) f.write(all_params.astype(np.float32)) f.close() f = open(data_dir + '/ParamsM.csv', 'w') first_param_m_s = StringIO() np.savetxt(first_param_m_s, first_param_m.reshape((first_param_m.shape[0] * first_param_m.shape[1],)), fmt='%f', newline=',') first_param_m_st = first_param_m_s.getvalue() f.write('%s\n' % first_param_m_st) # write .getvalue() of first_param_m_s f.close() f3 = open(data_dir + '/ParamMappings.txt', 'w') json.dump(param_mappings, f3); f3.close() return first_param_m, runModel_hoc_object