def norm_readout(pname, guess_range, time, d, model, norm):
"""
guess range should be a tuple of pmin pmax,
cond is a dict with model params
pname is name of parameter to be normalized
norm cond: value of area to normalize against
"""
sim_name = "dummy"
model_name = "dummy"
pmin, pmax = guess_range
d = dict(d)
d1 = dict(d)
d2 = dict(d)
d1[pname] = pmin
d2[pname] = pmax
df1 = run_exp(time, d1, model, sim_name, model_name)
df2 = run_exp(time, d2, model, sim_name, model_name)
area1 = get_area(df1.time, df1.cells)
area2 = get_area(df2.time, df2.cells)
#print(pname, cond, area1, area2)
if not(area1 < norm < area2):
guess = np.nan
print("param not in guess range or nan")
return guess
else:
guess = (pmin+pmax) / 2
crit = False
counter = 0
while crit == False:
counter = counter + 1
if counter > 50:
print("counter reached, stopping normalization")
break
d[pname] = guess
df = run_exp(time, d, model, sim_name, model_name)
area = get_area(df.time, df.cells)
#print("guess...area...pmin...pmax")
#print(guess, area, pmin, pmax)
if area < norm:
pmin = guess
guess = (guess+pmax)/2
else:
pmax = guess
guess = (guess+pmin)/2
if np.abs(area-norm) < 0.005:
crit = True
return guess
def get_readouts(self):
"""
get readouts from state array
"""
state = self.state
peak = readouts.get_peak_height(state.time, state.cells)
area = readouts.get_area(state.time, state.cells)
tau = readouts.get_peaktime(state.time, state.cells)
decay = readouts.get_duration(state.time, state.cells)
reads = [peak, area, tau, decay]
read_names = ["Peak Height", "Response Size", "Peak Time", "Decay"]
data = {"readout" : read_names, "read_val" : reads}
reads_df = pd.DataFrame(data = data)
reads_df["name"] = self.name
# deprecated, use only to compare menten vs threshold models
if "menten" in self.mode.__name__ :
modelname = "menten"
else:
modelname = "thres"
#
reads_df["model_name"] = modelname
return reads_df
def get_readouts_from_df(self, state):
"""
run simulation and then get readouts
Parameters
----------
state : TYPE
DESCRIPTION.
Returns
-------
reads_df : TYPE
DESCRIPTION.
"""
# get readouts
peak = readouts.get_peak_height(state.time, state.cells)
area = readouts.get_area(state.time, state.cells)
tau = readouts.get_peaktime(state.time, state.cells)
decay = readouts.get_decay(state.time, state.cells)
# convert readouts to dataframe
reads = [peak, area, tau, decay]
read_names = ["peak", "area", "tau", "decay"]
reads_df = pd.DataFrame(data=np.array([reads]), columns=read_names)
return reads_df
def norm(self, val, pname, norm):
"""
optimization function
calculate difference between simulated response size and wanted response size
val : parameter value
pname: str, parameter name
norm : wanted response size
returns float, difference in abs. values between wanted resp. size and calc. response size
"""
self.parameters[pname] = float(val)
state = self.state
area = readouts.get_area(state.time, state.cells)
return np.abs(area-norm)
def get_readouts(time, cells):
"""
get readouts from state array
"""
peak = readouts.get_peak_height(time, cells)
area = readouts.get_area(time, cells)
tau = readouts.get_peaktime(time, cells)
decay = readouts.get_duration(time, cells)
reads = [peak, area, tau, decay]
read_names = ["Peak Height", "Response Size", "Peak Time", "Decay"]
data = {"readout" : read_names, "read_val" : reads}
reads_df = pd.DataFrame(data = data)
return reads_df
def get_readouts(self):
"""
get readouts from state array
"""
state = self.state
peak = readouts.get_peak(state.time, state.cells)
area = readouts.get_area(state.time, state.cells)
tau = readouts.get_peaktime2(state.time, state.cells)
decay = readouts.get_duration(state.time, state.cells)
reads = [peak, area, tau, decay]
read_names = ["Peak", "Area", "Peaktime", "Decay"]
data = {"readout" : read_names, "read_val" : reads}
reads_df = pd.DataFrame(data = data)
reads_df["name"] = self.name
if "menten" in self.mode.__name__ :
modelname = "menten"
else:
modelname = "thres"
reads_df["model_name"] = modelname
return reads_df
import matplotlib.pyplot as plt import itertools # ============================================================================= # run experiment to get normalization conditions # ============================================================================= time = np.arange(0,20,0.01) sim_name = "dummy" model_name = "dummy" # readouts should be normalized for default param conditions between models model = C_model df = run_exp(time, d, model, sim_name, model_name) area_norm = get_area(df.time, df.cells) # ============================================================================= # heatmap conditions # ============================================================================= arr1 = np.logspace(-1, 1, 30) arr2 = arr1 name1 = "n_div" name2 = "r_diff" name3 = "gamma" pnames = [name1, name2, name3] model_list = [il2_model, timer_model, C_model] # =============================================================================
# systematic feedback analysis
fb_arr = np.geomspace(1.0, 10, 50)
fb_1 = []
fb_2 = []
fb_3 = []
fb_list = [fb_1, fb_2, fb_3]
# for all delay types vary feedback strength
for d, label, l in zip(dic, labels, fb_list):
# for each fb value get resposne size
for fb in fb_arr:
d["fb_strength"] = fb
state = run_model(time, d)
cells = get_cells(state, time, d)
cells = cells[["time", "eff"]]
area = get_area(time, cells.eff)
l.append(area)
df = pd.DataFrame({
"no_delay": fb_list[0],
"small_delay": fb_list[1],
"high_delay": fb_list[2],
"fb_fc": fb_arr
})
df = df.melt(id_vars="fb_fc",
value_name="population response",
var_name="delay")
df["mode"] = mode
df2_list.append(df)