def section_6_2_3(virtual_successes=1, disp=True):
if disp: print("\nSection 6.2.3")
df = pd.DataFrame(columns=['lifetime', 'evolutionary', 'log-uniform', 'from research', 'eq. wt. avg.'])
overall = []
for method in ['conservative', 'aggressive']:
datadict = {'lifetime': functions.lifetime_anchor(method, virtual_successes=virtual_successes),
'evolutionary': functions.evolutionary_anchor(method, virtual_successes=virtual_successes),
'log-uniform': functions.log_uniform(method),
'from research': functions.four_param_framework_comp(rel_imp_res_comp=5,
ftp_cal_equiv=1 / 300,
spend2036=method,
virtual_successes=virtual_successes)
}
average = float(np.average([value for value in datadict.values()]))
overall.append(average)
datadict['eq. wt. avg.'] = average
datadict = {k: float(v) for k, v in datadict.items()}
datadict = {k: round_sig(v * 100, 2, type=str) + "%" for k, v in datadict.items()}
row = pd.Series(data=datadict, name=method)
df = df.append(row)
if disp: print(df)
overall = np.average(overall)
if disp: print("\nAll things considered average:", round_sig(overall * 100, 2, type=str) + "%")
return overall
def section6_3_1_helper_comp(ftp_cal_equiv, rel_imp_res_comp, biggest_spends_method, rowname, df):
if biggest_spends_method == '50/50':
left = 1 / 2 * (functions.four_param_framework_comp(ftp_cal_equiv=ftp_cal_equiv,
rel_imp_res_comp=rel_imp_res_comp,
spend2036='aggressive') +
functions.four_param_framework_comp(ftp_cal_equiv=ftp_cal_equiv,
rel_imp_res_comp=rel_imp_res_comp,
spend2036='conservative'))
right = 1 / 2 * (functions.four_param_framework_comp(ftp_cal_equiv=ftp_cal_equiv,
rel_imp_res_comp=rel_imp_res_comp,
spend2036='aggressive',
virtual_successes=0.5) +
functions.four_param_framework_comp(ftp_cal_equiv=ftp_cal_equiv,
rel_imp_res_comp=rel_imp_res_comp,
spend2036='conservative',
virtual_successes=0.5))
else:
left = functions.four_param_framework_comp(ftp_cal_equiv=ftp_cal_equiv,
rel_imp_res_comp=rel_imp_res_comp,
spend2036=biggest_spends_method)
right = functions.four_param_framework_comp(ftp_cal_equiv=ftp_cal_equiv,
rel_imp_res_comp=rel_imp_res_comp,
spend2036=biggest_spends_method,
virtual_successes=0.5)
left = round_sig(left * 100, 2, type=str) + "%"
right = round_sig(right * 100, 2, type=str) + "%"
return df.append(pd.Series(name=rowname, data={'1 VS': left, '0.5 VS': right}))
def section6_2_comp_from_research_helper(biggest_spends_method):
print("\nBiggest_spends_method:", biggest_spends_method)
columns = [1, 5, 10]
df = pd.DataFrame(columns=columns)
for ftp_cal_equiv in [1 / 50, 1 / 100, 1 / 300, 1 / 1000, 1 / 3000]:
dict_comprehension = {rel_imp_res_comp:
functions.four_param_framework_comp(rel_imp_res_comp=rel_imp_res_comp,
regime_start=1956,
forecast_to=2036,
spend2036=biggest_spends_method,
ftp_cal_equiv=ftp_cal_equiv)
for rel_imp_res_comp in columns}
dict_comprehension = {k: round_sig(v * 100, 2, type=str) + "%" for k, v in dict_comprehension.items()}
row_name = 'ftp_cal = ' + to_fraction_strings(ftp_cal_equiv)
row = pd.Series(data=dict_comprehension, name=row_name)
df = df.append(row)
df.columns = ["X=" + str(x) for x in columns] # rename the columns
print(df)
def section_6_3_1_virtual_succ():
print("\nSection 6.3.1, virtual successes")
df = pd.DataFrame(columns=['1 VS', '0.5 VS'])
g_exp = 4.3 / 100
rowname = "Researcher-year, low"
g_act = 7 / 100
ftp_cal_equiv = 1 / 1000
df = section6_3_1_helper_research(g_act, ftp_cal_equiv, g_exp, rowname, df)
rowname = "Researcher-year, middle"
g_act = 11 / 100
ftp_cal_equiv = 1 / 300
df = section6_3_1_helper_research(g_act, ftp_cal_equiv, g_exp, rowname, df)
rowname = "Researcher-year, high"
g_act = 16 / 100
ftp_cal_equiv = 1 / 100
df = section6_3_1_helper_research(g_act, ftp_cal_equiv, g_exp, rowname, df)
rowname = 'Computation, low'
ftp_cal_equiv = 1 / 1000
rel_imp_res_comp = 10
biggest_spends_method = 'conservative'
df = section6_3_1_helper_comp(ftp_cal_equiv, rel_imp_res_comp, biggest_spends_method, rowname, df)
rowname = 'Computation, central'
left = 0.5 * (functions.lifetime_anchor('conservative', virtual_successes=1) + functions.lifetime_anchor('aggressive', virtual_successes=1))
right = 0.5 * (functions.lifetime_anchor('conservative', virtual_successes=.5) + functions.lifetime_anchor('aggressive', virtual_successes=.5))
left = round_sig(left * 100, 2, type=str) + "%"
right = round_sig(right * 100, 2, type=str) + "%"
df = df.append(pd.Series(name=rowname, data={'1 VS': left, '0.5 VS': right}))
rowname = 'Computation, high'
ftp_cal_equiv = 1 / 300
rel_imp_res_comp = 1
biggest_spends_method = 'aggressive'
df = section6_3_1_helper_comp(ftp_cal_equiv, rel_imp_res_comp, biggest_spends_method, rowname, df)
rowname = 'Computation, central, bracketed weigh. avg.'
left = section_6_2_3(virtual_successes=1, disp=False)
right = section_6_2_3(virtual_successes=.5, disp=False)
left = round_sig(left * 100, 2, type=str) + "%"
right = round_sig(right * 100, 2, type=str) + "%"
df = df.append(pd.Series(name=rowname, data={'1 VS': left, '0.5 VS': right}))
rowname = 'Computation, high, bracketed weigh. avg.'
left = 0.5 * (functions.four_param_framework_comp(rel_imp_res_comp=1,
ftp_cal_equiv=1 / 300,
spend2036='aggressive',
virtual_successes=1) +
functions.log_uniform('aggressive'))
right = 0.5 * (functions.four_param_framework_comp(rel_imp_res_comp=1,
ftp_cal_equiv=1 / 300,
spend2036='aggressive',
virtual_successes=.5) +
functions.log_uniform('aggressive'))
left = round_sig(left * 100, 2, type=str) + "%"
right = round_sig(right * 100, 2, type=str) + "%"
df = df.append(pd.Series(name=rowname, data={'1 VS': left, '0.5 VS': right}))
print(df)
def computation_callable(year): return functions.four_param_framework_comp(forecast_to=year, **kwargs)