Spaces:
Runtime error
Runtime error
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,237 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import pandas as pd
|
2 |
+
import numpy as np
|
3 |
+
from random import normalvariate, random
|
4 |
+
import plotly.express as px
|
5 |
+
|
6 |
+
from radcad import Model, Simulation, Experiment
|
7 |
+
import streamlit as st
|
8 |
+
|
9 |
+
|
10 |
+
# Additional dependencies
|
11 |
+
|
12 |
+
# For analytics
|
13 |
+
import numpy as np
|
14 |
+
# For visualization
|
15 |
+
import plotly.express as px
|
16 |
+
pd.options.plotting.backend = "plotly"
|
17 |
+
|
18 |
+
st.header('DeSci Value Flow Model')
|
19 |
+
|
20 |
+
def p_researcher1(params, substep, state_history, previous_state):
|
21 |
+
losses = 0
|
22 |
+
to_market = 0
|
23 |
+
to_researcher = 0
|
24 |
+
to_treasury = 0
|
25 |
+
to_other_researcher = 0
|
26 |
+
salary = 0
|
27 |
+
funding = 0
|
28 |
+
if (previous_state['timestep'] < params['timestep_switch']) and (previous_state['funding_pool'] > params['funding_round']):
|
29 |
+
funding = params['funding_round']
|
30 |
+
to_treasury -= funding
|
31 |
+
research_value = funding * (1-params['epsilon'])
|
32 |
+
losses += funding - research_value
|
33 |
+
|
34 |
+
salary = research_value * params['beta']
|
35 |
+
to_market = research_value
|
36 |
+
if (random() < params['probability_buying']) and (previous_state['researcher1_value'] > params['cost_buying']):
|
37 |
+
salary = salary - params['cost_buying']
|
38 |
+
tx_fee = params['cost_buying'] * params['tx_fee']
|
39 |
+
salary -= tx_fee
|
40 |
+
to_treasury += tx_fee
|
41 |
+
to_other_researcher += params['cost_buying']
|
42 |
+
to_researcher += salary + research_value
|
43 |
+
elif (previous_state['timestep'] > params['timestep_switch']) and (previous_state['researcher1_value'] > params['cost_buying']):
|
44 |
+
tx_fee = params['cost_buying'] * params['tx_fee']
|
45 |
+
to_researcher -= params['cost_buying'] - tx_fee
|
46 |
+
to_other_researcher += params['cost_buying']
|
47 |
+
to_treasury += tx_fee
|
48 |
+
return {'update_researcher1_funding': funding,
|
49 |
+
'update_researcher1_salary': salary,
|
50 |
+
'update_researcher1_value': to_researcher,
|
51 |
+
'update_funding_pool': to_treasury,
|
52 |
+
'update_market': to_market,
|
53 |
+
'update_researcher2_value': to_other_researcher,
|
54 |
+
'update_losses': losses}
|
55 |
+
|
56 |
+
def p_researcher2(params, substep, state_history, previous_state):
|
57 |
+
losses = 0
|
58 |
+
to_market = 0
|
59 |
+
to_researcher = 0
|
60 |
+
to_treasury = 0
|
61 |
+
to_other_researcher = 0
|
62 |
+
salary = 0
|
63 |
+
funding = 0
|
64 |
+
if (previous_state['timestep'] > params['timestep_switch']) and (previous_state['funding_pool'] > params['funding_round']):
|
65 |
+
funding = params['funding_round']
|
66 |
+
to_treasury -= funding
|
67 |
+
research_value = funding * (1-params['epsilon'])
|
68 |
+
losses += funding - research_value
|
69 |
+
|
70 |
+
salary = research_value * params['beta']
|
71 |
+
to_market = research_value
|
72 |
+
if (random() < params['probability_buying']) and (previous_state['researcher2_value'] > params['cost_buying']):
|
73 |
+
salary = salary - params['cost_buying']
|
74 |
+
tx_fee = params['cost_buying'] * params['tx_fee']
|
75 |
+
salary -= tx_fee
|
76 |
+
to_treasury += tx_fee
|
77 |
+
to_other_researcher += params['cost_buying']
|
78 |
+
to_researcher += salary + research_value
|
79 |
+
elif (previous_state['timestep'] < params['timestep_switch']) and (previous_state['researcher2_value'] > params['cost_buying']):
|
80 |
+
tx_fee = params['cost_buying'] * params['tx_fee']
|
81 |
+
to_researcher -= params['cost_buying'] - tx_fee
|
82 |
+
to_other_researcher += params['cost_buying']
|
83 |
+
to_treasury += tx_fee
|
84 |
+
return {'update_researcher2_funding': funding,
|
85 |
+
'update_researcher2_salary': salary,
|
86 |
+
'update_researcher2_value': to_researcher,
|
87 |
+
'update_funding_pool': to_treasury,
|
88 |
+
'update_market': to_market,
|
89 |
+
'update_researcher1_value': to_other_researcher,
|
90 |
+
'update_losses': losses}
|
91 |
+
|
92 |
+
def s_timestep(params, substep, state_history, previous_state, policy_input):
|
93 |
+
updated_timestep = previous_state['timestep'] + 1
|
94 |
+
return 'timestep', updated_timestep
|
95 |
+
|
96 |
+
def s_funding_pool(params, substep, state_history, previous_state, policy_input):
|
97 |
+
funding_pool = previous_state['funding_pool']
|
98 |
+
updated_funding_pool = funding_pool + policy_input['update_funding_pool']
|
99 |
+
return 'funding_pool', updated_funding_pool
|
100 |
+
|
101 |
+
def s_researcher1_value(params, substep, state_history, previous_state, policy_input):
|
102 |
+
r_value = previous_state['researcher1_value']
|
103 |
+
updated_researcher1_value = r_value + policy_input['update_researcher1_value']
|
104 |
+
return 'researcher1_value', updated_researcher1_value
|
105 |
+
|
106 |
+
def s_researcher1_funding(params, substep, state_history, previous_state, policy_input):
|
107 |
+
r_funding = previous_state['researcher1_funding']
|
108 |
+
updated_researcher1_funding = r_funding + policy_input['update_researcher1_funding']
|
109 |
+
return 'researcher1_funding', updated_researcher1_funding
|
110 |
+
|
111 |
+
def s_researcher1_salary(params, substep, state_history, previous_state, policy_input):
|
112 |
+
r_salary = previous_state['researcher1_salary']
|
113 |
+
updated_researcher1_salary = r_salary + policy_input['update_researcher1_salary']
|
114 |
+
return 'researcher1_salary', updated_researcher1_salary
|
115 |
+
|
116 |
+
def s_researcher2_value(params, substep, state_history, previous_state, policy_input):
|
117 |
+
r_value = previous_state['researcher2_value']
|
118 |
+
updated_researcher2_value = r_value + policy_input['update_researcher2_value']
|
119 |
+
return 'researcher2_value', updated_researcher2_value
|
120 |
+
|
121 |
+
def s_researcher2_funding(params, substep, state_history, previous_state, policy_input):
|
122 |
+
r_funding = previous_state['researcher2_funding']
|
123 |
+
updated_researcher2_funding = r_funding + policy_input['update_researcher2_funding']
|
124 |
+
return 'researcher2_funding', updated_researcher2_funding
|
125 |
+
|
126 |
+
def s_researcher2_salary(params, substep, state_history, previous_state, policy_input):
|
127 |
+
r_salary = previous_state['researcher2_salary']
|
128 |
+
updated_researcher2_salary = r_salary + policy_input['update_researcher2_salary']
|
129 |
+
return 'researcher2_salary', updated_researcher2_salary
|
130 |
+
|
131 |
+
def s_knowledge_market(params, substep, state_history, previous_state, policy_input):
|
132 |
+
value = previous_state['knowledge_market_value']
|
133 |
+
updated_market_value = value + policy_input['update_market']
|
134 |
+
return 'knowledge_market_value', updated_market_value
|
135 |
+
|
136 |
+
def s_losses(params, substep, state_history, previous_state, policy_input):
|
137 |
+
losses = previous_state['losses']
|
138 |
+
updated_losses = losses + policy_input['update_losses']
|
139 |
+
return 'losses', updated_losses
|
140 |
+
|
141 |
+
st.subheader('Initial Value Allocation')
|
142 |
+
funding_pool = st.slider('Initial Funding Pool', min_value=1000, max_value=10000, value=1000, step=10)
|
143 |
+
researcher1_value = st.slider('Researcher1 Tokens', min_value=0, max_value=1000, value=0, step=1)
|
144 |
+
researcher2_value = st.slider('Researcher2 Tokens', min_value=0, max_value=1000, value=0, step=1)
|
145 |
+
st.subheader('Simulation Parameters')
|
146 |
+
tx_fee = st.slider('Transaction fee collected by DAO treasury during each transaction in the knowledge market', min_value=0., max_value=1., value=0.1, step=0.0001)
|
147 |
+
st.write('Set the funding disbursed each round from the funding pool')
|
148 |
+
funding_round = st.slider('Funding Round', min_value=100, max_value=1000, value=100, step=1)
|
149 |
+
st.write('Set the relative value leakages in the model.')
|
150 |
+
epsilon = st.slider('Work Inefficiency Weight', min_value=0., max_value=1., value=0.1, step=0.0001)
|
151 |
+
st.write('Set the portion of grant funding to be used as researcher salary.')
|
152 |
+
beta = st.slider('Salary Weight', min_value=0., max_value=1., value=0.4, step=0.0001)
|
153 |
+
st.write('Set the cost of getting access to papers in the knowledge market.')
|
154 |
+
cost_buying = st.slider('Cost of Buying', min_value=10., max_value=100., value=10., step=0.1)
|
155 |
+
st.write('Set the probability a researcher will buy access to a paper at each timestep.')
|
156 |
+
probability_buying = st.slider('Researcher Probability of Buying', min_value=0., max_value=1., value=0.1, step=0.0001)
|
157 |
+
st.write('Set the number of timesteps in the simulation.')
|
158 |
+
timesteps = st.slider('Timesteps', min_value=10, max_value=1000, value=100, step=1)
|
159 |
+
|
160 |
+
initial_state = {
|
161 |
+
'funding_pool': funding_pool,
|
162 |
+
'researcher1_value': researcher1_value,
|
163 |
+
'researcher1_funding': 0,
|
164 |
+
'researcher1_salary': 0,
|
165 |
+
'researcher2_value': researcher2_value,
|
166 |
+
'researcher2_funding': 0,
|
167 |
+
'researcher2_salary': 0,
|
168 |
+
'knowledge_market_value': 0,
|
169 |
+
'timestep': 0,
|
170 |
+
'losses': 0
|
171 |
+
}
|
172 |
+
ts = int(timesteps/2)
|
173 |
+
|
174 |
+
system_params = {
|
175 |
+
'funding_pool': [funding_pool],
|
176 |
+
'funding_round': [funding_round],
|
177 |
+
'beta': [beta],
|
178 |
+
'epsilon': [epsilon],
|
179 |
+
'cost_buying': [cost_buying],
|
180 |
+
'probability_buying': [probability_buying],
|
181 |
+
'timestep_switch': [ts],
|
182 |
+
'tx_fee': [tx_fee]
|
183 |
+
}
|
184 |
+
|
185 |
+
def configure_and_run_experiment(initial_state,
|
186 |
+
partial_state_update_blocks,
|
187 |
+
timesteps):
|
188 |
+
model = Model(
|
189 |
+
# Model initial state
|
190 |
+
initial_state=initial_state,
|
191 |
+
# Model Partial State Update Blocks
|
192 |
+
state_update_blocks=partial_state_update_blocks,
|
193 |
+
# System Parameters
|
194 |
+
params=system_params
|
195 |
+
)
|
196 |
+
simulation = Simulation(
|
197 |
+
model=model,
|
198 |
+
timesteps=timesteps, # Number of timesteps
|
199 |
+
runs=1 # Number of Monte Carlo Runs
|
200 |
+
)
|
201 |
+
|
202 |
+
result = simulation.run()
|
203 |
+
return result
|
204 |
+
|
205 |
+
partial_state_update_blocks = [
|
206 |
+
{
|
207 |
+
'policies': {
|
208 |
+
'p_researcher1': p_researcher1,
|
209 |
+
'p_researcher2': p_researcher2
|
210 |
+
},
|
211 |
+
'variables': {
|
212 |
+
'timestep': s_timestep,
|
213 |
+
'funding_pool': s_funding_pool,
|
214 |
+
'researcher1_value': s_researcher1_value,
|
215 |
+
'researcher1_funding': s_researcher1_funding,
|
216 |
+
'researcher1_salary': s_researcher1_salary,
|
217 |
+
'researcher2_value': s_researcher2_value,
|
218 |
+
'researcher2_funding': s_researcher2_funding,
|
219 |
+
'researcher2_salary': s_researcher2_salary,
|
220 |
+
'knowledge_market_value': s_knowledge_market,
|
221 |
+
'losses': s_losses
|
222 |
+
}
|
223 |
+
}
|
224 |
+
]
|
225 |
+
|
226 |
+
if st.button('Run Simulation'):
|
227 |
+
raw_result = configure_and_run_experiment(initial_state, partial_state_update_blocks, timesteps)
|
228 |
+
df = pd.DataFrame(raw_result)
|
229 |
+
fig1 = df.plot(kind='line', x='timestep', y=['funding_pool', 'researcher1_value', 'researcher2_value'], width=1000)
|
230 |
+
fig2 = df.plot(kind='line', x='timestep', y=['funding_pool','knowledge_market_value'], width=1000)
|
231 |
+
fig3 = df.plot(kind='line', x='timestep', y=['funding_pool', 'losses'], width=1000)
|
232 |
+
fig4 = df.plot(kind='line', x='timestep', y=['researcher1_value', 'researcher2_value', 'losses'], width=1000)
|
233 |
+
st.subheader('Results')
|
234 |
+
st.plotly_chart(fig1)
|
235 |
+
st.plotly_chart(fig2)
|
236 |
+
st.plotly_chart(fig3)
|
237 |
+
st.plotly_chart(fig4)
|