Merge branch 'master' of https://github.com/BlockScience/Community_Inclusion_Currencies

README.md

@@ -1,9 +1,16 @@
 # Community_Inclusion_Currencies
-Repository for Complex Systems model of the Grassroots Economics Community Inclusion Currency project.
+Repository for Complex Systems model of the Grassroots Economics Community Inclusion Currency project. The Colab notebooks are able to be run and played with by anyone who uses the link. 
 
 ### Colab
-[Click here to get to interactive colab](https://colab.research.google.com/github/BlockScience/Community_Inclusion_Currencies/blob/master/Colab/CIC_Network_cadCAD_model.ipynb#scrollTo=UZEO2sP9Mhi-)
+[Click here to get to an interactive notebook](https://colab.research.google.com/github/BlockScience/Community_Inclusion_Currencies/blob/master/Colab/CIC_Network_cadCAD_model.ipynb#scrollTo=UZEO2sP9Mhi-)
+
+[Click here to for an interactive notebook with a parameter sweep](https://colab.research.google.com/github/BlockScience/Community_Inclusion_Currencies/blob/master/Colab/CIC_Network_cadCAD_model_params.ipynb)
+
 ### Simulation work
 [Click here](Simulation/CIC_Network_cadCAD_model.ipynb)
+
+#### Parameter sweep 
+[Click here](Simulation_param/CIC_Network_cadCAD_model_params.ipynb)
+
 ### Theory work
 [Click here](Theory/cic_initialization.ipynb)

Simulation_param/model/economyconfig.py

@@ -0,0 +1,36 @@
+import math
+from decimal import Decimal
+from datetime import timedelta
+import numpy as np
+from typing import Dict, List
+
+from cadCAD.configuration import append_configs
+from cadCAD.configuration.utils import bound_norm_random, ep_time_step, config_sim, access_block
+
+from .genesis_states import genesis_states
+from .partial_state_update_block import partial_state_update_block
+
+params: Dict[str, List[int]] = {
+    'drip_frequency': [30,60,90] # in days
+}
+
+
+sim_config = config_sim({
+    'N': 5,
+    'T': range(100), #day 
+    'M': params,
+})
+
+seeds = {
+    'p': np.random.RandomState(26042019),
+}
+env_processes = {}
+
+
+append_configs(
+    sim_configs=sim_config,
+    initial_state=genesis_states,
+    seeds=seeds,
+    env_processes=env_processes,
+    partial_state_update_blocks=partial_state_update_block
+)

Simulation_param/model/genesis_states.py

@@ -0,0 +1,23 @@
+from .parts.initialization import * 
+import pandas as pd
+
+genesis_states = { 
+    # initial states of the economy
+    'network': create_network(),# networkx market
+    'KPIDemand': {},
+    'KPISpend': {},
+    'KPISpendOverDemand': {},
+    'VelocityOfMoney':0,
+    'startingBalance': {},
+    '30_day_spend': {},
+    'withdraw':{},
+    'outboundAgents':[],
+    'inboundAgents':[],
+    'operatorFiatBalance': R0,
+    'operatorCICBalance': S0,
+    'fundsInProcess': {'timestep':[],'decision':[],'cic':[],'shilling':[]},
+    'totalDistributedToAgents':0,
+    'totalMinted':0,
+    'totalBurned':0
+}
+

Simulation_param/model/partial_state_update_block.py

@@ -0,0 +1,89 @@
+from .parts.exogenousProcesses import *
+from .parts.kpis import *
+from .parts.system import *
+from .parts.operatorentity import *
+
+partial_state_update_block = {
+    # Exogenous
+    'Exogenous': {
+        'policies': {
+        },
+        'variables': {
+            'startingBalance': startingBalance,
+            'operatorFiatBalance': redCrossDrop,
+            '30_day_spend': update_30_day_spend,
+            'network':clear_agent_activity
+        }
+    },
+    # Users
+    'Behaviors': {
+        'policies': {
+            'action': choose_agents
+        },
+        'variables': {
+        'network': update_agent_activity,
+        'outboundAgents': update_outboundAgents,
+        'inboundAgents':update_inboundAgents
+        }
+    },
+    'Spend allocation': {
+        'policies': {
+            'action': spend_allocation
+        },
+        'variables': {
+        'network': update_node_spend
+        }
+    },
+    'Withdraw behavior': {
+        'policies': {
+            'action': withdraw_calculation
+        },
+        'variables': {
+        'withdraw': update_withdraw,
+        'network':update_network_withraw
+        }
+    },
+    # Operator
+    'Operator Disburse to Agents': {
+        'policies': {
+            'action': disbursement_to_agents
+        },
+        'variables': {
+        'network':update_agent_tokens,
+        'operatorCICBalance':update_operator_FromDisbursements,
+        'totalDistributedToAgents':update_totalDistributedToAgents
+        }
+    },
+    'Operator Inventory Control': {
+        'policies': {
+            'action': inventory_controller
+        },
+        'variables': {
+        'operatorFiatBalance':update_operator_fiatBalance,
+        'operatorCICBalance':update_operator_cicBalance, 
+        'totalMinted': update_totalMinted,
+        'totalBurned':update_totalBurned,
+        'fundsInProcess':update_fundsInProcess
+        }
+    },
+    # KPIs
+    'KPIs': {
+        'policies': {
+            'action':kpis
+        },
+        'variables':{
+            'KPIDemand': update_KPIDemand,
+            'KPISpend': update_KPISpend,
+            'KPISpendOverDemand': update_KPISpendOverDemand        
+        }
+    },
+    'Velocity': {
+        'policies': {
+            'action':velocity_of_money
+        },
+        'variables':{
+
+            'VelocityOfMoney': update_velocity_of_money
+        }
+    }
+}

Simulation_param/model/parts/exogenousProcesses.py

@@ -0,0 +1,122 @@
+
+import numpy as np
+import pandas as pd
+from .initialization import *
+from .supportingFunctions import *
+    
+
+def startingBalance(params, step, sL, s, _input):
+    '''
+    Calculate agent starting balance every 30 days
+    '''
+    y = 'startingBalance'
+    network = s['network']
+
+    startingBalance = {}
+
+    timestep = s['timestep']
+
+    division =  timestep % 31  == 0
+
+    if timestep == 1:
+        for i in agents:
+            startingBalance[i] = network.nodes[i]['tokens']
+    elif division == True:
+        for i in agents:
+            startingBalance[i] = network.nodes[i]['tokens']
+    else:
+        startingBalance = s['startingBalance']
+    x = startingBalance
+
+    return (y, x)
+
+def update_30_day_spend(params, step, sL, s,_input):
+    '''
+    Aggregate agent spend. Refresh every 30 days.
+    '''
+    y = '30_day_spend'
+    network = s['network']
+
+    timestep = s['timestep']
+
+    division =  timestep % 31  == 0
+
+    if division == True:
+        outflowSpend, inflowSpend = iterateEdges(network,'spend')
+        spend =  outflowSpend 
+    else:
+        spendOld = s['30_day_spend']
+        outflowSpend, inflowSpend = iterateEdges(network,'spend')
+        spend = DictionaryMergeAddition(spendOld,outflowSpend) 
+
+    x = spend
+    return (y, x)
+
+def redCrossDrop(params, step, sL, s, _input):
+    '''
+    Every 30 days, the red cross drips to the grassroots operator node
+    '''
+    y = 'operatorFiatBalance'
+    fiatBalance = s['operatorFiatBalance']
+    
+    timestep = s['timestep']
+    
+    division =  timestep % params['drip_frequency']  == 0
+
+    if division == True:
+        fiatBalance = fiatBalance + drip
+    else:
+        pass
+
+    x = fiatBalance
+    return (y, x)
+
+
+def clear_agent_activity(params,step,sL,s,_input):
+    '''
+    Clear agent activity from the previous timestep
+    '''
+    y = 'network'
+    network = s['network']
+
+    if s['timestep'] > 0:
+        outboundAgents = s['outboundAgents']
+        inboundAgents = s['inboundAgents']
+        
+        try:
+            for i,j in zip(outboundAgents,inboundAgents):
+                network[i][j]['demand'] = 0
+        except:
+            pass
+
+        # Clear cic % demand edge weights
+        try:
+            for i,j in zip(outboundAgents,inboundAgents):
+                network[i][j]['fractionOfDemandInCIC'] = 0
+        except:
+            pass
+
+
+        # Clear utility edge types
+        try: 
+            for i,j in zip(outboundAgents,inboundAgents):
+                network[i][j]['utility'] = 0
+        except:
+            pass
+        
+        # Clear cic % spend edge weights
+        try:
+            for i,j in zip(outboundAgents,inboundAgents):
+                network[i][j]['fractionOfActualSpendInCIC'] = 0
+        except:
+            pass
+        # Clear spend edge types
+        try: 
+            for i,j in zip(outboundAgents,inboundAgents):
+                network[i][j]['spend'] = 0
+        except:
+            pass
+    else:
+        pass
+    x = network
+    return (y,x)

Simulation_param/model/parts/initialization.py

@@ -0,0 +1,118 @@
+
+# import libraries
+import networkx as nx
+import matplotlib.pyplot as plt
+import numpy as np
+from .supportingFunctions import *
+
+# Assumptions:
+# Amount received in shilling when withdraw occurs
+leverage = 1 
+
+# process time
+process_lag = 7 # timesteps
+
+# red cross drip amount
+drip = 4000
+
+# system initialization
+agents = ['a','b','c','d','e','f','g','h','i','j','k','l','m','o','p']
+
+# system actors
+system = ['external','cic']
+
+# chamas
+chama = ['chama_1','chama_2','chama_3','chama_4']
+
+# traders
+traders = ['ta','tb','tc'] #only trading on the cic. Link to external and cic not to other agents
+
+allAgents = agents + system
+
+mixingAgents = ['a','b','c','d','e','f','g','h','i','j','k','l','m','o','p','external']
+
+UtilityTypesOrdered ={'Food/Water':1,
+                    'Fuel/Energy':2,
+                    'Health':3,
+                    'Education':4,
+                    'Savings Group':5,
+                    'Shop':6}
+
+utilityTypesProbability = {'Food/Water':0.6,
+                    'Fuel/Energy':0.10,
+                    'Health':0.03,
+                    'Education':0.015,
+                    'Savings Group':0.065,
+                    'Shop':0.19}
+
+
+R0 =  500 #thousand xDAI
+kappa = 4 #leverage
+P0 = 1/100 #initial price
+S0 = kappa*R0/P0
+V0 = invariant(R0,S0,kappa)
+P = spot_price(R0, V0, kappa)
+
+# Price level
+priceLevel = 100
+
+fractionOfDemandInCIC = 0.5
+fractionOfActualSpendInCIC = 0.5
+
+def create_network():
+    # Create network graph
+    network = nx.DiGraph()
+
+    # Add nodes for n participants plus the external economy and the cic network
+    for i in agents:
+        network.add_node(i,type='Agent',tokens=400, native_currency = int(np.random.uniform(low=20, high=500, size=1)[0]))
+        
+        
+    network.add_node('external',type='Contract',native_currency = 100000000,tokens = 0,delta_native_currency = 0, pos=(1,50))
+    network.add_node('cic',type='Contract',tokens= S0, native_currency = R0,pos=(50,1))
+
+    for i in chama:
+        network.add_node(i,type='Chama')
+        
+    for i in traders:
+        network.add_node(i,type='Trader',tokens=20, native_currency = 20, 
+                        price_belief = 1, trust_level = 1)
+        
+    # Create bi-directional edges between all participants
+    for i in allAgents:
+        for j in allAgents:
+            if i!=j:
+                network.add_edge(i,j)
+
+    # Create bi-directional edges between each trader and the external economy and the cic environment            
+    for i in traders:
+        for j in system:
+            if i!=j:
+                network.add_edge(i,j)
+                
+    # Create bi-directional edges between some agent and a chama node representing membershio      
+    for i in chama:
+        for j in agents:
+            if np.random.choice(['Member','Non_Member'],1,p=[.50,.50])[0] == 'Member':
+                network.add_edge(i,j)
+
+    # Type colors 
+    colors = ['Red','Blue','Green','Orange']
+    color_map = []
+    for i in network.nodes:
+        if network.nodes[i]['type'] == 'Agent':
+            color_map.append('Red')
+        elif network.nodes[i]['type'] == 'Cloud':
+            color_map.append('Blue')
+        elif network.nodes[i]['type'] == 'Contract':
+            color_map.append('Green')
+        elif network.nodes[i]['type'] == 'Trader':
+            color_map.append('Yellow')
+        elif network.nodes[i]['type'] == 'Chama':
+            color_map.append('Orange')
+            
+    pos = nx.spring_layout(network,pos=nx.get_node_attributes(network,'pos'),fixed=nx.get_node_attributes(network,'pos'),seed=10)
+    nx.draw(network,node_color = color_map,pos=pos,with_labels=True,alpha=0.7)
+    plt.savefig('images/graph.png')
+    plt.show()
+    return network

Simulation_param/model/parts/kpis.py

@@ -0,0 +1,92 @@
+
+import numpy as np
+from .initialization import *
+from .supportingFunctions import *
+import networkx as nx
+
+
+# Behaviors
+def kpis(params, step, sL, s):
+    ''''''
+    # instantiate network state
+    network = s['network']
+
+    KPIDemand = {}
+    KPISpend = {}
+    KPISpendOverDemand = {}
+    for i in mixingAgents:
+        demand = []
+        for j in network.adj[i]:
+            try:
+                demand.append(network.adj[i][j]['demand'])
+            except:
+                pass
+
+        spend = []
+        for j in network.adj[i]:
+            try:
+                spend.append(network.adj[i][j]['spend'])
+            except:
+                pass
+
+        sumDemand = sum(demand)
+        sumSpend = sum(spend)
+        try:
+            spendOverDemand = sumSpend/sumDemand
+        except:
+            spendOverDemand = 0
+
+        KPIDemand[i] = sumDemand
+        KPISpend[i] = sumSpend
+        KPISpendOverDemand[i] = spendOverDemand
+
+    #print(nx.katz_centrality_numpy(G=network,weight='spend'))
+    return {'KPIDemand':KPIDemand,'KPISpend':KPISpend,'KPISpendOverDemand':KPISpendOverDemand}
+
+def velocity_of_money(params, step, sL, s):
+    ''''''
+    # instantiate network state
+    network = s['network']
+
+    KPISpend = s['KPISpend']
+
+    # TODO: Moving average for state variable
+    T = []
+    for i,j in KPISpend.items():
+        T.append(j)
+        
+    T = sum(T)
+    
+    # TODO Moving average for state variable 
+    M = []
+    for i in agents:
+        M.append(network.nodes[i]['tokens'] + network.nodes[i]['native_currency'])
+        
+    M = sum(M)
+    
+    V_t = (priceLevel *T)/M
+
+    return {'V_t':V_t,'T':T,'M':M}
+
+
+# Mechanisms
+def update_KPIDemand(params, step, sL, s,_input):
+    y = 'KPIDemand'
+    x = _input['KPIDemand']
+    return (y,x)
+
+def update_KPISpend(params, step, sL, s,_input):
+    y = 'KPISpend'
+    x = _input['KPISpend']
+    return (y,x)
+
+def update_KPISpendOverDemand(params, step, sL, s,_input):
+    y = 'KPISpendOverDemand'
+    x = _input['KPISpendOverDemand']
+    return (y,x)
+
+
+def update_velocity_of_money(params, step, sL, s,_input):
+    y = 'VelocityOfMoney'
+    x = _input['V_t']
+    return (y,x)

Simulation_param/model/parts/operatorentity.py

@@ -0,0 +1,287 @@
+
+import numpy as np
+import pandas as pd
+from cadCAD.configuration.utils import access_block
+from .initialization import *
+from .supportingFunctions import *
+from collections import OrderedDict
+
+# Parameters
+FrequencyOfAllocation = 45 # every two weeks
+idealFiat = 5000
+idealCIC = 200000
+varianceCIC = 50000
+varianceFiat = 1000
+unadjustedPerAgent = 50
+
+
+
+
+agentAllocation = {'a':[1,1],'b':[1,1],'c':[1,1], # agent:[centrality,allocationValue]
+                   'd':[1,1],'e':[1,1],'f':[1,1],
+                   'g':[1,1],'h':[1,1],'i':[1,1],
+                   'j':[1,1],'k':[1,1],'l':[1,1],
+                   'm':[1,1],'o':[1,1],'p':[1,1]}
+
+# Behaviors
+def disbursement_to_agents(params, step, sL, s):
+    '''
+    Distribute every FrequencyOfAllocation days to agents based off of centrality allocation metric
+    '''
+    fiatBalance = s['operatorFiatBalance']
+    cicBalance = s['operatorCICBalance']
+    timestep = s['timestep']
+
+    division =  timestep % FrequencyOfAllocation  == 0
+
+    if division == True:
+        agentDistribution ={} # agent: amount distributed
+        for i,j in agentAllocation.items():
+            agentDistribution[i] = unadjustedPerAgent * agentAllocation[i][1]
+        distribute = 'Yes'
+        
+    else:
+        agentDistribution = 0
+        distribute = 'No'
+
+
+    return {'distribute':distribute,'amount':agentDistribution}
+
+
+def inventory_controller(params, step, sL, s):
+    '''
+    Monetary policy hysteresis conservation allocation between fiat and cic reserves.
+    
+    # TODO: If scarcity on both sides, add feedback to reduce percentage able to withdraw, frequency you can redeem, or redeem at less than par. 
+    '''
+    fiatBalance = s['operatorFiatBalance']
+    cicBalance = s['operatorCICBalance']
+    timestep = s['timestep']
+    fundsInProcess = s['fundsInProcess']
+
+
+    updatedCIC = cicBalance
+    updatedFiat = fiatBalance
+
+    #decision,amt = mint_burn_logic_control(idealCIC,updatedCIC,variance,updatedFiat)
+    decision,amt = mint_burn_logic_control(idealCIC,updatedCIC,varianceCIC,updatedFiat,varianceFiat,idealFiat)
+
+    if decision == 'burn':
+        try:
+            deltaR, realized_price = withdraw(amt,updatedFiat,updatedCIC, V0, kappa)
+            # update state
+            # fiatBalance = fiatBalance - deltaR
+            # cicBalance = cicBalance - amt
+            fiatChange = abs(deltaR)
+            cicChange = amt
+
+        except:
+            print('Not enough to burn')
+
+            fiatChange = 0
+            cicChange = 0
+        
+    elif decision == 'mint':
+        try:
+            deltaS, realized_price = mint(amt,updatedFiat,updatedCIC, V0, kappa)
+            # update state
+            # fiatBalance = fiatBalance + amt
+            # cicBalance = cicBalance + deltaS
+            fiatChange = amt
+            cicChange = abs(deltaS)
+
+        except:
+            print('Not enough to mint')
+            fiatChange = 0
+            cicChange = 0
+
+    else:
+        fiatChange = 0
+        cicChange = 0
+        decision = 'none'
+        pass
+
+    if decision == 'mint':
+        fundsInProcess['timestep'].append(timestep + process_lag)
+        fundsInProcess['decision'].append(decision)
+        fundsInProcess['cic'].append(fiatChange)
+        fundsInProcess['shilling'].append(cicChange)
+    elif decision == 'burn':
+        fundsInProcess['timestep'].append(timestep +process_lag)
+        fundsInProcess['decision'].append(decision)
+        fundsInProcess['cic'].append(fiatChange)
+        fundsInProcess['shilling'].append(cicChange)
+    else:
+        pass
+    
+    return {'decision':decision,'fiatChange':fiatChange,'cicChange':cicChange,'fundsInProcess':fundsInProcess}
+
+
+
+# Mechanisms 
+def update_agent_tokens(params,step,sL,s,_input):
+    '''
+    '''
+    y = 'network'
+    network = s['network']
+
+    distribute = _input['distribute']
+    amount = _input['amount']
+
+    if distribute == 'Yes':
+        for i in agents:
+            network.nodes[i]['tokens'] = network.nodes[i]['tokens'] + amount[i]
+    else:
+        pass
+
+    return (y,network)
+
+def update_operator_FromDisbursements(params,step,sL,s,_input):
+    '''
+    '''
+    y = 'operatorCICBalance'
+    x = s['operatorCICBalance']
+    timestep = s['timestep']
+    
+    distribute = _input['distribute']
+    amount = _input['amount'] 
+
+    if distribute == 'Yes':
+        totalDistribution = []
+        for i,j in amount.items():
+            totalDistribution.append(j)
+        
+        totalDistribution = sum(totalDistribution)
+        x = x - totalDistribution
+
+    else:
+        pass
+
+    return (y,x)
+
+def update_totalDistributedToAgents(params,step,sL,s,_input):
+    '''
+    '''
+    y = 'totalDistributedToAgents'
+    x = s['totalDistributedToAgents']
+    timestep = s['timestep']
+    
+    distribute = _input['distribute']
+    amount = _input['amount'] 
+
+    if distribute == 'Yes':
+        totalDistribution = []
+        for i,j in amount.items():
+            totalDistribution.append(j)
+        
+        totalDistribution = sum(totalDistribution)
+        x = x + totalDistribution
+    else:
+        pass
+
+    return (y,x)
+
+def update_operator_fiatBalance(params,step,sL,s,_input):
+    '''
+    '''
+    y = 'operatorFiatBalance'
+    x = s['operatorFiatBalance']
+    fundsInProcess = s['fundsInProcess']
+    timestep = s['timestep']
+    if _input['fiatChange']:
+        try:
+            if fundsInProcess['timestep'][0] == timestep + 1:
+                if fundsInProcess['decision'][0] == 'mint':
+                    x = x - abs(fundsInProcess['shilling'][0])
+                elif fundsInProcess['decision'][0] == 'burn':
+                    x = x + abs(fundsInProcess['shilling'][0])
+            else:
+                pass
+        except:
+            pass
+    else:
+        pass
+
+
+    return (y,x)
+
+def update_operator_cicBalance(params,step,sL,s,_input):
+    '''
+    '''
+    y = 'operatorCICBalance'
+    x = s['operatorCICBalance']
+    fundsInProcess = s['fundsInProcess']
+    timestep = s['timestep']
+
+    if _input['cicChange']:
+        try:
+            if fundsInProcess['timestep'][0] == timestep + 1:
+                if fundsInProcess['decision'][0] == 'mint':
+                    x = x + abs(fundsInProcess['cic'][0])
+                elif fundsInProcess['decision'][0] == 'burn':
+                    x = x - abs(fundsInProcess['cic'][0])
+            else:
+                pass
+        except:
+            pass
+    else:
+        pass
+
+    return (y,x)
+
+def update_totalMinted(params,step,sL,s,_input):
+    '''
+    '''
+    y = 'totalMinted'
+    x = s['totalMinted']
+    timestep = s['timestep']
+    try:
+        if _input['fundsInProcess']['decision'][0] == 'mint':
+            x = x + abs(_input['fundsInProcess']['cic'][0])
+        elif _input['fundsInProcess']['decision'][0] == 'burn':
+            pass
+    except:
+        pass
+
+
+    return (y,x)
+
+def update_totalBurned(params,step,sL,s,_input):
+    '''
+    '''
+    y = 'totalBurned'
+    x = s['totalBurned']
+    timestep = s['timestep']
+    try:
+        if _input['fundsInProcess']['decision'][0] == 'burn':
+            x = x + abs(_input['fundsInProcess']['cic'][0])
+        elif _input['fundsInProcess']['decision'][0] == 'mint':
+            pass
+    except:
+        pass
+
+    return (y,x)
+
+def update_fundsInProcess(params,step,sL,s,_input):
+    '''
+    '''
+    y = 'fundsInProcess'
+    x = _input['fundsInProcess']
+    timestep = s['timestep']
+
+    if _input['fundsInProcess']:
+        try:
+            if x['timestep'][0] == timestep:
+                del x['timestep'][0]
+                del x['decision'][0]
+                del x['cic'][0]
+                del x['shilling'][0]
+            else:
+                pass
+        except:
+            pass
+    else:
+        pass
+
+    return (y,x)
+

Simulation_param/model/parts/supportingFunctions.py

@@ -0,0 +1,442 @@
+import numpy as np
+from scipy.stats import gamma
+import matplotlib.pyplot as plt
+import seaborn as sns
+import pandas as pd
+
+default_kappa= 4
+default_exit_tax = .02
+
+#value function for a given state (R,S)
+def invariant(R,S,kappa=default_kappa):
+    
+    return (S**kappa)/R
+
+#given a value function (parameterized by kappa)
+#and an invariant coeficient V0
+#return Supply S as a function of reserve R
+def reserve(S, V0, kappa=default_kappa):
+    return (S**kappa)/V0
+
+#given a value function (parameterized by kappa)
+#and an invariant coeficient V0
+#return Supply S as a function of reserve R
+def supply(R, V0, kappa=default_kappa):
+    return (V0*R)**(1/kappa)
+
+#given a value function (parameterized by kappa)
+#and an invariant coeficient V0
+#return a spot price P as a function of reserve R
+def spot_price(R, V0, kappa=default_kappa):
+    return kappa*R**((kappa-1)/kappa)/V0**(1/kappa)
+
+#for a given state (R,S)
+#given a value function (parameterized by kappa)
+#and an invariant coeficient V0
+#deposit deltaR to Mint deltaS
+#with realized price deltaR/deltaS
+def mint(deltaR, R,S, V0, kappa=default_kappa):
+    deltaS = (V0*(R+deltaR))**(1/kappa)-S
+    if deltaS ==0:
+        realized_price = spot_price(R+deltaR, V0, kappa)
+    else:
+        realized_price = deltaR/deltaS
+    deltaS = round(deltaS,2)
+    return deltaS, realized_price
+
+#for a given state (R,S)
+#given a value function (parameterized by kappa)
+#and an invariant coeficient V0
+#burn deltaS to Withdraw deltaR
+#with realized price deltaR/deltaS
+def withdraw(deltaS, R,S, V0, kappa=default_kappa):
+    deltaR = R-((S-deltaS)**kappa)/V0
+    if deltaS ==0:
+        realized_price = spot_price(R+deltaR, V0, kappa)
+    else:
+        realized_price = deltaR/deltaS
+    deltaR = round(deltaR,2)
+    return deltaR, realized_price
+
+
+
+def iterateEdges(network,edgeToIterate):
+    '''
+    Description:
+    Iterate through a network on a weighted edge and return
+    two dictionaries: the inflow and outflow for the given agents
+    in the format:
+    
+    {'Agent':amount}
+    '''
+    outflows = {}
+    inflows = {}
+    for i,j in network.edges:
+        try:
+            amount = network[i][j][edgeToIterate]
+            if i in outflows:
+                outflows[i] = outflows[i] + amount
+            else:
+                outflows[i] =  amount
+            if j in inflows:
+                inflows[j] = inflows[j] + amount
+            else:
+                inflows[j] = amount
+        except:
+            pass
+    return outflows,inflows
+
+
+def inflowAndOutflowDictionaryMerge(inflow,outflow):
+    '''
+    Description:
+    Merge two dictionaries and return one dictionary with zero floor'''
+    
+    merged = {}
+
+    inflowsKeys = [k for k,v in inflow.items() if k not in outflow]
+    for i in inflowsKeys:
+        merged[i] = inflow[i]
+    outflowsKeys = [k for k,v in outflow.items() if k not in inflow]
+    for i in outflowsKeys:
+        merged[i] = outflow[i]
+    overlapKeys = [k for k,v in inflow.items() if k in outflow]
+    for i in overlapKeys:
+        amt = outflow[i] - inflow[i] 
+        if amt < 0:
+            merged[i] = 0
+        else:
+            merged[i] = amt
+            pass
+        
+    return merged
+
+        
+def spendCalculation(agentToPay,agentToReceive,rankOrderDemand,maxSpendCurrency,maxSpendTokens,cicPercentage):
+    '''
+    Function to calculate if an agent can pay for demand given token and currency contraints
+    '''
+    if (rankOrderDemand[agentToReceive] * (1-cicPercentage)) > maxSpendCurrency[agentToPay]:
+        verdict_currency = 'No'
+    else:
+        verdict_currency = 'Enough'
+        
+    if (rankOrderDemand[agentToReceive] * cicPercentage) > maxSpendTokens[agentToPay]:
+        verdict_cic = 'No'
+    else:
+        verdict_cic = 'Enough'
+    
+    if verdict_currency == 'Enough'and verdict_cic == 'Enough':
+        spend = rankOrderDemand[agentToReceive]
+    
+    elif maxSpendCurrency[agentToPay] > 0:
+        spend = maxSpendCurrency[agentToPay]
+    else:
+        spend = 0
+        
+    return spend
+
+
+def spendCalculationExternal(agentToPay,agentToReceive,rankOrderDemand,maxSpendCurrency):
+    '''
+    '''
+    if rankOrderDemand[agentToReceive] > maxSpendCurrency[agentToPay]:
+        verdict_currency = 'No'
+    else:
+        verdict_currency = 'Enough'
+    
+    if verdict_currency == 'Enough':
+        spend = rankOrderDemand[agentToReceive]
+        
+    elif maxSpendCurrency[agentToPay] > 0:
+        spend = maxSpendCurrency[agentToPay]
+    else:
+        spend = 0
+        
+    return spend
+
+
+def DictionaryMergeAddition(inflow,outflow):
+    '''
+    Description:
+    Merge two dictionaries and return one dictionary'''
+    
+    merged = {}
+
+    inflowsKeys = [k for k,v in inflow.items() if k not in outflow]
+    for i in inflowsKeys:
+        merged[i] = inflow[i]
+    outflowsKeys = [k for k,v in outflow.items() if k not in inflow]
+    for i in outflowsKeys:
+        merged[i] = outflow[i]
+    overlapKeys = [k for k,v in inflow.items() if k in outflow]
+    for i in overlapKeys:
+        merged[i] = outflow[i] + inflow[i] 
+        
+    return merged
+
+def mint_burn_logic_control(ideal,actual,variance,fiat,fiat_variance,ideal_fiat):
+    '''
+    Inventory control function to test if the current balance is in an acceptable range. Tolerance range 
+    '''
+    if ideal - variance <= actual <= ideal + (2*variance):
+        decision = 'none'
+        amount = 0
+    else:
+        if (ideal + variance) > actual:
+            decision = 'mint'
+            amount = (ideal + variance) - actual
+        else:
+            pass
+        if actual > (ideal + variance):
+            decision = 'burn'
+            amount = actual - (ideal + variance) 
+        else:
+            pass
+
+    if decision == 'mint':
+        if fiat < (ideal_fiat - fiat_variance):
+            if amount > fiat:
+                decision = 'none'
+                amount = 0
+            else:
+                pass
+    if decision == 'none':
+        if fiat < (ideal_fiat - fiat_variance):
+            decision = 'mint'
+            amount = (ideal_fiat-fiat_variance)
+        else:
+            pass
+    
+    amount = round(amount,2)
+    return decision, amount
+    
+#NetworkX functions
+def get_nodes_by_type(g, node_type_selection):
+    return [node for node in g.nodes if g.nodes[node]['type']== node_type_selection]
+
+def get_edges_by_type(g, edge_type_selection):
+    return [edge for edge in g.edges if g.edges[edge]['type']== edge_type_selection]
+
+def get_edges(g):
+    return [edge for edge in g.edges if g.edges[edge]]
+
+def get_nodes(g):
+    '''
+    df.network.apply(lambda g: np.array([g.nodes[j]['balls'] for j in get_nodes(g)]))
+    '''
+    return [node for node in g.nodes if g.nodes[node]]
+
+def aggregate_runs(df,aggregate_dimension):
+    '''
+    Function to aggregate the monte carlo runs along a single dimension.
+    Parameters:
+    df: dataframe name
+    aggregate_dimension: the dimension you would like to aggregate on, the standard one is timestep.
+    Example run:
+    mean_df,median_df,std_df,min_df = aggregate_runs(df,'timestep')
+    '''
+    df = df[df['substep'] == df.substep.max()]
+    mean_df = df.groupby(aggregate_dimension).mean().reset_index()
+    median_df = df.groupby(aggregate_dimension).median().reset_index()
+    std_df = df.groupby(aggregate_dimension).std().reset_index()
+    min_df = df.groupby(aggregate_dimension).min().reset_index()
+
+    return mean_df,median_df,std_df,min_df
+
+
+
+def plot_averaged_runs(df,aggregate_dimension,x, y,run_count,lx=False,ly=False, suppMin=False):
+    '''
+    Function to plot the mean, median, etc of the monte carlo runs along a single variable.
+    Parameters:
+    df: dataframe name
+    aggregate_dimension: the dimension you would like to aggregate on, the standard one is timestep.
+    x = x axis variable for plotting
+    y = y axis variable for plotting
+    run_count = the number of monte carlo simulations
+    lx = True/False for if the x axis should be logged
+    ly = True/False for if the x axis should be logged
+    suppMin: True/False for if the miniumum value should be plotted
+    Note: Run aggregate_runs before using this function
+    Example run:
+    '''
+    mean_df,median_df,std_df,min_df = aggregate_runs(df,aggregate_dimension)
+
+    plt.figure(figsize=(10,6))
+    if not(suppMin):
+        plt.plot(mean_df[x].values, mean_df[y].values,
+             mean_df[x].values,median_df[y].values,
+             mean_df[x].values,mean_df[y].values+std_df[y].values,
+             mean_df[x].values,min_df[y].values)
+        plt.legend(['mean', 'median', 'mean+ 1*std', 'min'],bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)
+
+    else:
+        plt.plot(mean_df[x].values, mean_df[y].values,
+             mean_df[x].values,median_df[y].values,
+             mean_df[x].values,mean_df[y].values+std_df[y].values,
+             mean_df[x].values,mean_df[y].values-std_df[y].values)
+        plt.legend(['mean', 'median', 'mean+ 1*std', 'mean - 1*std'],bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)
+
+    plt.xlabel(x)
+    plt.ylabel(y)
+    title_text = 'Performance of ' + y + ' over all of ' + str(run_count) + ' Monte Carlo runs'
+    plt.title(title_text)
+    if lx:
+        plt.xscale('log')
+
+    if ly:
+        plt.yscale('log')
+
+def plot_median_with_quantiles(df,aggregate_dimension,x, y):
+    '''
+    Function to plot the median and 1st and 3rd quartiles of the monte carlo runs along a single variable.
+    Parameters:
+    df: dataframe name
+    aggregate_dimension: the dimension you would like to aggregate on, the standard one is timestep.
+    x = x axis variable for plotting
+    y = y axis variable for plotting
+
+    Example run:
+    plot_median_with_quantiles(df,'timestep','timestep','AggregatedAgentSpend')
+    '''
+    
+    df = df[df['substep'] == df.substep.max()]
+    firstQuantile = df.groupby(aggregate_dimension).quantile(0.25).reset_index()
+    thirdQuantile = df.groupby(aggregate_dimension).quantile(0.75).reset_index()
+    median_df = df.groupby(aggregate_dimension).median().reset_index()
+    
+    fig, ax = plt.subplots(1,figsize=(10,6))
+    ax.plot(median_df[x].values, median_df[y].values, lw=2, label='Median', color='blue')
+    ax.fill_between(firstQuantile[x].values, firstQuantile[y].values, thirdQuantile[y].values, facecolor='black', alpha=0.2)
+    ax.set_title(y + ' Median')
+    ax.legend(loc='upper left')
+    ax.set_xlabel('Timestep')
+    ax.set_ylabel('Amount')
+    ax.grid()
+    
+def plot_median_with_quantiles_annotation(df,aggregate_dimension,x, y):
+    '''
+    Function to plot the median and 1st and 3rd quartiles of the monte carlo runs along a single variable.
+    Parameters:
+    df: dataframe name
+    aggregate_dimension: the dimension you would like to aggregate on, the standard one is timestep.
+    x = x axis variable for plotting
+    y = y axis variable for plotting
+
+    Example run:
+    plot_median_with_quantiles(df,'timestep','timestep','AggregatedAgentSpend')
+    '''
+    
+    df = df[df['substep'] == df.substep.max()]
+    firstQuantile = df.groupby(aggregate_dimension).quantile(0.25).reset_index()
+    thirdQuantile = df.groupby(aggregate_dimension).quantile(0.75).reset_index()
+    median_df = df.groupby(aggregate_dimension).median().reset_index()
+    
+    fig, ax = plt.subplots(1,figsize=(10,6))
+    ax.axvline(x=30,linewidth=2, color='r')
+    ax.annotate('Agents can withdraw and Red Cross Drip occurs', xy=(30,2), xytext=(35, 1),
+            arrowprops=dict(facecolor='black', shrink=0.05))
+    
+    ax.axvline(x=60,linewidth=2, color='r')
+    ax.axvline(x=90,linewidth=2, color='r')
+    ax.plot(median_df[x].values, median_df[y].values, lw=2, label='Median', color='blue')
+    ax.fill_between(firstQuantile[x].values, firstQuantile[y].values, thirdQuantile[y].values, facecolor='black', alpha=0.2)
+    ax.set_title(y + ' Median')
+    ax.legend(loc='upper left')
+    ax.set_xlabel('Timestep')
+    ax.set_ylabel('Amount')
+    ax.grid()
+
+
+def first_five_plot(df,aggregate_dimension,x,y,run_count):
+    '''
+    A function that generates timeseries plot of at most the first five Monte Carlo runs.
+    Parameters:
+    df: dataframe name
+    aggregate_dimension: the dimension you would like to aggregate on, the standard one is timestep.
+    x = x axis variable for plotting
+    y = y axis variable for plotting
+    run_count = the number of monte carlo simulations
+    Note: Run aggregate_runs before using this function
+    Example run:
+    first_five_plot(df,'timestep','timestep','revenue',run_count=100)
+    '''
+    mean_df,median_df,std_df,min_df = aggregate_runs(df,aggregate_dimension)
+    plt.figure(figsize=(10,6))
+    if run_count < 5:
+        runs = run_count
+    else:
+        runs = 5
+    for r in range(1,runs+1):
+        legend_name = 'Run ' + str(r)
+        plt.plot(df[df.run==r].timestep, df[df.run==r][y], label = legend_name )
+    plt.plot(mean_df[x], mean_df[y], label = 'Mean', color = 'black')
+    plt.legend(bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)
+    plt.xlabel(x)
+    plt.ylabel(y)
+    title_text = 'Performance of ' + y + ' over the First ' + str(runs) + ' Monte Carlo Runs'
+    plt.title(title_text)
+    #plt.savefig(y +'_FirstFiveRuns.jpeg')
+    
+    
+def aggregate_runs_param_mc(df,aggregate_dimension):
+    '''
+    Function to aggregate the monte carlo runs along a single dimension.
+    Parameters:
+    df: dataframe name
+    aggregate_dimension: the dimension you would like to aggregate on, the standard one is timestep.
+    Example run:
+    mean_df,median_df,std_df,min_df = aggregate_runs(df,'timestep')
+    '''
+    df = df[df['substep'] == df.substep.max()]
+    mean_df = df.groupby(aggregate_dimension).mean().reset_index()
+    median_df = df.groupby(aggregate_dimension).median().reset_index()
+    #min_df = df.groupby(aggregate_dimension).min().reset_index()
+    #max_df = df.groupby(aggregate_dimension).max().reset_index()
+    return mean_df,median_df
+
+def param_dfs(results,params,swept):
+    mean_df,median_df = aggregate_runs_param_mc(results[0]['result'],'timestep')
+    mean_df[swept] = params[0]
+    median_df[swept] = params[0]
+    #max_df[swept] = params[0]
+    #min_df[swept] = params[0]
+    for i in range(1,len(params)):
+        mean_df_intermediate,median_df_intermediate = aggregate_runs_param_mc(results[i]['result'],'timestep')
+        mean_df_intermediate[swept] = params[i]
+        median_df_intermediate[swept] = params[i]
+        #max_df_intermediate[swept] = params[i]
+        #min_df_intermediate[swept] = params[i]
+        mean_df= pd.concat([mean_df, mean_df_intermediate])
+        median_df= pd.concat([median_df, median_df_intermediate])
+        #max_df= pd.concat([max_df, max_df_intermediate])
+        #min_df= pd.concat([min_df, min_df_intermediate])
+    return mean_df,median_df
+
+
+def param_plot(results,state_var_x, state_var_y, parameter, save_plot = False,**kwargs):
+    '''
+    Results (df) is the dataframe (concatenated list of results dictionaries)
+    length = intreger, number of parameter values
+    Enter state variable name as a string for x and y. Enter the swept parameter name as a string.
+    y_label kwarg for custom y-label and title reference
+    x_label kwarg for custom x-axis label
+    '''
+    sns.scatterplot(x=state_var_x, y = state_var_y, hue = parameter, style= parameter, palette  = 'coolwarm',alpha=1, data = results, legend="full")
+    title_text = 'Effect of ' + parameter + ' Parameter Sweep on ' + state_var_y
+    for key, value in kwargs.items():
+        if key == 'y_label':
+            plt.ylabel(value)
+            title_text = 'Effect of ' + parameter + ' Parameter Sweep on ' + value
+        if key == 'x_label':
+            plt.xlabel(value)
+    plt.title(title_text)
+    if save_plot == True:
+        filename = state_var_y + state_var_x + parameter + 'plot.png'
+# #         plt.savefig('static/images/' + filename)
+#         plt.savefig(filename)
+        lgd = plt.legend(bbox_to_anchor=(1.05, 1), loc=2, borderaxespad=0.)
+        #title_text = 'Market Volatility versus Normalized Liquid Token Supply for All Runs'
+        plt.title(title_text)
+        plt.savefig('static/images/' + filename, bbox_extra_artists=(lgd,), bbox_inches='tight')

Simulation_param/model/parts/system.py

@@ -0,0 +1,279 @@
+
+import numpy as np
+import pandas as pd
+from cadCAD.configuration.utils import access_block
+from .initialization import *
+from .supportingFunctions import *
+from collections import OrderedDict
+
+# Parameters
+agentsMinus = 2
+# percentage of balance a user can redeem
+redeemPercentage = 0.5
+
+# Behaviors
+def choose_agents(params, step, sL, s):
+    '''
+    Choose agents to interact during the given timestep and create their demand from a uniform distribution. 
+    Based on probability, choose utility. 
+    '''
+    outboundAgents = np.random.choice(mixingAgents,size=len(mixingAgents)-agentsMinus).tolist()
+    inboundAgents = np.random.choice(mixingAgents,size=len(mixingAgents)-agentsMinus).tolist()
+    stepDemands = np.random.uniform(low=1, high=500, size=len(mixingAgents)-agentsMinus).astype(int)
+    
+
+    stepUtilities = np.random.choice(list(UtilityTypesOrdered.keys()),size=len(mixingAgents)-agentsMinus,p=list(utilityTypesProbability.values())).tolist()
+
+    return {'outboundAgents':outboundAgents,'inboundAgents':inboundAgents,'stepDemands':stepDemands,'stepUtilities':stepUtilities}
+
+
+def spend_allocation(params, step, sL, s):
+    '''
+    Take mixing agents, demand, and utilities and allocate agent shillings and tokens based on utility and scarcity. 
+    '''
+    # instantiate network state
+    network = s['network']
+
+    spendI = []
+    spendJ = []
+    spendAmount = []
+
+    # calculate max about of spend available to each agent
+    maxSpendShilling = {}
+    for i in mixingAgents:
+        maxSpendShilling[i] = network.nodes[i]['native_currency']
+        
+    maxSpendCIC = {}
+    for i in mixingAgents:
+        maxSpendCIC[i] = network.nodes[i]['tokens']
+
+
+    for i in mixingAgents: 
+        rankOrder = {}
+        rankOrderDemand = {}
+        for j in network.adj[i]:
+            try:
+                rankOrder[j] = UtilityTypesOrdered[network.adj[i][j]['utility']]
+                rankOrderDemand[j] = network.adj[i][j]['demand']
+                rankOrder = dict(OrderedDict(sorted(rankOrder.items(), key=lambda v: v, reverse=False)))
+                for k in rankOrder:
+                    # if i or j is external, we transact 100% in shilling
+                    if i == 'external':
+                        amt = spendCalculationExternal(i,j,rankOrderDemand,maxSpendShilling)
+                        spendI.append(i)
+                        spendJ.append(j)
+                        spendAmount.append(amt)
+                        maxSpendShilling[i] = maxSpendShilling[i] - amt 
+                    elif j == 'external':
+                        amt = spendCalculationExternal(i,j,rankOrderDemand,maxSpendShilling)
+                        spendI.append(i)
+                        spendJ.append(j)
+                        spendAmount.append(amt)
+                        maxSpendShilling[i] = maxSpendShilling[i] - amt 
+                    else:
+                        amt = spendCalculation(i,j,rankOrderDemand,maxSpendShilling,maxSpendCIC,fractionOfDemandInCIC)
+                        spendI.append(i)
+                        spendJ.append(j)
+                        spendAmount.append(amt)
+                        maxSpendShilling[i] = maxSpendShilling[i] - amt * (1- fractionOfDemandInCIC)
+                        maxSpendCIC[i] = maxSpendCIC[i] - (amt * fractionOfDemandInCIC)
+            except:
+                pass
+    return {'spendI':spendI,'spendJ':spendJ,'spendAmount':spendAmount}
+
+
+def withdraw_calculation(params, step, sL, s):
+    ''''''
+    # instantiate network state
+    network = s['network']
+
+    # Assumptions:
+    # * user is only able to withdraw up to 50% of balance, assuming they have spent 50% of balance
+    # * Agents will withdraw as much as they can.
+    withdraw = {}
+
+    fiftyThreshold = {}
+
+    startingBalance = s['startingBalance']
+
+    spend = s['30_day_spend']
+    timestep = s['timestep']
+
+    division =  timestep % 30  == 0
+
+    if division == True:
+        for i,j in startingBalance.items():
+            fiftyThreshold[i] = j * 0.5
+        if s['timestep'] > 7:
+            for i,j in fiftyThreshold.items():
+                if spend[i] > 0 and fiftyThreshold[i] > 0:
+                    if spend[i] * fractionOfActualSpendInCIC >= fiftyThreshold[i]:
+                        spent = spend[i]
+                        amount = spent * redeemPercentage
+                        if network.nodes[i]['tokens'] > amount:
+                            withdraw[i] = amount
+                        elif  network.nodes[i]['tokens'] < amount:
+                            withdraw[i] = network.nodes[i]['tokens']
+                    else:
+                        pass
+                else:
+                    pass
+        else:
+            pass
+    else:
+        pass
+
+
+    return {'withdraw':withdraw}
+
+# Mechanisms 
+def update_agent_activity(params,step,sL,s,_input):
+    '''
+    Update the network for interacting agent, their demand, and utility.
+    '''
+    y = 'network'
+    network = s['network']
+
+    outboundAgents = _input['outboundAgents']
+    inboundAgents = _input['inboundAgents']
+    stepDemands = _input['stepDemands']
+    stepUtilities = _input['stepUtilities']
+    
+    # create demand edge weights
+    try:
+        for i,j,l in zip(outboundAgents,inboundAgents,stepDemands):
+            network[i][j]['demand'] = l
+    except:
+        pass
+
+    # Create cic % edge weights
+    try:
+        for i,j in zip(outboundAgents,inboundAgents):
+            # if one of the agents is external, we will transact in 100% shilling
+            if i == 'external':
+                network[i][j]['fractionOfDemandInCIC'] = 1
+            elif j == 'external':
+                network[i][j]['fractionOfDemandInCIC'] = 1
+            else:
+                network[i][j]['fractionOfDemandInCIC'] = fractionOfDemandInCIC
+    except:
+        pass
+
+    # Create utility edge types
+    try: 
+        for i,j,l in zip(outboundAgents,inboundAgents,stepUtilities):
+            network[i][j]['utility'] = l
+    except:
+        pass
+
+    x = network
+    return (y,x)
+
+
+def update_outboundAgents(params,step,sL,s,_input):
+    '''
+    Update outBoundAgents state variable
+    '''
+    y = 'outboundAgents'
+
+    x = _input['outboundAgents']
+
+    return (y,x)
+
+def update_inboundAgents(params,step,sL,s,_input):
+    '''
+    Update inBoundAgents state variable
+    '''
+    y = 'inboundAgents'
+
+    x = _input['inboundAgents']
+    return (y,x)
+
+
+def update_node_spend(params, step, sL, s,_input):
+    '''
+    Update network with actual spend of agents.
+    '''
+    y = 'network'
+    network = s['network']
+    
+    spendI = _input['spendI']
+    spendJ = _input['spendJ']
+    spendAmount = _input['spendAmount']
+
+    for i,j,l in zip(spendI,spendJ,spendAmount):   
+        network[i][j]['spend'] = l
+        if i == 'external':
+            network[i][j]['fractionOfActualSpendInCIC'] = 1
+        elif j == 'external':
+            network[i][j]['fractionOfActualSpendInCIC'] = 1
+        else:
+            network[i][j]['fractionOfActualSpendInCIC'] = fractionOfActualSpendInCIC
+
+    outflowSpend, inflowSpend = iterateEdges(network,'spend')
+
+    for i, j in inflowSpend.items():
+        if i == 'external':
+            network.nodes[i]['native_currency'] = network.nodes[i]['native_currency'] +  inflowSpend[i]
+        elif j == 'external':
+            network.nodes[i]['native_currency'] = network.nodes[i]['native_currency'] +  inflowSpend[i]
+        else:
+            network.nodes[i]['native_currency'] = network.nodes[i]['native_currency'] +  inflowSpend[i] * (1- fractionOfDemandInCIC)
+            network.nodes[i]['tokens'] = network.nodes[i]['tokens'] + (inflowSpend[i] * fractionOfDemandInCIC)
+        
+    for i, j in outflowSpend.items():
+        if i == 'external':
+            network.nodes[i]['native_currency'] = network.nodes[i]['native_currency'] - outflowSpend[i]
+        elif j == 'external':
+            network.nodes[i]['native_currency'] = network.nodes[i]['native_currency'] - outflowSpend[i]
+        else:
+            network.nodes[i]['native_currency'] = network.nodes[i]['native_currency'] - outflowSpend[i]* (1- fractionOfDemandInCIC)
+            network.nodes[i]['tokens'] = network.nodes[i]['tokens'] - (outflowSpend[i] * fractionOfDemandInCIC)
+
+    # Store the net of the inflow and outflow per step
+    network.nodes['external']['delta_native_currency'] = sum(inflowSpend.values()) - sum(outflowSpend.values())
+
+    x = network
+    return (y,x)
+
+
+def update_withdraw(params, step, sL, s,_input):
+    '''
+    Update flow sstate variable with the aggregated amount of shillings withdrawn
+    '''
+    y = 'withdraw'
+    x = s['withdraw']
+    if _input['withdraw']:
+        x = _input['withdraw']
+    else:
+        x = 0
+
+    return (y,x)
+
+def update_network_withraw(params, step, sL, s,_input):
+    '''
+    Update network for agents withdrawing 
+    '''
+    y = 'network'
+    network = s['network']
+    withdraw = _input['withdraw']
+
+    if withdraw:
+        for i,j in withdraw.items():
+            # update agent nodes
+            network.nodes[i]['tokens'] = network.nodes[i]['tokens'] - j
+            network.nodes[i]['native_currency'] = network.nodes[i]['native_currency'] + (j * leverage)
+
+        withdrawnCICSum = []
+        for i,j in withdraw.items():
+            withdrawnCICSum.append(j)
+        
+        # update cic node
+        network.nodes['cic']['native_currency'] = network.nodes[i]['native_currency'] - (sum(withdrawnCICSum) * leverage)
+        network.nodes['cic']['tokens'] = network.nodes[i]['tokens'] + (sum(withdrawnCICSum) * leverage)
+
+    else:
+        pass
+    x = network
+    return (y,x)
+