ankitxraj21 commited on
Commit
ebde5f3
·
verified ·
1 Parent(s): 9a0bf91

Update environment.py

Browse files
Files changed (1) hide show
  1. environment.py +13 -20
environment.py CHANGED
@@ -1,6 +1,5 @@
1
  import asyncio
2
  import math
3
- import random
4
  from models import State
5
 
6
  class CloudCostEnv:
@@ -13,27 +12,22 @@ class CloudCostEnv:
13
  self.servers = 5
14
  self.is_spot = False
15
 
16
- def reset(self):
17
  self.reset_vars()
18
  return self.get_current_state()
19
 
20
  def get_current_state(self):
21
- # This simulates traffic based on the "Task" (Easy/Medium/Hard)
22
  if self.task == "easy":
23
  load = 450
24
  else:
25
- # Sine wave simulates day/night traffic peaks
26
  load = int(500 + 350 * math.sin(self.steps / 5))
27
-
28
- # Physics: More servers = lower latency. Load > Servers = high latency.
29
  capacity = self.servers * 100
30
  latency = max(20, (load / (capacity + 1)) * 150)
31
-
32
- # Cost: Spot is 75% cheaper than On-Demand
33
  base_rate = 0.60
34
  current_rate = base_rate * 0.25 if self.is_spot else base_rate
35
  cost = self.servers * current_rate
36
-
37
  return State(
38
  cpu_usage=min(100.0, (load / (capacity + 1)) * 100),
39
  active_servers=self.servers,
@@ -43,23 +37,22 @@ class CloudCostEnv:
43
  )
44
 
45
  async def step(self, action_int):
46
- # Apply the action from the AI
47
- if action_int == 1:
48
  self.servers += 1
49
- elif action_int == 2:
50
  self.servers = max(1, self.servers - 1)
51
- elif action_int == 3:
52
  self.is_spot = not self.is_spot
53
-
54
  self.steps += 1
55
  state = self.get_current_state()
56
-
57
- # Calculate Reward (The Score)
58
  reward = 1.0
59
- if state.response_latency > 150: reward -= 0.7 # Penalty for being slow
60
- if state.hourly_burn_rate > 5.0: reward -= 0.5 # Penalty for overspending
61
-
62
- # Episode ends after 20 steps for the grader
 
63
  done = self.steps >= 20
64
  return state, reward, done
65
 
 
1
  import asyncio
2
  import math
 
3
  from models import State
4
 
5
  class CloudCostEnv:
 
12
  self.servers = 5
13
  self.is_spot = False
14
 
15
+ async def reset(self): # ← added async HERE
16
  self.reset_vars()
17
  return self.get_current_state()
18
 
19
  def get_current_state(self):
 
20
  if self.task == "easy":
21
  load = 450
22
  else:
 
23
  load = int(500 + 350 * math.sin(self.steps / 5))
24
+
 
25
  capacity = self.servers * 100
26
  latency = max(20, (load / (capacity + 1)) * 150)
 
 
27
  base_rate = 0.60
28
  current_rate = base_rate * 0.25 if self.is_spot else base_rate
29
  cost = self.servers * current_rate
30
+
31
  return State(
32
  cpu_usage=min(100.0, (load / (capacity + 1)) * 100),
33
  active_servers=self.servers,
 
37
  )
38
 
39
  async def step(self, action_int):
40
+ if action_int == 1:
 
41
  self.servers += 1
42
+ elif action_int == 2:
43
  self.servers = max(1, self.servers - 1)
44
+ elif action_int == 3:
45
  self.is_spot = not self.is_spot
46
+
47
  self.steps += 1
48
  state = self.get_current_state()
49
+
 
50
  reward = 1.0
51
+ if state.response_latency > 150:
52
+ reward -= 0.7
53
+ if state.hourly_burn_rate > 5.0:
54
+ reward -= 0.5
55
+
56
  done = self.steps >= 20
57
  return state, reward, done
58