September 17th

Organisms group numbers

Group 1:
    Strategy: Old strategy uses age and generation to dictate actions. The organisms moved in an diagonal path. Continued to work on the decision tree and the age classification.
   
    Jinesh: Not in class
    Vishwa: Also uses age of organism to determines actions. The longer you have been around for the more conservative your actions. The intuition behind this is that the latter in the game the less food is around. Old strategy would win with larger board.
    Hao: The decision tree did not work because there was no way to create a static tree and could not pass the information on to the children. 

Group 2:

    Di: 3 different roles, the New Born that looks for food, but stays on the same path.
    Lauren: Split organisms into different roles denoted by the different states. The Scavenger moves toward food. The Camper stays put and will only move if it needs food and there is something around it. It tried to leave room between other campers. Roles can be switched based on energy or food available
    Sam:


Group 3:

    Markus:  Age is used to determine weather to move or reproduce. Only moves when org is young and then it will stop, to make the organism localized, and create a blocking structure. Organisms are shy and try and move away from each other. Organism spreads out well and creates a checker board.
    Yigit:
    Patric: Keeps track of how long it has been until the org has eaten as a proxy of how much food is on the board and how many other organisms. This beats sheep since it is overly aggressive. Does not look at states, but tries to spoof states. Tried to implement machine learning on parameters, but took too long. r

Group 4:
    Tanay: Organism is very good if P=1, Q=0.
    Franklin: Initial approach was scrapped. Noticed they were being starved sheep because the sheep was blocking them. So, every 20 rounds the organism will change direction. Moving in one direction is good since it allowed them to spread out on the board. Set very high thresholds for reproduction.
    Tiernan: Tries to find the density of the board (other organisms) and using that to modify actions. Not caring about different species. Thresholds will be thrown off if amount of energy per food is lowered.



Group 5:
    Strategy:

    Tim: When the organism in alone on the board the organism should think in terms of groups, or farms. This works when the energy is abundant or the organisms will die out. When the food is scarce the org needs to be very greedy. It is hard to switch between the single organism strategy and the multi organisms. In the organism is by itself, the longer you are on the board the better the energy and pattern is. Minimize the number of organisms with multiple free spaces around it.
    Yang: old strategy had a low reproduction limit, increased this. Still using the generational strategy to determine actions, but the generations can change depending on the available food.
    Harjot

    Demo: Does very well in the regular configurations

Group 6:
    Strategy: Farming with a notion of a colony with solders. Works well without other players or with low food. If too many people around, farm did not get a change to start.
   
    Tanveer: Samples current board status
    Hari: Decided not to go with the farming. Recapped Strategy. Uses states for communication. Uses age to help determine whether to switch to farmer. Organism moves to aggressive player and them will convert back to farms

    Demo: Player is not switching to aggressive player. Did very well with very low food

September 10th

Class Discussion Recap

Group 1: 

Improvements: made two more strategies 

Conditions: works will with increase with board size and cost of moving. Fails if food is scarce 

Jinesh: Zone approach: not out preforming old strategy. Splitting up board into zones for each organism. Communicate with with states to tell if zone is occupied, using bits to communicate location of zones. Has smaller pop them other strategies. Any strategy that favors reproduction decreases overall energy 

Vishwa: Recapped strategy. 

Hao: Decision tree approach. Create a decision tree based on information the organism has immediately. Will determine if the organism moves or reproduces. Each organism learns on their own. 

Group 2: 

Improvements:  Have a ratio, food vs. number of turns looking, the “healthy-ness” of the board. This affects if you are waiting for food around or looking for food. 

Conditions: have trouble with low costs of moving because being conservative about moving. 

Di: made improvements, moves based on hungry-ness and how healthy the board is. 

Lauren: Tried to stabilize the energy. Reproduce based on neighbors. Each organisms starts with a certain number of reproductions to try ad get more population against other organisms.  Look to implement state species communication. 

Sam: Estimating the state of the board (how much food is around). Recapped strategy. Did demo

Demo: Very Large board with sparse conditions: wait for a long time does not reproduce until about 1.5K

Group 3: 

Improvements: Generational states, different actions based on 

Demo: moves around, eats only when hungry. reproduce very slowly. 

Markus: Recapped strategy. Think of current strategy as switching between 2 behaviors, exploring and reproducing. 

Yigit: Recapped strategy

Patric:  Discussed improvements, in the beginning move around randomly until found food of a certain amount (as proxy for time) then reproduce faster. Equilibrate at 110k. Leave food alone unless the food is over a certain amount. Food situations is mostly an issue at the start. Reproducing is very costly.

Group 4: 

Improvements: Dynamic thresholds for actions based on the number of organisms around you. Communicating with states 

Tanay: This is not always the best strategy, you want to reproduce more as the board gets full. Moving in one way limits what the organisms can do. Try to estimate p & q. Should hibernate, minimize reproduction, and scouting. 

Franklin: Recapped initial strategy.  

Tiernan: Another problem is there is a population cap when current strategy. Sitting around is a non starter if there are other organisms on the board

Group 5: 

Improvements:  Tries to calculate the cost to do actions, and make decisions about that. organisms are not learning as well as hoped. Estimating p & q. ONly implemented communication if farmer is blocked in.

Demo: stays still for a while, reproduce to the north and move to the west 

Conditions: does not do as well with plentiful boards 

Tim: all communication is based on state. Working on data robustness. p & q are very important. 

Yang: The most important thing is to survive, whether to move or to stay still can be calculated with the probably. At the beginning you want to stay still and then once you move you want to and keep moving that direction. Should you eat food immediately or wait? You should wait, the longer you wait the more it will double and you will have more food to eat. 

Harjot: Recapped strategy  and presented improvements. Board size is not worth knowing. Have heuristic to determine when to reproduce. 

Group 6: 

Improvements: made a very defined farming structure with communication. Create multiple “types” of organisms: farmers, scouts, and soldiers. Has problems with “deadlock” where two farms will lock each other out. 

Conditions: sometimes goes extinct with less food. Does well with bigger board

Current Strategy: Structured colonies with different types of “organisms” 

Tanveer: Improvements and improvement demo. Energy remains constant at 30k

Hari: Summary: stayed still and reproduce. When organism moved around a lot it survived better

Demo: Showed diamond farm, and colonies. organisms make a checkered board like pattern. 

Class Notes - September 5th 2013

Group 1: 

Strategy: Heuristic: look for food and go to it. Do no move until there is food around. If 2 or more squared of food, reproduce.

Jinesh: Did demo

Vishwa: discussed strategy 

Hao: similar to chess game, use a game tree

Demo: Staying still until food. Does not go find food, reproduce according to food around. 

Group 2:

Initial observation find best heuristic, always moving west, always reproduce south. Over population is a problem. Tried to reproduce in terms of neighbors

Di: only reproduce if reached a cell with a certain amount of food

Lauren: Discussed strategy 

Sam: each organism has a “domain” for food. Having problems if no food appears around, will go extinct. Did demo

Demo: organism moves randomly, until finds food. eats all squares and moves to random square. If food on next square is more then 30% reproduce. Does not reproduce until very late in game, then takes over board

Group 3: 

Strategy: threshold for reproduce; stay still is better, don’t move unless food; Eat until there is only 2 food

Markus: discus strategy, did Demo

Yigit: next step, incorporate communication with neighbors

Patric: new to group

Demo: Stays still, then moves randomly or towards food. 

Group 4: 

Initial observations: always move west, always reproduce North, reproduce at energy level. This allows the food to reproduce in the different area of the board. This is current strategy 

Tanay: 

Franklin: discus Strategy

Tiernan: did Demo

Demo: Reproduce first then wait. Move to the west. a lot of food accumulates, then population boom. Might have a problem if food has lower probability    

Group 5: 

Strategy: limit for reproduce in one direction otherwise stay still until you have energy for 6 moves (based on trial and error) Using M and v. Reproduce so that there is an organism on every row of the board. Always move west and reproduce down

Tim: population based terms of food, “farming”, best to move in one direction. Did Demo

Yang

Harjot

Demo: reproduce first, then hang out(both below threshold) and then only move when need energy. Takes a long time to take over board

Group 6: 

Strategy: Stay still until food is around, if food is around, reproduce

Tanveer: Farming, using four organisms around the food

Hari: discussed strategy. Did demo

Nirmal: new to group

Demo: Stays still, will just sit there and die if no food around it. Reproduced onto food. Biggest drawback is that organism does not move until it dies so there is no food appearing on that square. Stable at 200 is organisms