next-steps.txt

**WHAT WE HAVE DONE**
-> Implementation of the DCT algorithm
  - Quantizers are successfully being trained for encoding and decoding the DCT transform
  - Using the the optimal bit allocation matrices for B = 76 bits
  - Keaton also simulated the BSC and the images look to be re-constructing properly
  - Rob also added the polya channel simulation to this algorithm

-> General Lloyd's algorithm that accounts for BSC or Polya channel error
  - Rob added the polya channel conditional probabilities
  - The test to see if the algorithm and if we are simulating the polya channel properly is in
    the polya-channel.py file. We train the quantizer using the polya channel and then send the
    codebook through the polya channel and compare the distortions. They seem to match up.
  - We may need more testing, but things are looking good there
  - Rob would still like other eyes on the polya channel implementation to see if he missed anything

**WHAT WE NEED TO MODIFY IN OUR IMPLEMENTATION**
-> We only use 1000 samples to train the quantizers in the DCT algorithm. This is too little
  - If we increase the number of samples, the algorithm will take way too long.
  - We need to port the general_lloyd_algorithm(...) function to C that Liam/Matthew have been working on.
  - We may need to leverage the CPUs of the supercomputer and see if that speeds anything up.

-> We need relevant images
  - We need grayscale images that better reflect our application.
  - Can somebody look into potential image candidates that we can use?
  
-> We need to add timing tests for thesis 
  - We should probably modify the algorithms to include timings for each of the iterations
  - There are probably things Linder will ask for that we need to be prepared to add in terms of testing
    and showing things like graphs or other timings for the thesis

**LOOKING AHEAD TO THESIS**
-> We need to start drafting out a plan for the report
  - We should set up a meeting this week and start designing the report / designating areas for people to work on