Eric Reiss

General: This is the home page of a channel. It is similar to someone's profile "wall" in a social network context. Posts created by the channel are displayed according to the observer's viewing permissions.
Create a Post: If you have permission to create posts on the channel page, then you will see the post editor at the top.

Help

Eric Reiss

ericr@left.engr.usu.edu

Undergraduate Research Assistant

Eric Reiss

Utah, Wed, 05 Jul 2023 18:39:25 -0600

I have added an option that generates the label, state, and transition files that can be used to build an explicit model. I am fairly confident that the transient and steady state analysis outputs give the the probability that the model is in a given at the end of the analysis where the given state is the one in parentheses. Currently the explicit states do not stop once the finish condition is reached because I thought it may be interesting to analyze the model when it starts in the all zero state and see the probability that it generates an error. I don't know if that will be useful or not, but I plan on adding an option that recognizes the finish condition tomorrow. The next thing that may be interesting that I noticed in the documentation is the -simpath tag. I am going to dig into that as well to see if it can more closely match NGDBF iterations since PRISM's iterations are different. I have also added more trapping sets to load_trapping_sets.m.

#^https://github.com/fluentverification/usu_stochastic_case_studies/tree/master/ngdbf_models

NGDBF

Eric Reiss

Thu, 06 Jul 2023 14:47:03 -0600

I have added a final condition option and made the functions compatible with Octave.

Eric Reiss

Utah, Sat, 24 Jun 2023 13:53:31 -0600

I am still trying to decipher the transient analysis output. I added 50 to all the states to see what that would do to the labeling of the output and for some reason, it made the states 100,101,... I wonder if PRISM adds the smallest state to the output states because state was originally [1..8] and that caused the output states to start at 2. I changed state to be in the range [0..7], and that caused the output states to start at 0. I will keep looking into it.

I also changed the script so that it saves every model. I have attached an example of the output file and one of the models. I plan to look more into importing the state space next week. I have also added a script called "load_trapping_sets.m" that I am slowly adding all the trapping set adjacency matrices to.

NGDBF

NGDBF Script Update

Eric Reiss

Utah, Tue, 13 Jun 2023 16:54:25 -0600

In the latest commit, I generalized run_ngdbf so that it can handle other trapping sets. I turned it into a function that takes a transition matrix as an argument. I couldn't remember the way that the trapping set is represented as a matrix, so I just used my intuition and went with what made sense to me. A '1' is placed to represent a check node between two variable nodes, with '1's on the diagonal representing degree one check nodes. For example, I represent the (4,4) trapping set with the matrix
1 1 0 1
1 1 1 0
0 1 1 1
1 0 1 1
If there is a more correct way to represent them, it shouldn't be too hard to change that. I want to add the other channel parameters as arguments next and then wrap that in another script that will sweep the SNR and maybe some other parameters.

Github: #^https://github.com/fluentverification/usu_stochastic_case_studies/blob/master/ngdbf_models/run_ngdbf.m

NGDBF

Winstead

Wed, 14 Jun 2023 09:57:08 -0600

The adjacency matrix should be formatted with one row for each parity-check, and one column for each symbol node. A ‘1’ is placed whenever there is an edge between the corresponding check and symbol nodes. A degree-one check node should appear as a row with only a single non-zero position. A degree-two check node should appear as a row with exactly two non-zero positions.

Example the “triangular” (3,3) trapping set:


C1 - - - S1 - - - - C2 - - - - S2 - - - C3
         |                     |
         |                     |
         C4 - - - - S3 - - - - C5
                    |
                    C6

This graph should have a 3x6 adjacency matrix:



      S1  S2  S3
C1    1   0   0
C2    1   1   0
C3    0   1   0
C4    1   0   1
C5    0   1   1
C6    0   0   1