# A Family of Algorithms for Computing Information-Theoretic Forms of Strong Converse Exponents in Channel Coding and Lossy Source Coding

This paper studies the computation of error and correct decoding probability exponents in channel coding and lossy source coding. For channel coding, we show that the recently proposed algorithm of Tridenski and Zamir for computing strong converse exponent has the attractive property that the objective function alternately takes the forms appearing in Arimoto's and Dueck and Körner's exponents. Because of this property, the convergence of the algorithm directly implies the match of the two exponents. Then, we consider a special case of Tridenski et al.'s generalized algorithm of that has not been examined in depth. We show that the objective function of this algorithm also has the interesting property that it takes the forms of Dueck and Körner's exponent and another representation of the strong converse exponent derived by Arimoto's algorithm. This particular case is important because the objective function in this case can be used to prove that the Gallager and Csiszár and Körner error exponents agree. For lossy source coding, we propose two new algorithms for computing the Csiszár and Körner's strong converse exponent. We also define a function similar to Blahut's error exponent with a negative slope parameter for source coding. We show that one of the proposed algorithms has the property that the objective function alternately takes the forms of Csiszár and Körner's exponent and the newly defined exponent function. The convergence of the algorithm proves that the new exponent function coincides with the Csiszár and Körner exponent. We also prove that Arimoto algorithm for computing error exponent of lossy source coding can work for negative ρ∈ [-1,0). and thus can be used to compute the new exponent function. Computation of Arikan and Merhav's guessing exponent is also discussed.

• 3 publications
• 14 publications
research
01/16/2020

### Proof of Convergence for Correct-Decoding Exponent Computation

For a discrete memoryless channel with finite input and output alphabets...
research
09/21/2022

### Tight Exponential Strong Converse for Source Coding Problem with Encoded Side Information

The source coding problem with encoded side information is considered. A...
research
04/23/2023

### Computing the optimal error exponential function for fixed-length lossy coding in discrete memoryless sources

The error exponent of fixed-length lossy source coding was established b...
research
01/17/2019

### Joint Source-Channel Coding for the Multiple-Access Channel with Correlated Sources

This paper studies the random-coding exponent of joint source-channel co...
research
08/21/2002

### Winner-Relaxing Self-Organizing Maps

A new family of self-organizing maps, the Winner-Relaxing Kohonen Algori...
research
10/23/2019

### On the Bee-Identification Error Exponent with Absentee Bees

The "bee-identification problem" was formally defined by Tandon, Tan and...
research
11/28/2017

### Expurgated Bounds for the Asymmetric Broadcast Channel

This work contains two main contributions concerning the expurgation of ...