Discrete memoryless channel capacity
Capacity is the maximal asymptotically reliable rate.
Key knowledge
For a finite discrete memoryless channel p(y|x), C=max_pₓ I(X;Y) bits per channel use. Rates below C allow vanishing average block error; rates above C do not.
Conditions and scope
- Channel uses follow the same transition law independently conditional on inputs.
- No feedback or extra input-cost constraint is included; finite-block performance is a separate question.
Sources · 2
References
Sources retain their recorded scope. A source link alone does not establish verification of every statement.
Tsachy Weissman / Stanford University2 sources
Evidence and review
Definitions 4–6 and Theorem 7: achievable rate, memoryless channel and maximal mutual information. Opening channel-coding theorem: rates below C achievable, rates above C not achievable with vanishing average block error.
2026-10-10 · Finite DMC capacity and coding threshold.
Entry details
Permanent link to this entry ↗- Record identity
- info-channel-capacity
- Revision
- university-reference.1
- Structured reference record
- University reference