TOPICS
Search

Tree Reconstruction


Tree reconstruction asks whether observations far from the root retain information about a random root state transmitted through noisy independent edge channels. In the four-state ferromagnetic channel, the root is uniform on four states and a child has conditional transition probability

 P_lambda(j|i)=lambdadelta_(ij)+(1-lambda)/4,

where 0<lambda<1 and delta_(ij) is the Kronecker delta. Different children are independent conditional on their parent's state. Reconstruction occurs if the expected total variation distance of the root posterior, conditional on depth-n spins, from the uniform distribution has a positive limit as n->infty.


See also

Cayley Tree, Conditional Probability, Potts Model, Rooted Tree, Total Variation Distance

Explore with Wolfram|Alpha

References

Sly, A. "Reconstruction for the Potts Model." Ann. Probab. 39, 1365-1406, 2011. https://doi.org/10.1214/10-AOP584.

Cite this as:

Weisstein, Eric W. "Tree Reconstruction." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/TreeReconstruction.html

Subject classifications