6 problems
- 0 votes0 replies0 views
The conjecture that the normalized infimum C(u^m) equals one
Conjecture that . For every such ,
- 0 votes0 replies0 views
Logarithmic sample-complexity conjecture for efficient multichannel sparse blind deconvolution
Consider multichannel sparse blind deconvolution with signal vectors having unknown nonzero supports, and let denote the number of measurements or channe…
- 0 votes0 replies1 view
Negligible trapping of manifold gradient descent near saddle points
Consider manifold gradient descent (MGD) applied to the paper's objective function on the sphere, and let its iterates start from the initialization used in the analysis. The objec…
- 0 votes0 replies0 views
General subgaussian bounds for multichannel sparse blind deconvolution
The channels are modeled by sparse random vectors, with the Bernoulli-Rademacher model used in the paper as a special case of Bernoulli-subgaussian distributions. General subgaussi…
- 0 votes0 replies0 views
Extension of the analysis to other subgaussian nonzero entries
The channels are assumed to have a Bernoulli-subgaussian model: their entries are zero with Bernoulli sparsity and have subgaussian nonzero values; the paper's analysis currently t…
- 0 votes0 replies0 views
The 4N-4 conjecture for generalized phase retrieval
The 4N-4 conjecture. The smallest explicit measurement ensemble for generalized phase retrieval should have size .