Sergio Sánchez Ramírez

icon indicating a website link https://mofeing.xyz

icon company Barcelona Supercomputing Center - @bsc-quantic icon location Barcelona, Spain icon location PhD student @bsc-quantic / Quantum Computing, Tensor Networks, HPC, ...

Results 119 comments of Sergio Sánchez Ramírez

Fix type escaping in `@import_frule`, `@import_rrule`

Okay, fixed now.

CalDAV sync

+1 for this

Vectorize batch tensor network contractions

An alternative for batch TN contraction is to add a "batch" hyperindex

Vectorize batch tensor network contractions

Reactant.jl will be capable of this when we add proper support for it.

Implement HOSVD factorization for `Tensor`

Moved to [bsc-quantic/Tensors.jl](https://github.com/bsc-quantic/Tensors.jl)

Implement Canonical Polyadic Decomposition (CPD) factorization for `Tensor`

Moved to [bsc-quantic/Tensors.jl](https://github.com/bsc-quantic/Tensors.jl)

Fix excessive synchronization in `contract` on `Dagger.DArray`

Updated the code to the last `Tenet` and `Dagger` changes. Also it now calls `Tenet.contract` instead of `OMEinsum.EinCode`. @jofrevalles I also added some tests.

Implement Tensor splitting methods (QR, SVD, LU, ...)

Moved to [bsc-quantic/Tensors.jl](https://github.com/bsc-quantic/Tensors.jl)

Investigate TN contraction precompilation latency

The source of compilation time seems to come from 2 sources: - `OMEinsum` uses dynamic dispatch with runtime-generated `Val` values, that's why first contraction always takes so much time -...

Enhance `replace!(tn::TensorNetwork, ...)` function for more flexible `Tensor` replacements

#55 includes a new `replace!` method that can replace a `Tensor` with a `TensorNetwork` as long as the open indices of the `TensorNetwork` match the labels of the `Tensor`.