InfiniteMPS excitations

[jake]

Hi ITensors Devs,

Very quick question. Are any plans to implement functionality for excitations/quasi-particle ansatz outlined in the literature? If not, if I we’re to write something and provide some benchmarks could I contribute to the package? Thanks.

Best,
Jake

[ryanlevy]

Regardless of what the ITensor devs respond, let me advocate for writing your own public library (shameless example plug: GitHub - ryanlevy/ITensorIMPSTools.jl: Tools for ITensorInfiniteMPS ) and then working with them to merge things in parallel/later (the new UMPS slicing came from this). You may also find while writing the package your API thoughts change and it’ll give you a testing ground which is much more fluid.

Note the underlying ecosystem is being rewritten (see GitHub - ITensor/ITensorDocsNext: Documentation for an in-progress rewrite of the ITensor ecosystem. ITensor is supported by the Simons Foundation's Flatiron Institute. ) and so making something as extensible as possible is a good idea, while simultaneously suiting your needs.

[miles]

Hi Jake,
Ryan’s answer is the right one. The short answer from the developer side is that we don’t have plans in the near term to implement the exicitation / quasiparticle ansatz.

The longer answer is that we are generally moving away from building the core library around physics concepts and black-box physics algorithms, and more toward the math concepts underpinning these algorithms. So what we hope to do is build in better support for things like infinite tensor networks and for ideas relating to taking derivatives of elements of tensor networks and contracting these with orthogonal vectors / tensors. (These are the main concepts underlying building up ‘tangent spaces’ in which excitations live.)

Then with such sub-structure, it becomes easier to build domain-specific and/or black box codes on top.

Meanwhile, we strongly encourage people in the community to build domain-specific overlay libraries for themselves and others to use. A good example is Ryan’s ITensorEntropyTools library which offers special but fairly general features for computing reduced density matrices from tensor networks.