orthogonalization for tensors

[mganahl]

Initialization of orthogonal tensors with respect to a pivot

[alewis]

Let's maybe meet sometime and write out all the initialization methods we want?

[aeekayy]

@alewis @mganahl is this issue still relevant? If so, any details to share that may be helpful for me?

[GouravWadhwa]

Hi @alewis, I am new to open source and since this is a good first issue, I would like to contribute. Please let me know the details and relevance of this issue

[Alfa-Shashank]

Hi @alewis @mganahl, I am new to open source, I would like to contribute too. Can you please share the relevant details?

[pragyasrivastava0805]

Hi @alewis @mganahl ,I am very new to open source.This issue seems interesting to me,please share the details.I would like to work on it.I also need to do it as a part of my Technical Assessment as a part of application process for a Research Intern Position.Please assign it to me.

[alewis]

Maybe you guys can work together? Basically an initializer must be added so that when a tensor is contracted about a given pivot index, the result is orthogonal. That is, the initializer should create a random tensor, construct its QR decomposition with a given pivot argument, and then return the Q factor.

[pragyasrivastava0805]

@alewis do the changes need to be done here https://github.com/google/TensorNetwork/blob/master/tensornetwork/linalg/initialization.py I am very new to open source.Pls help

[pragyasrivastava0805]

931 Please review my PR

[pragyasrivastava0805]

@alewis @mganahl

[pragyasrivastava0805]

@alewis could you please review my PR.I need it as a part of MSR recruiting

[pragyasrivastava0805]

also please give me an intuitive explanation for doing such an initialization

[pragyasrivastava0805]

@alewis @mganahl