BUG: fixed IndexError in pwcca when number of neurons do not match

[maxwell-aladago]

Given two activations z_i \in \mathbb{R}^{n_i, T} and z_j \in \mathbb{R}^{n_j, T} where n_j > n_i, pwcca(z_i, z_j) raises an IndexError whereas pwcca(z_j, z_i) works fine.

This pull request fixes that bug.

[google-cla]

Thanks for your pull request. It looks like this may be your first contribution to a Google open source project (if not, look below for help). Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

📝 Please visit https://cla.developers.google.com/ to sign.

Once you've signed (or fixed any issues), please reply here with @googlebot I signed it! and we'll verify it.

---

What to do if you already signed the CLA

Individual signers

• It's possible we don't have your GitHub username or you're using a different email address on your commit. Check your existing CLA data and verify that your email is set on your git commits.

Corporate signers

• Your company has a Point of Contact who decides which employees are authorized to participate. Ask your POC to be added to the group of authorized contributors. If you don't know who your Point of Contact is, direct the Google project maintainer to go/cla#troubleshoot (Public version).
• The email used to register you as an authorized contributor must be the email used for the Git commit. Check your existing CLA data and verify that your email is set on your git commits.
• The email used to register you as an authorized contributor must also be attached to your GitHub account.

ℹ️ Googlers: Go here for more info.

[brando90]

@arimorcos @rubai5 @yosinski can this be merged?

[brando90]

Given two activations z_i \in \mathbb{R}^{n_i, T} and z_j \in \mathbb{R}^{n_j, T} where n_j > n_i, pwcca(z_i, z_j) raises an IndexError whereas pwcca(z_j, z_i) works fine.

This pull request fixes that bug.

@maxwell-aladago Does this affect the code or what it just a notational bug?

[maxwell-aladago]

Given two activations z_i \in \mathbb{R}^{n_i, T} and z_j \in \mathbb{R}^{n_j, T} where n_j > n_i, pwcca(z_i, z_j) raises an IndexError whereas pwcca(z_j, z_i) works fine.
This pull request fixes that bug.

@maxwell-aladago Does this affect the code or what it just a notational bug?

It affected the code.

[brando90]

@maxwell-aladago I am confused, if the if statement that you left unchanged, why does it matter which of the two layers we choose if they remove the small values anyway in their cca code?

i.e.

    # - choose layer that had less small values removed (since the sum of true indices would be less)
    # to me this is puzzling because since we removed them when computing the
    # sigma_xxs, sigma_yys etc. why does it matter which one we choose? perhaps becuase at the end it uses
    # the actual acts1 value bellow which makes a difference
    if np.sum(sresults["x_idxs"]) <= np.sum(sresults["y_idxs"]):

[brando90]

discussion of this here: moskomule/anatome#27

[brando90]

@maxwell-aladago I also noticed you didn't center the acts1, acts2 (nor did the original code)...isn't that a bug? the cca_core.get_cca_similarity does center but the compute_pwcca does not...