Update README.md + fixed taking square roots to avoid NANs

README.md

@@ -15,6 +15,11 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
 limitations under the License.
 
+## Install
+```
+pip install -U git+https://github.com/google/svcca
+```
+
 ## (SV)CCA for Representational Insights in Deep Neural Networks
 
 This repository contains code and [jupyter notebook tutorials](https://github.com/google/svcca/tree/master/tutorials) on results in the following papers, as well as suggested extensions and open problems.

cca_core.py

@@ -148,8 +148,13 @@ def compute_ccas(sigma_xx, sigma_xy, sigma_yx, sigma_yy, epsilon,
 
   if verbose:
     print("taking square root")
-  invsqrt_xx = positivedef_matrix_sqrt(inv_xx)
-  invsqrt_yy = positivedef_matrix_sqrt(inv_yy)
+  # - gives NAN errors
+  # invsqrt_xx = positivedef_matrix_sqrt(inv_xx)
+  # invsqrt_yy = positivedef_matrix_sqrt(inv_yy)
+  import scipy
+  # print('npy\'s sqrt')
+  invsqrt_xx = scipy.linalg.sqrtm(inv_xx)
+  invsqrt_yy = scipy.linalg.sqrtm(inv_yy)
 
   if verbose:
     print("dot products...")