Fix README rand_sparse_tri QuickStart example (#92)

[NusaybahA]

Summary

• Replaced the invalid unit_triangular parameter with strict in the README Quickstart examples to align with the current rand_sparse_tri() API.
• Updated example usage to reflect the correct supported parameters.
• Clarified the sparsity comment to accurately describe the generated 1000×1000 dense matrix size and avoid confusion around event_size=1000.

Testing

• Verified that the README examples now use the correct strict keyword.
• Confirmed the sparsity comment accurately reflects the matrix dimensions and sparsity ratio.

[aymuos15]

Congrats on your first contribution! 🎉

The strict fix is correct, and requires_grad_(True) fixes a second latent bug: without it, the trailing scale_tril.grad.shape line hits AttributeError on None. I ran the snippet end-to-end and it works now.

Maybe another small change - Three comments still say "unit triangular," but strict=True gives a zero diagonal, not a unit one.

-# Create sparse lower triangular matrix (unit triangular, no diagonal)
+# Create sparse lower triangular matrix (strictly lower triangular, no diagonal)

 dist_ldlt = SparseMultivariateNormal(
     loc=loc,
     diagonal=diagonal,
-    scale_tril=scale_tril  # Unit lower triangular
+    scale_tril=scale_tril  # Strictly lower triangular
 )

 dist_precision = SparseMultivariateNormal(
     loc=loc,
     diagonal=precision_diagonal,
-    precision_tril=precision_tril  # Unit triangular precision factor
+    precision_tril=precision_tril  # Strictly triangular precision factor
 )

[NusaybahA]

After reviewing the code I updated any remaining comments that included "unit triangular" ensuring they align with the rest of the code and so that the comments are clear to read.

[aymuos15]

@theo-barfoot LGTM

[theo-barfoot]

Thanks for this, this looks good. The replacement of unit_triangular with strict correctly matches the current rand_sparse_tri API, and the values used for the LDLᵀ and LLᵀ examples are appropriate. Adding requires_grad_(True) also fixes the gradient section at the end of the example.

I found one small pre-existing terminology issue, that was my mistake, which it would be good to correct. In the updated sparsity comment: 5,000 nonzeros in a 1000 × 1000 matrix corresponds to 0.5% density, or 99.5% sparsity.

So I would change the comment to:

nnz=5000, # 5000 nonzeros per 1000x1000 matrix (99.5% sparsity)

Once that wording is corrected, this looks ready to merge.

[NusaybahA]

I have now corrected the sparsity comment to state a 99.5% sparsity, to make the wording clear.

[aymuos15]

Thanks for the review @theo-barfoot !

[theo-barfoot]

Looks good! Congrats on your first contribution 😃