Description

This addresses a Future Warning in pypfopt/hierarchical_portfolio.py caused by multiplying integer-type weights with float values. Due to stricter type enforcement, this would cause errors in future pandas versions.

Files Changed

• pypfopt/hierarchical_portfolio.py

Fix

Explicitly cast weights array to float64 before arithmetic operations:

# Before
w[first_cluster] *= alpha  # weight 1

# After
w = w.astype('float64')  # ensure the weights array is float64
w[first_cluster] *= alpha  # weight 1

Testing Performed

• Ran full test suite with pytest (no failures)
• Manually verified warning elimination
• Confirmed HRP optimization consistency

Checklist

• Code follows project style guidelines
• Documentation updated (if applicable)
• Unit tests pass
• No unintended side effects

This patch ensures floating-point arithmetic and prevents future dtype conflicts in pandas.