Skip to content

Comparison with Pareto Analysis

How ASCICat relates to Pareto frontier methods.

Overview

Aspect Pareto ASCICat
Output Non-dominated set Ranked list
Preferences Not required Explicit weights
Comparability Within dataset Across studies
Final selection Subjective Deterministic

Pareto Frontier

A solution is Pareto-optimal if no other solution is better on all objectives.

Stability
    ↑
    |   ★ Pareto frontier
    |  ★ ★
    | ★   ★
    |★     ★
    |       ★
    |_________★___→ Activity

Complementary Methods

Pareto First, ASCI Second

  1. Identify Pareto-optimal catalysts
  2. Apply ASCI ranking within Pareto set
  3. Get prioritized list of non-dominated solutions
# Find Pareto-optimal
pareto_mask = is_pareto_optimal(objectives)
pareto_results = results[pareto_mask]

# Rank within Pareto set
pareto_ranked = pareto_results.sort_values('ASCI', ascending=False)

Validation

Top ASCI catalysts should be predominantly Pareto-optimal:

  • If not → scoring functions may be miscalibrated
  • Typical overlap: 80-95% of top 100

When to Use Each

Use Pareto When:

  • Exploring trade-off space
  • Stakeholders disagree on priorities
  • No clear weight rationale
  • Generating options for discussion

Use ASCI When:

  • Priorities can be quantified
  • Need single prioritized list
  • Cross-study comparison needed
  • Reproducibility required

Empirical Observations

From HER catalyst screening:

Metric Value
Total catalysts 48,312
Pareto-optimal ~2,000 (4%)
Top 100 ASCI in Pareto ~90%
Top 10 ASCI in Pareto ~100%

Key Insight

Complementarity

ASCICat and Pareto methods are complementary, not competing.

  • Pareto shows the possible trade-offs
  • ASCI provides actionable priorities

References

  • Miettinen, K. Nonlinear Multiobjective Optimization (Springer, 1998)
  • Ehrgott, M. Multicriteria Optimization (Springer, 2005)