EOMC Home

Research Areas

Research Facilities

People

Collaborators

Prospective Students

Support

Events

Contact






SMF - Space Mapping Framework

SMF (Space Mapping Framework) is a user-friendly, GUI-based software system that implements state-of-the-art optimization algorithms and modeling procedures based on space mapping technology.

SMF performs constrained minimax, L-square or custom optimization, engineering model enhancement and statistical analysis.  It is suitable for engineering modeling and design optimization problems.  It contains a number of features that make it especially suitable for microwave engineering.

SMF Highlights
  • SM-based optimization with flexible surrogate model definition (input/output SM, implicit SM, frequency scaling), and optional trust region methods
  • Adaptive space mapping with automated selection of the surrogate model
  • SM-based local modeling for statistical design and analysis
  • Advanced SM-based modeling including combinations with RBFs, SVR, and fuzzy systems
  • Drivers for the commercial simulators: Sonnet em, Agilent ADS, FEKO, Ansoft HFSS, Ansoft Maxwell, CST Microwave Studio, CST Design Studio, MEFiSTo
  • Database support for models using commercial simulators
  • SM-based interpolation for evaluating a fine model at off-grid points
  • Direct optimization modules for fine, coarse and surrogate models
  • Interactive SM interface for SM-based modeling, data acquisition, space mapping assessment, statistical analysis, interactive SM optimization, and space mapping tuning
  • Exporting graphs, generating report files
  • Storing work in project files
SMF Architecture

SMF system contains five main parts including the setup interfaces, core algorithms, auxiliary algorithms, model drivers, and visualization interfaces. Core algorithms include a range of built-in space mapping and generic surrogate-based optimization procedures, interfaces for automated and interactive optimization and modeling as well as the script interpreter that allows the user to construcst dedicated algorithms using various SMF procedures and capabilities as the building blocks.




There is a number of auxiliary procedured pre-implemented in the SMF system that include parameter extraction algorithms, surrogate model optimization algorithms, interfaces for direct optimization of the high- and low-fidelity models, as well as basic and advanced modeling procedures.

SMF system contains drivers for several commercial electromagnetic and circuit simulators that allow the user to include these simulators in an automated optimization loop. It is possible to define several models and use them for multi-fidelity/multi-level optimization/modeling. All evaluations of the simulator-based models are stored in the SMF database for possible reuse.

SMF provides visualization interfaces that allow the user to observe the fine/coarse/surrogate model responses, visually evaluate the model matching, as well as conduct step-by-step interactive optimization and surrogate model construction and testing.

SMF Snapshots

Automated space mapping optimization interface:



Interactive interface:



Surrogate model optimization interface:



Fine model (direct) optimization interface:



Frequency scaling interface:



Surrogate model testing interface:









2009 Reykjavik University. All rights reserved. Contact koziel@ru.is regarding any issues concerning this site.