Changes between Initial Version and Version 1 of Ticket #5099
- Timestamp:
- 2018-08-29T16:58:47Z (7 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
Ticket #5099 – Description
initial v1 1 Min/max attributes can have an important influence on the correctness and performance of Modelica models. In particular, this is critical for the optimization of the {{{inStream()}}} , see Section 9.2 of the Modelica Specification.1 Min/max attributes can have an important influence on the correctness and performance of Modelica models. In particular, this is critical for the optimization of the {{{inStream()}}} operator, see Section 9.2 of the Modelica Specification. 2 2 3 The current implementation of inStream (see #3885) opimizes away terms with min >=0 and max <=0after alias elimination. It is thus essential that min/max information is not lost in the process: when many variables in an alias set are represented by one variable in the problem that will be solved numerically, the min/max attributes of this variable should span the intersection of the min/max intervals of each variable in the set.3 The current implementation of {{{inStream()}}} (see #3885) opimizes away terms with {{{min >= 0}}} and {{{max <= 0}}} after alias elimination. It is thus essential that min/max information is not lost in the process: when many variables in an alias set are represented by one variable in the problem that will be solved numerically, the min/max attributes of this variable should span the intersection of the min/max intervals of each variable in the set. 4 4 5 5 In particular, it is important to merge the min/max attributes of variables showing up in alias equations that can stem from connect statements and from typical modelling equations involving flow variables, that is: