Add PEtab math parser and sympy converter (#260)

Related to PEtab-dev/PEtab#579

* ANTLR grammar for PEtab math expressions + derived lexer/parser/visitor
* Test cases
* PEtab -> SymPy converter

Note: All sympy symbols in PEtab match expressions now have `real=True`.

Author: dweindl

petab/calculate.py

@@ -5,7 +5,7 @@
 
 import numpy as np
 import pandas as pd
-import sympy
+import sympy as sp
 
 import petab
 
@@ -144,7 +144,7 @@ def calculate_residuals_for_table(
     return residual_df
 
 
-def get_symbolic_noise_formulas(observable_df) -> dict[str, sympy.Expr]:
+def get_symbolic_noise_formulas(observable_df) -> dict[str, sp.Expr]:
     """Sympify noise formulas.
 
     Arguments:
@@ -167,7 +167,7 @@ def get_symbolic_noise_formulas(observable_df) -> dict[str, sympy.Expr]:
 
 def evaluate_noise_formula(
     measurement: pd.Series,
-    noise_formulas: dict[str, sympy.Expr],
+    noise_formulas: dict[str, sp.Expr],
     parameter_df: pd.DataFrame,
     simulation: numbers.Number,
 ) -> float:
@@ -192,16 +192,18 @@ def evaluate_noise_formula(
     )
     # fill in measurement specific parameters
     overrides = {
-        f"noiseParameter{i_obs_par + 1}_{observable_id}": obs_par
+        sp.Symbol(
+            f"noiseParameter{i_obs_par + 1}_{observable_id}", real=True
+        ): obs_par
         for i_obs_par, obs_par in enumerate(observable_parameter_overrides)
     }
 
     # fill in observables
-    overrides[observable_id] = simulation
+    overrides[sp.Symbol(observable_id, real=True)] = simulation
 
     # fill in general parameters
     for row in parameter_df.itertuples():
-        overrides[row.Index] = row.nominalValue
+        overrides[sp.Symbol(row.Index, real=True)] = row.nominalValue
 
     # replace parametric measurement specific parameters
     for key, value in overrides.items():

petab/math/PetabMathExprLexer.g4

@@ -0,0 +1,35 @@
+// Lexer grammar for PEtab math expressions
+// run `regenerate.sh` to regenerate the lexer
+lexer grammar PetabMathExprLexer;
+
+
+NUMBER          : EXPONENT_FLOAT | INTEGER | POINT_FLOAT | INF;
+INTEGER         : DIGITS ;
+EXPONENT_FLOAT   : (INTEGER | POINT_FLOAT) EXPONENT ;
+POINT_FLOAT      : DIGITS '.' DIGITS ;
+fragment EXPONENT: ('e' | 'E') ('+' | '-')? DIGITS ;
+FLOAT_NUMBER: POINT_FLOAT | EXPONENT_FLOAT;
+fragment DIGITS : [0-9]+ ;
+
+WS      : [ \t\r\n]+ -> skip ;
+TRUE    : 'true' ;
+FALSE   : 'false' ;
+INF     : 'inf' ;
+NAME : [a-zA-Z_][a-zA-Z0-9_]* ;
+OPEN_PAREN : '(' ;
+CLOSE_PAREN : ')' ;
+BOOLEAN_OR : '||' ;
+BOOLEAN_AND : '&&' ;
+GT : '>' ;
+LT : '<' ;
+GTE : '>=' ;
+LTE : '<=' ;
+EQ : '==' ;
+NEQ : '!=' ;
+PLUS : '+' ;
+MINUS : '-' ;
+ASTERISK : '*' ;
+SLASH : '/' ;
+CARET: '^';
+EXCLAMATION_MARK: '!';
+COMMA: ',';

petab/math/PetabMathExprParser.g4

@@ -0,0 +1,42 @@
+// Parser grammar for PEtab math expressions
+// run `regenerate.sh` to regenerate the parser
+parser grammar PetabMathExprParser;
+
+options { tokenVocab=PetabMathExprLexer; }
+
+petabExpression:
+    expr EOF ;
+
+expr:
+    <assoc=right> expr '^' expr             # PowerExpr
+    | ('+'|'-') expr                        # UnaryExpr
+    | '!' expr                              # BooleanNotExpr
+    | expr ('*'|'/') expr                   # MultExpr
+    | expr ('+'|'-') expr                   # AddExpr
+    | '(' expr ')'                          # ParenExpr
+    | expr comp_op expr                     # ComparisonExpr
+    | expr (BOOLEAN_AND | BOOLEAN_OR) expr  # BooleanAndOrExpr
+    | number                                # Number_
+    | booleanLiteral                        # BooleanLiteral_
+    | functionCall                          # functionCall_
+    | var                                   # VarExpr_
+    ;
+
+comp_op:
+    GT
+    | LT
+    | GTE
+    | LTE
+    | EQ
+    | NEQ
+    ;
+
+argumentList: expr (',' expr)* ;
+functionCall: NAME OPEN_PAREN argumentList CLOSE_PAREN ;
+
+booleanLiteral:
+    TRUE
+    | FALSE
+    ;
+number: NUMBER ;
+var: NAME ;