Fermat MCP

Instructions

Implementation Reference

• fmcp/sympy_mcp/core/matrices.py:59-178 (handler)

  The handler function matrix_operation that executes the tool logic for various SymPy matrix operations including det, inv, rref, eigenvals, and create.
  def matrix_operation( operation: MatrixOperation, data: Union[str, List, Tuple], # Create parameters rational: bool = True, nrows: Optional[int] = None, ncols: Optional[int] = None, # Common parameters simplify: bool = True, ) -> Any: """ Unified interface for matrix operations. Args: operation: The matrix operation to perform. One of: - 'create': Create a matrix from data - 'det': Calculate the determinant - 'inv': Calculate the inverse - 'rref': Compute reduced row echelon form - 'eigenvals': Compute eigenvalues data: The matrix data in one of these formats: - String: "1 2; 3 4" or "[1, 2] [3, 4]" - List of lists: [[1, 2], [3, 4]] - Single list: [1, 2, 3, 4] (will be converted to a row matrix) rational: If True, convert to rational numbers (default: True) nrows, ncols: For creating matrices of specific dimensions simplify: If True, simplify the result (default: True) Returns: The result of the operation in a JSON-serializable format. For 'create', returns the matrix as a list of lists. For 'det', returns the determinant as a float or symbolic expression. For 'inv', returns the inverse matrix as a list of lists. For 'rref', returns a tuple (RREF_matrix, pivot_columns). For 'eigenvals', returns a dictionary of eigenvalues with multiplicities. Examples: >>> matrix_operation('create', '1 2; 3 4') [[1, 2], [3, 4]] >>> matrix_operation('det', [[1, 2], [3, 4]]) -2 >>> matrix_operation('inv', '1 2; 3 4') [[-2.0, 1.0], [1.5, -0.5]] >>> matrix_operation('rref', '1 2 3; 4 5 6') ([[1, 0, -1], [0, 1, 2]], (0, 1)) >>> matrix_operation('eigenvals', '3 -2; 4 -1') {'1 - 2*I': 1, '1 + 2*I': 1} """ # Parse matrix data matrix_data = _parse_matrix_data(data) # Create matrix, optionally converting to rational numbers if rational: matrix = Matrix(matrix_data).applyfunc(lambda x: nsimplify(x, rational=True)) else: matrix = Matrix(matrix_data) # Handle matrix reshaping if nrows/ncols are provided if nrows is not None and ncols is not None: matrix = matrix.reshape(nrows, ncols) elif nrows is not None: matrix = matrix.reshape(nrows, -1) elif ncols is not None: matrix = matrix.reshape(-1, ncols) # Handle different operations if operation == "create": return _convert_to_json_serializable(matrix.tolist()) # For other operations, ensure the matrix is square if needed if operation in ["det", "inv", "eigenvals"] and matrix.rows != matrix.cols: raise ValueError(f"Matrix must be square for {operation} operation") if operation == "det": try: det = matrix.det() return float(det) if det.is_real else str(det) except (TypeError, AttributeError): return str(det) elif operation == "inv": try: inv_matrix = matrix.inv() return _convert_to_json_serializable(inv_matrix.tolist()) except ValueError as e: if "matrix is not invertible" in str(e).lower(): raise ValueError("Matrix is not invertible (determinant is zero)") raise elif operation == "rref": rref_matrix, pivot_columns = matrix.rref() return ( _convert_to_json_serializable(rref_matrix.tolist()), tuple(pivot_columns), ) elif operation == "eigenvals": try: eigenvals = matrix.eigenvals() # Convert to a serializable format with better handling of complex numbers result = {} for val, mult in eigenvals.items(): # Convert to complex and format nicely cval = complex(val.evalf()) if abs(cval.imag) < 1e-10: # Effectively real key = f"{cval.real:.6f}".rstrip("0").rstrip(".") else: # Complex number real_part = f"{cval.real:.6f}".rstrip("0").rstrip(".") imag_part = f"{abs(cval.imag):.6f}".rstrip("0").rstrip(".") sign = " + " if cval.imag >= 0 else " - " key = f"{real_part}{sign}{imag_part}j" result[key] = int(mult) return result except Exception as e: raise ValueError(f"Failed to compute eigenvalues: {str(e)}") else: valid_ops = get_args(MatrixOperation) raise ValueError(f"Invalid operation. Must be one of: {valid_ops}")
• fmcp/sympy_mcp/server.py:26-29 (registration)

  Registers the matrix_operation function as the sympy_mcp_matrix_operation tool in the FastMCP server.
  sympy_mcp.tool( matrix_operation, description="Do symbolic matrix operations like create, det, inv, rref, eigenvals", )
• fmcp/sympy_mcp/core/matrices.py:4-68 (schema)

  Type definitions for the tool parameters, including MatrixOperation literal and function signature serving as input schema.
  # Define operation types for type hints MatrixOperation = Literal["create", "det", "inv", "rref", "eigenvals"] def _parse_matrix_data(data: Union[List, Tuple, str]) -> List[List[Any]]: """Parse matrix data from various input formats.""" if isinstance(data, str): # Handle string input like "1 2; 3 4" or "[1, 2] [3, 4]" if ";" in data: # Handle MATLAB-style format: "1 2; 3 4" rows = [row.strip() for row in data.split(";") if row.strip()] return [[sympify(x) for x in row.split()] for row in rows] else: # Handle list of lists format: "[1, 2] [3, 4]" import ast try: # Try to parse as a list of lists parsed = ast.literal_eval("[" + data.replace(" ", ",") + "]") if all(isinstance(row, (list, tuple)) for row in parsed): return [[sympify(x) for x in row] for row in parsed] elif all(not isinstance(x, (list, tuple)) for x in parsed): # Single row matrix return [[sympify(x) for x in parsed]] except (ValueError, SyntaxError): pass # Try space-separated values return [ [sympify(x) for x in row.split()] for row in data.split("\n") if row.strip() ] # Handle list/tuple input if all(isinstance(row, (list, tuple)) for row in data): return [[sympify(x) for x in row] for row in data] elif all(not isinstance(x, (list, tuple)) for x in data): # Single row matrix return [[sympify(x) for x in data]] raise ValueError("Invalid matrix format") def _convert_to_json_serializable(obj): """Convert SymPy objects to JSON-serializable types.""" if hasattr(obj, "as_immutable"): return str(obj) elif isinstance(obj, (list, tuple)): return [_convert_to_json_serializable(x) for x in obj] elif isinstance(obj, dict): return {str(k): _convert_to_json_serializable(v) for k, v in obj.items()} return obj def matrix_operation( operation: MatrixOperation, data: Union[str, List, Tuple], # Create parameters rational: bool = True, nrows: Optional[int] = None, ncols: Optional[int] = None, # Common parameters simplify: bool = True, ) -> Any:
• fmcp/sympy_mcp/core/matrices.py:8-46 (helper)

  Helper function to parse matrix data from different input formats into SymPy-compatible lists.
  def _parse_matrix_data(data: Union[List, Tuple, str]) -> List[List[Any]]: """Parse matrix data from various input formats.""" if isinstance(data, str): # Handle string input like "1 2; 3 4" or "[1, 2] [3, 4]" if ";" in data: # Handle MATLAB-style format: "1 2; 3 4" rows = [row.strip() for row in data.split(";") if row.strip()] return [[sympify(x) for x in row.split()] for row in rows] else: # Handle list of lists format: "[1, 2] [3, 4]" import ast try: # Try to parse as a list of lists parsed = ast.literal_eval("[" + data.replace(" ", ",") + "]") if all(isinstance(row, (list, tuple)) for row in parsed): return [[sympify(x) for x in row] for row in parsed] elif all(not isinstance(x, (list, tuple)) for x in parsed): # Single row matrix return [[sympify(x) for x in parsed]] except (ValueError, SyntaxError): pass # Try space-separated values return [ [sympify(x) for x in row.split()] for row in data.split("\n") if row.strip() ] # Handle list/tuple input if all(isinstance(row, (list, tuple)) for row in data): return [[sympify(x) for x in row] for row in data] elif all(not isinstance(x, (list, tuple)) for x in data): # Single row matrix return [[sympify(x) for x in data]] raise ValueError("Invalid matrix format")