To solve John's problem, we first need to convert the natural language description into a symbolic representation of the optimization problem.

Let's denote:
- $x_1$ as the pounds of pears John should eat.
- $x_2$ as the pounds of broccoli John should eat.

The objective function is to minimize the total cost, which can be represented algebraically as $6x_1 + 8x_2$, since a pound of pears costs $6 and a pound of broccoli costs $8.

The constraints are:
- Calcium requirement: $2x_1 + 4x_2 \geq 15$
- Potassium requirement: $5x_1 + 3x_2 \geq 20$
- Magnesium requirement: $3x_1 + 6x_2 \geq 17$
- Non-negativity constraint (since John cannot eat a negative amount of food): $x_1, x_2 \geq 0$

Thus, the symbolic representation of the problem is:
```json
{
    'sym_variables': [('x1', 'pounds of pears'), ('x2', 'pounds of broccoli')],
    'objective_function': '6*x1 + 8*x2',
    'constraints': [
        '2*x1 + 4*x2 >= 15',
        '5*x1 + 3*x2 >= 20',
        '3*x1 + 6*x2 >= 17',
        'x1 >= 0',
        'x2 >= 0'
    ]
}
```

Now, let's implement this problem using Gurobi in Python:
```python
from gurobipy import *

# Create a new model
m = Model("Johns_Diet")

# Define the variables
x1 = m.addVar(vtype=GRB.CONTINUOUS, name="pounds_of_pears")
x2 = m.addVar(vtype=GRB.CONTINUOUS, name="pounds_of_broccoli")

# Set the objective function
m.setObjective(6*x1 + 8*x2, GRB.MINIMIZE)

# Add the constraints
m.addConstr(2*x1 + 4*x2 >= 15, "calcium_requirement")
m.addConstr(5*x1 + 3*x2 >= 20, "potassium_requirement")
m.addConstr(3*x1 + 6*x2 >= 17, "magnesium_requirement")

# Optimize the model
m.optimize()

# Print the results
if m.status == GRB.OPTIMAL:
    print("Optimal solution found.")
    print(f"Pounds of pears: {x1.x}")
    print(f"Pounds of broccoli: {x2.x}")
    print(f"Total cost: ${6*x1.x + 8*x2.x:.2f}")
else:
    print("No optimal solution found.")

```