This is my mental picture on how we can make predictions from a theory (I'm not a physicist so this might be quite wrong) :

Typically, we solve a partial differential equation (analytically if we can, but often we can only do numerical calculations), where in the **input** of the problem we put the **known quantities** (or the quantities *we don't know but want to simulate*), and in the output, we get the numerical quantities that we can then confront to reality through experiments. 
It's really easy when we have a simple equation such as F=Ma (Newton), and the numerical input is just one quantity (say the intial position and speed of a particle, etc.), the numerical computation is simply a 2-order differential equation, and the numerical output is also one quantity (say the position of the particle at time t) and the experiment involves no statistical error, etc.

But I suppose things get much more complicated in reality with a huge theory such as the Standard Model of particle physics. 
**My question** is precisely this: 

- in the case of the LHC experiments in the discovery of the Higgs, **what is the input ?** 

- **What is the "numerical computation"** (and where is it? 

- How is the theory implemented? 

- Do the experimentalists use specific software? 

- Does it require huge CPU power?) (I'm guessing it's not solving a partial differential equation?)? 

- **What is typically the output ?** 

My mental picture on how they determined the mass of the Higgs is that they confronted whatever output the theory gives when there is a Higgs of mass X (a numerical variable in the computation), and they change the value of X in the numerical computation in order to fit the output with experimental data.

- Is my mental picture correct?