Fit with constraints in TMinuit

Hi rooters,

I am currently performing an unbinned maximum likelihood fit to extract two parameters (a1, a2) from N events. The likelihood function is constructed as:

-lnL=Sum_{i=0}^{N} ln W(\xi_i; a1,a2)

I then attempted to incorporate a constraint between a1 and a2 using the Lagrange multiplier method, modifying the likelihood function to:

-lnL=Sum_{i=0}^{N} ln W(\xi_i; a1,a2) + lambda Func(a1, a2)

However, when minimizing -lnL using TMinuit, the fit consistently hits the boundary regardless of how I adjust the range of lambda, resulting in failed convergence. I would like to ask:

  1. Whether my construction of the likelihood function using the Lagrange multiplier method is correct
  2. Whether TMinuit is an appropriate tool for this constrained minimization approach

thanks for your attention,

Yupeng