Matrix Product Toolkit | Itools / Mp-icorrelation

mp-icorrelation is the main tool for calculating real-space local expectation values and correlation functions. It requires a previously hard-coded model, which describes the local basis of the lattice, and the name of the observable.

Usage: mp-icorrelation [options] <model> <Op1> <Op2> <psi>
Allowed options:
  --help                show this help message
  --nmax arg            Maximum number of bosons per site (for bh model)
  --string arg          make a string correlation function with this operator
  --spin arg            spin per site, for spin models
  --nosite              don't calculate the on-site correlations <op1(i)op2(i)>
  -l [ --length ] arg   length of correlation [default 100]

The --string argument calculates a string correlation function, using this operator as the central site. This would also be needed to calculate a fermionic correlator, where the string will be 'P'.

By default, the program calculates all correlations in the unit cell <op1(x) op2(y)> where x varies over all sites in the first unit cell, and y varies over all sites in unit cells up to the one containing the site specified by -l. The correlation points where x=y are also calculated by default, turn this off with --nosite.

The required parameters are the model, the first and second operators of the two-point correlator, and the wavefunction file. The model is not the same as the Hamiltonian used for mp-idmrg5, but is related: the model contains information only about the local basis of the lattice sites, and a single model may be used with many different Hamiltonian's. As at 18-11-2009, the available models are sf-u1, spin, spin-su2, spin-u1, spin-z2, tj-u1, tj-u1su2, klm-u1su2, hubbard-so4, hubbard-u1, bh, bh-u1.

The --spin and --nmax options are specific to particular models. For spin models, the --spin parameter specifies the local spin at each site. The --nmax parameter specifies the number of bosons per site, for the Bose-Hubbard model.

To calculate local expectation values, use the identity operator 'I' as one of the operators of the two-point correlator.

The output is 4 columns, x y re im being the x,y coordinates of the correlator (starting from site 1), and the real and imaginary components of the correlator.