I just started to use the OMPS package but I cannot understant how to write an exciton-boson Hamiltonian, or if it is possible at all.
The Hamiltonian is technically comprised of a first site with and arbitrary (Hermitian) Hamiltonian H_e and additional boson sites interacting with the first site. The formal structure of the Hamiltonian would be
H = ∑{n,m} H |n>< m| + ∑k ω_k n_k − ∑ |n><n|g_{nk} (b_k + b_k)
In the above expression the parameters H_{nm} define the exction part and should be arbitrary, and moreover each bosonic site should allow for a different size of the basis (i.e. a different maximum excitation number); n_k is the k-th occupation number; ω_k the frequency of the k-th boson; the constants g_k defines the coupling between the state |n> of the first site and the k-th boson.
Any help is appreciated,
Raffaele
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
the main problem I am facing at the moment is how to define a copuling operator between the first site and all the other sites. Due to lack of symmetries of such a model I think it should be possible to use the rule "ProductTerm" to construct very general exciton-boson MPOs and then add all of them toghether.
However, I don't understand how OMPS performs the summation of MPO operators. Is it possible to perform addition and truncation of operators using SVD to keep bond dimensions at a reasonable level?
Thanks
R
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
the ProductTerm is an n-body interaction, not a two-site term. From your
question, I would have thought that is not what you want.
But I would look into (and the weights for the coupling are crucial):
- ExpTerm with site-dependent coupling only activating the first site. You
could also get a decay depending on the distance. With the proper decay
parameter, it does not decay.
- InfiniteFunction with site-dependent coupling activating the first site.
It uses ExpRule to fit other decay functions and is more expensive.
- TTerm is an interaction (equal strength) of sites 1 .. (kk - 1) with site
k, where you could probably which kk as the last site instead of the first
as you had in mind. I hope there is no check that kk must be smaller than
the system. (This one is less used)
There are no SVDs on the representation of the MPO, but for the
infiniteFunction you have cut-off criteria how many functions you want to
use. If look a bit into our OSMPS paper in CPC, you can estimate the bond
dimension of the MPO to get an estimate of how expensive the simulations
will be.
Let me know if you have further questions once you had a chance to look
into the terms.
the main problem I am facing at the moment is how to define a copuling
operator between the first site and all the other sites. Due to lack of
symmetries of such a model I think it should be possible to use the rule
"ProductTerm" to construct very general exciton-boson MPOs and then add all
of them toghether.
However, I don't understand how OMPS performs the summation of MPO
operators. Is it possible to perform addition and truncation of operators
using SVD to keep bond dimensions at a reasonable level?
Hi,
I just started to use the OMPS package but I cannot understant how to write an exciton-boson Hamiltonian, or if it is possible at all.
The Hamiltonian is technically comprised of a first site with and arbitrary (Hermitian) Hamiltonian H_e and additional boson sites interacting with the first site. The formal structure of the Hamiltonian would be
H = ∑{n,m} H |n>< m| + ∑k ω_k n_k − ∑ |n><n|g_{nk} (b_k + b_k)
In the above expression the parameters H_{nm} define the exction part and should be arbitrary, and moreover each bosonic site should allow for a different size of the basis (i.e. a different maximum excitation number); n_k is the k-th occupation number; ω_k the frequency of the k-th boson; the constants g_k defines the coupling between the state |n> of the first site and the k-th boson.
Any help is appreciated,
Raffaele
Hi,
the main problem I am facing at the moment is how to define a copuling operator between the first site and all the other sites. Due to lack of symmetries of such a model I think it should be possible to use the rule "ProductTerm" to construct very general exciton-boson MPOs and then add all of them toghether.
However, I don't understand how OMPS performs the summation of MPO operators. Is it possible to perform addition and truncation of operators using SVD to keep bond dimensions at a reasonable level?
Thanks
R
Hi,
the ProductTerm is an n-body interaction, not a two-site term. From your
question, I would have thought that is not what you want.
But I would look into (and the weights for the coupling are crucial):
- ExpTerm with site-dependent coupling only activating the first site. You
could also get a decay depending on the distance. With the proper decay
parameter, it does not decay.
- InfiniteFunction with site-dependent coupling activating the first site.
It uses ExpRule to fit other decay functions and is more expensive.
- TTerm is an interaction (equal strength) of sites 1 .. (kk - 1) with site
k, where you could probably which kk as the last site instead of the first
as you had in mind. I hope there is no check that kk must be smaller than
the system. (This one is less used)
You can find the details in the documentation.
https://openmps.sourceforge.io/pyautodoc/mpoterm.html?highlight=expterm#mpoterm.ExpTerm
https://openmps.sourceforge.io/pyautodoc/mpoterm.html?highlight=infinite#mpoterm.InfiniteFunc
https://openmps.sourceforge.io/pyautodoc/mpoterm.html?highlight=tterm#mpoterm.TTerm
There are no SVDs on the representation of the MPO, but for the
infiniteFunction you have cut-off criteria how many functions you want to
use. If look a bit into our OSMPS paper in CPC, you can estimate the bond
dimension of the MPO to get an estimate of how expensive the simulations
will be.
Let me know if you have further questions once you had a chance to look
into the terms.
Best regards,
Daniel
On Wed, Sep 25, 2019 at 2:25 PM lello lelloborrelli@users.sourceforge.net
wrote: