Hello, if I use the cTPQ method to run HPhi, can I set the LargeValue parameter in the modpara.def file to a very small value, such as 0.1 or even 0.01?

[shuailiu199966]

No description provided.

[tmisawa]

Sure. It is, of course, possible to set a smaller value.
However, please note that using a smaller value will increase the number of iterations required to reach a low temperature.

[shuailiu199966]

Sure. It is, of course, possible to set a smaller value. However, please note that using a smaller value will increase the number of iterations required to reach a low temperature.

Hello, I noticed that when I set LargeValue to 0.1 and Lanczosmax to 200, the calculated results seem to be incorrect. What could be the reason for this?

My energy turned positive, and as the temperature decreases, the energy keeps increasing.

[tmisawa]

I am sorry that I misunderstood the definition of LargeValue.

LargeValue corresponds to the inverse of the imaginary time evolution step size, Δtau (i.e., LargeValue = 1/Δτ).
If LargeValue is set too small, the imaginary time evolution becomes unstable.

In the modpara.def file, you can specify the expansion order of the Taylor expansion using ExpandCoef.
By increasing ExpandCoef, it is possible to use a smaller LargeValue.
However, increasing ExpandCoef results in higher computational costs.

Typically, when setting ExpandCoef to around 10,
it is recommended to set LargeValue to a value greater than 1, such as around 10.