umfpack-scikit

[ta95....@gmail.com]

Hi, i want to use umfpack in steady state solver then i must install umfpack-scikit for ubunt 15.04. How i can install it?

[Paul Nation]

You actually want Anaconda 2.5 or higher and the development version of qutip 4.0. Then you will have the Intel pardiso solver built-in

On Nov 2, 2016 1:34 PM, <ta95....@gmail.com> wrote:

Hi, i want to use umfpack in steady state solver then i must install umfpack-scikit for ubunt 15.04. How i can install it?

--
You received this message because you are subscribed to the Google Groups "QuTiP: Quantum Toolbox in Python" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qutip+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Pu ZHANG]

Hi, Paul! 

I'm also interested in umfpack usage. So do you mean Anaconda 2.5 + qutip 4.0 automatically includes umfpack? 

Thanks! 

Best regards, Pu Zhang

--

Faculty at School of Physics, Huazhong University of Science and Technology

Room 819 (N.), Yifu Science and Technology Building

1037 Luoyu Road, Wuhan, China

E-mail: puzha...@hust.edu.cn

Homepage: www.researchgate.net/profile/Pu_Zhang4

Phone: +86 18871860394

[Paul Nation]

No, it uses the intel math library instead.

[Pu ZHANG]

Thanks! Is the performance as good as with umfpack? 

BTW, I have another question on qutip performance. I will write another email to the email list. 

Best regards, Pu Zhang

--

Faculty at School of Physics, Huazhong University of Science and Technology

Room 819 (N.), Yifu Science and Technology Building

1037 Luoyu Road, Wuhan, China

E-mail: puzha...@hust.edu.cn

Homepage: www.researchgate.net/profile/Pu_Zhang4

Phone: +86 18871860394

[TLN]

Hi, recently I found that intel pardiso solver is not stable in handling large systems while umfpack still gives good performance (maybe in my specific case). Can you tell me why qutip abandoned umfpack and how can I improve the stability of pardiso?

[Paul Nation]

Can you send the example?

It was dropped because, at the time, Python  support was not very good.  It was difficult to get 64-bit support as well.  And finally, it required users doing another install.  The mkl comes with Anaconda by default.

That being said, should be straight forward to add back again.

-P

On Jul 14, 2017 06:28, "TLN" <long...@gmail.com> wrote:

Hi, recently I found that intel pardiso solver is not stable in handling large systems while umfpack still gives good performance (maybe in my specific case). Can you tell me why qutip abandoned umfpack and how can I improve the stability of pardiso?

--

[NGUYEN Thanh Long]

Thanks for your reply. As an example, I use a chain of spins couple to a common oscillator. The code is enclosed. I know the problem comes from pardiso solver since I traced back the inputs and output of the solver.

Everything is fine with ufmpack. Its right that the user needs to install extra package but not that difficult, at least on linux.

Thanks,

TLN

Here are the outputs of the example code:

N =  2
Nm =  10
Eigen solver:
    <n> =  4.228085894962971      residual_norm: 1.10667173515e-14
Pardiso solver:
    iter 0: <n> =  -4363.206582223964      residual_norm: 28691.38627413263
    iter 1: <n> =  -9336.655010222541      residual_norm: 46653.26204663699
    iter 2: <n> =  8479.99597120219      residual_norm: 25798.952230581053
    iter 3: <n> =  -3601.4477248703256      residual_norm: 13536.177411174811
    iter 4: <n> =  -3601.4477248703256      residual_norm: 13536.177411174811
    iter 5: <n> =  -3601.4477248703256      residual_norm: 13536.177411174811
    iter 6: <n> =  -4363.206582223964      residual_norm: 28691.38627413263
    iter 7: <n> =  -4363.206582223964      residual_norm: 28691.38627413263
    iter 8: <n> =  -3601.4477248703256      residual_norm: 13536.177411174811
    iter 9: <n> =  8479.99597120219      residual_norm: 25798.952230581053
umfpack solver (if installed):
    iter 0: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16
    iter 1: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16
    iter 2: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16
    iter 3: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16
    iter 4: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16
    iter 5: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16
    iter 6: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16
    iter 7: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16
    iter 8: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16
    iter 9: <n> =  4.228085894962969      residual_norm: 2.5077890381016095e-16

--
You received this message because you are subscribed to a topic in the Google Groups "QuTiP: Quantum Toolbox in Python" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/qutip/axGZ-7vtnhI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to qutip+unsubscribe@googlegroups.com.

[Paul Nation]

Should be fixed in Pull #720

To unsubscribe from this group and stop receiving emails from it, send an email to qutip+un...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

<SimulateArray.py>

[long...@gmail.com]

work perfectly now. Thanks

To unsubscribe from this group and stop receiving emails from it, send an email to qutip+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to a topic in the Google Groups "QuTiP: Quantum Toolbox in Python" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/qutip/axGZ-7vtnhI/unsubscribe.

To unsubscribe from this group and all its topics, send an email to qutip+un...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.