Advice from Dr. Gerald

Here is the advise from Dr. Gerald related to Cloud Computing.

These are the things you need to do.
This is the virtual machine software to prepare the image
– Windows operating system installation Create a windows (7 or 8)…. machine
–  Install the MPI in the windows (Just in case is not installed by default)
– Install your Delft 3D with compatibility to MPI library used (for parallel computing)
– Copy your information of your case study. (Run it once to test if it works on your machine – > Please measure the time)
                                   Size of the problem:
                                   Time to run
                                    Operating System
                                    Machine size used:
                                    Memory used in the Machines
                                    Define the units or information that makes the problem complex?
Think about place where your results will be stored.

___

So Let me be diligent. 🙂

Meeting with Mr. Cloud

On Wednesday, I met Dr. Gerald Corzo.

Prof. Dimitri ask him to guide me and Prof. Dimitri also ask me to learn from Dr. Gerald.

He explain me a lot about the chance to run my model in SurfSARA server in Amsterdam. He also explain about how cloud computation is done. We need Virtual Machine. Some of the server use Linux instead of Windows. How we can Apply MPI method to our model. And He also give me several literature related to that.

I learn about sharing mouse and keyboard through several screen. And I have to learn about cloud computing to do my thesis. I hope my computer will be still and bear the load. 😀

New Trial

In the next day, 291013, Dr. Mick send me email about the Columbia Case.

I am also surprised that he had make big concern. He said that:

” by removing the first line in the *.cmp files the model should work..pls let me know…

Suddenly I make some effort to deal with the problem in DFLOW FM.

There is 3 type of file with the extension of *.cmp, which are:

bw_tdm
sea_je
west_bnd

But by examining the content, I guess the sea_je sets are the ones.

After Before
MSF       3.8762000e-002  1.4276000e+002
Q1        5.5356000e-002  2.0668000e+002
O1        2.9836000e-001  2.2216000e+002
NO1       2.3448000e-002  2.0519000e+002

….and so forth

0.0       0.0000000e+000 0.0
MSF       3.8762000e-002  1.4276000e+002
Q1        5.5356000e-002  2.0668000e+002
O1        2.9836000e-001  2.2216000e+002
NO1       2.3448000e-002  2.0519000e+002

….and so forth

I delete the first line. However, I am so sorry, the program is still crash. Hereby I send you the 001.dia file.

Thank you very much, I will keep trying

Sometimes Business

On Monday (281013), I meet Dr. Mick again for consultation.

We try to solve the Columbia Case Study Model that is still stuck. How ever, we still struggling with unknown reason why the model is crush. The bad thing is that Dr. Mick take around 45 minutes for doing other case though we had make an appointment. I am ok, just doing the DFLOW FM tutorial and let him doing his work.

Finally, he join me working with Columbia Problem, however, until lunch time we had not have any solution yet. He said that he will let me know if there is some progress.

Before I go, actually I had forgotten and came back, I ask Dr. Mick Permission for attending the short course about poor data basin with Dr. Luigia Brandimarte.

The Geeks were exhausted

I had a weekly meeting with Dr. Mick and that was terrible 😀 .

I came 5 minutes late and what interesting is that Dr. Mick was just arrived. He sent an apologize email because he thought the appointment was 30 minutes before. And Mr. Johan came one our later, great.

OK, now come to the serious part. Dr. Mick had arrange the Columbia Estuary model in the flexible mesh form. However, we don’t know why it can’t be run. It always crash when we try to conduct the computation. We try to modify the parameter setting and the boundary condition file, and yet the model is still stuck.

About one hour later, Mr. Johan join our group work. On the other hand, Dr. Mick said that he should go to the Deltares in a few minutes. So the handlebar, move to Mr. Johan. He start to debug the code because might be the error was caused by the logic it self and not the input file.

It took around two hour. We stay until the lunch time, until the computer battery is off. And the good news is the case is not yet solved. He send the complain the staff in Deltares. He gave the code and the tutorial for the DFLOW flexible mesh model.

Lately in the afternoon, I check that Mr. Johan delete unstruc.exe and several essential file other file. It made me stuck until the end of the work time. I try to send Dr. Mick email about this problem.

Then I decide to take a cup of coffee for my break. Fortunately, I saw Abdi working in the restaurant. I decide to ask him whether he also has the file that I need at that time.

Lucky me I he has it and I can copy the file. My model then can be run. Means that I have a homework for my week end. 🙂

On Saturday Night, I got a reply from Dr. Mick. He sent all the required file, which also I already got from Abdi. He send unstruc.123 instead of unstruc.exe to avoid security problem in sending the message.

So Let me do my homework. 🙂

 

Running your model in the cloud

Prof. Dimitri is so interested to my idea of physical based model optimization using cloud computing. He ask me to consult with Dr. Gerald Corzo.

Moreover, he also promote several seminar on that subject, such as:

IWSG & Hydroinformatics Lab Technology seminar on parallel computing, tomorrow 22 October, 12:45 , room D3

and

Hydroinformatics Lab Technology Seminar (yet another one). SURFsara high performance computing: Tomorrow, Wednesday, 10:00, room B1

About Volumetric Approach

I decide to provide new approach in my thesis instead of just modelling performance comparison approach.

My First idea is to introduce volumetric salinity criteria in comparison with salt intrusion length criteria, towards estuary salinity analysis.

Dr. Mick seems to support my idea by his comment:

“The proposal looks promising. I am not fully aware of the volume weight method compared to the salt length, but I am eager to explore the methods.”

And it’s time to explain my idea in the better way. 😮

#simple thought in the second week of october

Needs Inovation, again and again

Dr. Mick said: “ok; can we meet tomorrow at 12.00? I have to leave at 12.15…

It just some strange thought on Tuesday 01/10, before I send the full proposal.

On Wednesday 02/10, he was convinced that I had fulfill the project requirement. However a Master Thesis should be has more innovative points.

And that was my homework to think about.