Voro++ failed due to a: java.lang.Exception: Parsing failed: First line not "Basis:"

[Former user]

I tried to run the code from this paper: "Including crystal structure attributes in machine learning models of formation energies via Voronoi tessellations". I want to generate the attributes only so I tried to run voro-ml-si/scripts/csv-output/generate-attributes.in. Then it comes with this error: Voro++ failed due to a: java.lang.Exception: Parsing failed: First line not "Basis:". The csv file generated has many lines as NaN.

[Logan Ward]

Thanks for reaching out!

The "First line not 'Basis'" error happens when the Voro++ executable provided with Magpie does not start. I have seen this problem for users running Magpie on OS X before. Are you using OS X? If so, we can fix this by getting you a version of Voro++ compiled for Macs.

[buptxz]

Thank you for your reply!

Yes, I am running on OS X. I have tried to compile Vassal and Magpie on my Mac but ended up with the same error. It is highly appreciated if you could provide me a Mac version :-)

[Logan Ward]

Are you familiar with compiling C++ on Mac?

What you would need to do is:

1. Download Voro++, unpack it in the voro++ folder of Vassal, and then compile the voro++ library (make all)
2. Compile voroExt.cpp (also found in the voro++ folder) by calling: g++ voroExt.cpp -I src/ src/libvoro++.a -o voroExt.exe.
3. Copy voroExt.cpp to Magpie’s “exes” folder, and Magpie should run after you recompile it.

If this would be a problem, I can ask around to see if one of my coworkers with a Mac can compile Voro++ for you.

[buptxz]

Thank you!

I followed your instruction. In the second step, the only voroExt.cpp file I could find was already in the voro++ folder of Vassal, not in the Voro++ source code I just downloaded. I think you mean voroExt.exe in the third step. After I recompiled Magpie, I had another error when I ran it: "Voro++ failed due to a: java.lang.Exception: Voro++ ran the wrong kind of tessellation". I am not sure if it has already solved the previous error.

[Logan Ward]

You are correct about voroExt.exe. Sorry about the typo.

I haven't seen this 'wrong kind of tessellation' error before. Could you run the test.file program in the voro++ folder of Vassal by feeding test.file via stdin (i.e., ./voroExe.exe < test.file), and tell me whether it contains the line "Polydisperse: True"?

[buptxz]

Yes. The output contains this line.

[Logan Ward]

Great, that's what it should do. If you change line 4 of test.file from "poly true" to "poly False"? The output should change to "Polydisperse: False".

It sounds like the previous error of Voro++ not running on your system is solved. I'm not sure why it seems to be reading the incorrect options

[buptxz]

I found that when I changed the line 4 from True to False, the output didn't change correspondingly.

[Logan Ward]

Great! So, we found the problem. I just verified that voroExt.exe works on my computer.

This is pretty weird. It seems your voroExt.exe is detecting a 'T' in 'False' when it shouldn't. I'll put together some debugging code, and an alternative implementation of voroExt soon to help figure this out. Since I haven't yet been able to reproduce this problem on my computer, would you be alright with running that debug code for me?

Just to confirm, does your test.file look like this:

basis 2.88 0 2.88 0 0 2.88
periodicity True true True
nAtoms 2
poly False
0 0 0 1
0.5 0.5 0.5 1

[buptxz]

Sure, I will be happy to.

Yes, that's exactly what I see on my end.

[Logan Ward]

Thanks! I think the problematic part of voroExt.cpp starts at Line 114. Could you replace the following code:

    if (strchr(temp, 'T') != NULL || strchr(temp, 't') != NULL) {
        poly = true;
    }

with this? As you'll likely notice, I changed how we detect what the user input is, and I added a debugging statement.

    std::cerr << "Polydisperse, read: " << temp << std::endl;
    if (strcmp(temp, "True") == 0 || strcmp(temp, "true") == 0) {
        poly = true;
    } else if (strcmp(temp, "False") == 0 || strcmp(temp, "false") == 0) {
        poly = false;
    } else {
        std::cerr << "Polydispersity choice not recognized: " << temp << std::endl;
        return 4;
    }

Could you let me know what the "Polydisperse, read:" line says when you change the "poly" line in the input?

[buptxz]

Thank you! After modifying, Polydisperse output changes corresponding to the input and Magpie runs smoothly after recompiling now.

[Logan Ward]

Glad to hear it worked! Thanks for your patience about this and for testing the changed code!

I'll update Vassal/Magpie to use this newer version, in case someone else has this problem!

[buptxz]

Sounds great. Thank you for your kindly help! By the way, what's a good way to contact you if I have some questions about your paper?

[Logan Ward]

Feel free to email me at [e-mail].

[Logan Ward]

Updated to new version of Voro++

Fixes #18. Thanks @buptxz!

→ <<cset 4e709f37c7ac>>