2

1I am currently using a structured array to save some measurements from a sensor. The array (named "data") is of the dimension 2000x3 with the three fields: "samples", "timestamp" and "labels", where samples is a vector of 6 elements. For example, one row looks like this: 

([-19.837763650963275, -19.61692779005053, -18.5301618270122, -13.413484985874076, -13.192649124961326, -12.105883161923], 0.0, 0)
[('samples', '<f8', (6,)), ('timestamp', '<f8'), ('labels', '<i4')]

If I now want to have all samples of one row i could access the row like this: 

data["samples"][10]

which works fine but if i turn it around and write it like this: 

data[10]["samples"]

I get the following error:

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Does anyone know why this is happening?

EDIT: For a better understanding here are the first 10 rows of the "data" array: 

[ ([-19.837763650963275, -19.61692779005053, -18.5301618270122,       -13.413484985874076, -13.192649124961326, -12.105883161923], 0.0, 0)
 ([-18.66282477705446, -18.449317421024432, -17.369283339067675, -12.357118609142269, -12.145937637117552, -11.068086720774204], 0.0, 0)
 ([-17.69388207920866, -17.49198382816449, -16.417085567075173, -11.53193799727324, -11.33976221074188, -10.268890664091229], 0.0, 0)
 ([-16.868088042481606, -16.683019283158636, -15.610803357043569, -10.88697503359368, -10.732729221349611, -9.663174869208511], 0.0, 0)
 ([-16.152597338007514, -15.99074244228478, -14.917542852993487, -10.420257243109129, -10.371835643625495, -9.274470774420056], 0.0, 0)
 ([-15.527885804583931, -15.39736918876727, -14.317660328685264, -10.451759205557037, -10.011944429521288, -9.006031495483906], 0.0, 0)
 ([-14.981993405744573, -14.894234811642814, -13.799157258392123, -10.01393043959338, -9.563854671143899, -8.587392657229502], 0.0, 0)
 ([-14.508475562047407, -14.48315918653869, -13.355984261155621, -9.577456476813333, -9.16920566037696, -8.205174149255923], 0.0, 0)
 ([-14.106856958780497, -14.199089434545343, -12.989626950396643, -9.214711746255777, -8.944244361010687, -7.942834820090279], 0.0, 0)
 ([-13.789298779585817, -13.943732248940886, -12.72321280228509, -9.208476598556874, -8.629970466866272, -7.690122078593869], 0.0, 0)]

EDIT 2: I'm using numpy version 1.10.4

And I'm actually not using 

data[0]["samples"]

but I tried it and it doesn't work as well. I'm using it this way: 

for row in data:
    print(row["samples"])

I did the most basic example that I could think of and it this works surprisingly:

I did the most basic example that I could think of and it this works surprisingly: 

a = np.zeros(10, dtype = [("a", "5f8"), ("b", "i4")])
    print(a["a"][5])   # Works 
    print(a[5]["a"])   # Surprisingly works as well
BStadlbauer
  • 1,287
  • 6
  • 18
  • Show more of data. Right now, you show one line, but three columns – Merlin Jun 11 '16 at 14:01
  • I edited the post to show more rows of the data array, as mentioned above it has about 2000 rows. – BStadlbauer Jun 11 '16 at 14:16
  • What is your numpy version? That looks like a bug. More of the error message stack might help. – hpaulj Jun 11 '16 at 14:28
  • Show entire code that uses 'data[10]["samples"]' – Merlin Jun 11 '16 at 14:54
  • There was a bug in earlier numpy versions in which the order of field and row indexing did matter. I can look that up. But we need to make sure it's a version issue. I have no problems with this kind of indexing on my newish numpy version. – hpaulj Jun 11 '16 at 15:06
  • @hpaulj - Numpy Version is 1.10.4 and I added the post above with a small example that surprisingly works – BStadlbauer Jun 12 '16 at 13:46
  • I suppose the error stack tells us nothing, since the error is occurring in compiled code (`row.__getitem__`). – hpaulj Jun 12 '16 at 16:32

0 Answers0