[ADOL-C] inverse_tensor_eval memory error
Andrea Walther
andrea.walther at uni-paderborn.de
Fri Aug 4 02:14:09 EDT 2017
Dear Brad,
thanks for the hint!
we will have a look at it as soon as possible.
Best regards
Andrea
Am 03.08.2017 um 20:32 schrieb Brad Bell:
> I have am trying to run a simple example using inverse_tensor_eval. To
> be specific
> F(z) = sin(z)
> I have been unable to figure out why I am getting memory errors when I
> run the
> code included below. When I run it in valgrind, I get the following output:
> =================================================================================================
>
> build>valgrind ./adolc
> ==29183== Memcheck, a memory error detector
> ==29183== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
> ==29183== Using Valgrind-3.12.0 and LibVEX; rerun with -h for copyright
> info
> ==29183== Command: ./adolc
> ==29183==
> ==29183== Invalid free() / delete / delete[] / realloc()
> ==29183== at 0x4C2ED4A: free (vg_replace_malloc.c:530)
> ==29183== by 0x4F8B731: inverse_tensor_eval (taylor.c:741)
> ==29183== by 0x401094: main (adolc.cpp:37)
> ==29183== Address 0x699cf38 is 8 bytes inside a block of size 32 alloc'd
> ==29183== at 0x4C2FA50: calloc (vg_replace_malloc.c:711)
> ==29183== by 0x4E63FA5: myalloc2 (adalloc.c:125)
> ==29183== by 0x4F8B1B5: inverse_tensor_eval (taylor.c:688)
> ==29183== by 0x401094: main (adolc.cpp:37)
> ==29183==
> ==29183== Invalid free() / delete / delete[] / realloc()
> ==29183== at 0x4C2ED4A: free (vg_replace_malloc.c:530)
> ==29183== by 0x4F8B74C: inverse_tensor_eval (taylor.c:743)
> ==29183== by 0x401094: main (adolc.cpp:37)
> ==29183== Address 0x699cf98 is 8 bytes inside a block of size 32 alloc'd
> ==29183== at 0x4C2FA50: calloc (vg_replace_malloc.c:711)
> ==29183== by 0x4E63FA5: myalloc2 (adalloc.c:125)
> ==29183== by 0x4F8B1D2: inverse_tensor_eval (taylor.c:689)
> ==29183== by 0x401094: main (adolc.cpp:37)
> ==29183==
> tensor[0][0] = 0.523599
> tensor[0][1] = 1.1547
> tensor[0][2] = 0.7698
> z = 0.523599
> zp = 1.1547
> zpp = 0.7698
> adolc: Done
> ==29183==
> ==29183== HEAP SUMMARY:
> ==29183== in use at exit: 1,094 bytes in 20 blocks
> ==29183== total heap usage: 107 allocs, 89 frees, 19,481,090 bytes
> allocated
> ==29183==
> ==29183== LEAK SUMMARY:
> ==29183== definitely lost: 104 bytes in 6 blocks
> ==29183== indirectly lost: 8 bytes in 1 blocks
> ==29183== possibly lost: 0 bytes in 0 blocks
> ==29183== still reachable: 982 bytes in 13 blocks
> ==29183== suppressed: 0 bytes in 0 blocks
> ==29183== Rerun with --leak-check=full to see details of leaked memory
> ==29183==
> ==29183== For counts of detected and suppressed errors, rerun with: -v
> ==29183== ERROR SUMMARY: 2 errors from 2 contexts (suppressed: 0 from 0)
> build>
> =================================================================================================
>
> # include <adolc/adolc.h>
> int main(void)
> {
> int n = 1; // number of independent and dependent variables
> int d = 2; // highest order derivative
> int p = 1; // number of directions
> int size = 3; // [(p + d) choose d] = (p+d-0) *...* (p+1) / (1
> *...* d)
> //
> double* S[1]; // seed matrix S[n][p]
> S[0] = new double[p];
> //
> double* tensor[1]; // partials of z w.r.t x tensor[n][size]
> tensor[0] = new double[size];
>
> // Set Seed matrix to identity
> S[0][0] = 1.0;
>
> // independent and dependent variables
> adouble az[1], ay[1]; // az[n], ay[n]
> double z[1], y[1]; // z[n], y[n]
>
> // record operations for y = F(z) = sin(z)
> short int tag = 1; // tape identifier
> trace_on(tag); // start recording
> z[0] = 1.0; // value during recording
> az[0] <<= z[0]; // independent variable
> ay[0] = sin(az[0]); // function evaluation
> ay[0] >>= y[0]; // dependent variable
> trace_off(); // turn off recording
>
> // evaluate inverse of F at x0
> double x0 = 0.5; // argument for inverse
> z[0] = std::asin(x0); // F^-1 (x0) = z0
> inverse_tensor_eval(tag, n, d, p, z, tensor, S);
>
> // print tensor
> for(int j = 0; j < size; j++)
> { std::cout << "tensor[0][" << j << "] = " << tensor[0][j] << "\n";
> }
>
> // derivative of F^-1 (x0) = 1.0 / sqrt( 1 - x0 * x0 )
> double zp = 1.0 / std::sqrt(1.0 - x0 * x0 );
>
> // second derivative of F^-1 (x0) = x0 / sqrt( 1 - x0 * x0 )^3
> double zpp = x0 * zp * zp * zp;
>
> // print derivatives for z
> std::cout << "z = " << z[0] << "\n";
> std::cout << "zp = " << zp << "\n";
> std::cout << "zpp = " << zpp << "\n";
> //
> delete [] S[0];
> delete [] tensor[0];
> //
> std::cout << "adolc: Done\n";
> return 0;
> }
>
> _______________________________________________
> ADOL-C mailing list
> ADOL-C at list.coin-or.org
> https://urldefense.proofpoint.com/v2/url?u=https-3A__list.coin-2Dor.org_mailman_listinfo_adol-2Dc&d=DwICAg&c=Ngd-ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=sszAvj5-wJal_YaNkOXa8y4Mts3w-L-xYDXwPYeYptQ&m=eI6wsRAC-v0R8vwRH9XGHzd6xQ3BcC-chdkcdB4kWYo&s=taZPyAvFLleFWuqkK2QcIVoqzT2WsGBZEm3jk310N9o&e=
>
--
Prof. Dr. Andrea Walther
Lehrstuhl fuer Mathematik und ihre Anwendungen
Institut fuer Mathematik
Universitaet Paderborn
Warburger Str. 100
33098 Paderborn
Email: andrea.walther at uni-paderborn.de
Phone: ++49 5251 602721
Fax: ++49 5251 603728
**********
More information about the ADOL-C
mailing list