plateform stringclasses 1
value | repo_name stringlengths 13 113 | name stringlengths 3 74 | ext stringclasses 1
value | path stringlengths 12 229 | size int64 23 843k | source_encoding stringclasses 9
values | md5 stringlengths 32 32 | text stringlengths 23 843k |
|---|---|---|---|---|---|---|---|---|
github | microsoft/ADBench-master | a_cumsum.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_cumsum.m | 367 | utf_8 | dc9e4b4f50747d81177ed2d7a1f6d3e7 | % function z = a_cumsum(adj, ind, a, b)
%
% Copyright (C) 2015 Johannes Willkomm
function z = a_cumsum(adj, ind, a, b)
if nargin < 4
b = adimat_first_nonsingleton(a);
end
switch ind
case 1
J = partial_cumsum(a, b);
z = reshape(adj(:).' * J, size(a));
case 2
z = a_zeros(0);
end
end
% $Id: a... |
github | microsoft/ADBench-master | a_fftn.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_fftn.m | 484 | utf_8 | a3beeb74ecdbe9045d23adcda81e99c4 | % function adj = a_fftn(adj, ind, a, sz)
%
% Compute adjoint of z = fftn(x, sz?).
%
% see also a_fft
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2013 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
function adj = a_fftn(adj, ind, a, szf)
if nargin < 4... |
github | microsoft/ADBench-master | dpartial_expm1.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/dpartial_expm1.m | 556 | utf_8 | ff877ce8f75e1a3abaf96633cb536192 | % function [dpartial y] = partial_expm1(x)
%
% Compute partial derivative diagonal of y = expm1(x).
%
% This returns the diagonal of the partial derivative. The partial
% derivative matrix can be obtained by using diag(partial).
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkom... |
github | microsoft/ADBench-master | a_adimat_hess_H.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_adimat_hess_H.m | 20,762 | utf_8 | 0657407f5ac88ed837da904ffd1aa11f | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | admMatMultV3TMc.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admMatMultV3TMc.m | 393 | utf_8 | 69cb1bb7e6f9aafda1a1f5289f68b5d0 | % function Z = admMatMultV3TMc(A, B)
%
% see also admMatMultV3
%
% Copyright (C) 2014 Johannes Willkomm
%
function Z = admMatMultV3TMc(A, B)
% A is (m x n x k) tensor
% B is (n x p) matrix
szA = size(A);
szB = size(B);
Z = zeros([szA(1), szB(2), size(A,3)]);
for k=1:size(A,3)
Z(:,:,k) = A(:,:,k) *... |
github | microsoft/ADBench-master | admTaylorVFor.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admTaylorVFor.m | 4,144 | utf_8 | 4bf100b1e984610cd2a6caae4bd6206c | % function [Jacobian, varargout] = admTaylorVFor(handle, seedMatrix, varargin)
%
% Derivatives computed with this function are precise up to machine
% precision. The runtime depends linearly on the number of columns of
% the seedMatrix, but only with a relativly small factor. In order to
% compute products with the... |
github | microsoft/ADBench-master | adimat_adjdiv.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_adjdiv.m | 699 | utf_8 | cdae9bb08adaecd403a21ab04674a39e | % function [adj1 adj2] = adimat_a_mrdivide(divident, divisor, adj)
%
% This determines the adjoints of expr = divident / divisor where adj
% is the adjoint of expr.
%
% see also adimat_a_mldivide
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2010-2014 Johannes Willkomm
%
function [adj1 adj2] = ... |
github | microsoft/ADBench-master | adimat_min22.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_min22.m | 751 | utf_8 | 8e853996737c1f030bc52c502710d1d9 | % function adj = adimat_min22(val1, val2, adj)
%
% This determines the adjoint of val2 in z = min(val1, val2), where
% the adjoint of z is given as parameter adj.
%
% see also a_mean
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2010-2013 Johannes Willkomm, Institute for Scientific Computing
... |
github | microsoft/ADBench-master | adimat_leigvect.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_leigvect.m | 458 | utf_8 | 42cc61f6436db225f7f76640aa2c1de6 | % function [L] = adimat_leigvect(A, lambda, R)
%
% This file is part of the ADiMat runtime environment
%
% Copyright (C) 2015 Johannes Willkomm
function [L, l_out] = adimat_leigvect(A, lambda, R)
neigs = 1;
tol = 1e-13;
[L Ds] = eigs(A.', neigs, lambda);
[closest, i] = min(abs(diag(Ds) - lambda));
L = L(:,i).... |
github | microsoft/ADBench-master | adimat_adjmultr.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_adjmultr.m | 720 | utf_8 | fbbdfeecf15ad2472e823b27af8c21d5 | % function r = adimat_adjmultr(val, factor, adj)
%
% This determines the adjoint value of val in expr = factor * val
% where adj is the adjoint of expr.
%
% If val is a scalar return adimat_allsum(adj .* factor)
% otherwise return factor' * adj.
%
% see also adimat_adjmultl, adimat_adjmultm, adimat_adjred,
% ad... |
github | microsoft/ADBench-master | partial_logm.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/partial_logm.m | 635 | utf_8 | fc61bc4dc565fd4c789e57abb9db19cf | % function [partial z] = partial_logm(x)
%
% Compute partial derivative of z = logm(x). Also return the function
% result z.
%
% partial is computed as the matrix inverse of partial_expm(logm(x)).
%
% see also g_logm, a_logm
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, I... |
github | microsoft/ADBench-master | admClearFunctions.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admClearFunctions.m | 677 | utf_8 | 09a0cc9e8436cb8d57377ab5380153eb | % function r = admClearFunctions(functionName, prefix)
%
% see also admOptions, admDiffFor, admDiffVFor, admDiffRev.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2011 Johannes Willkomm, Scientific Computing Group
% TU Darmstadt
function r = admClearFunctions(functionName, prefix... |
github | microsoft/ADBench-master | admComputeDerivComplex.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admComputeDerivComplex.m | 2,110 | utf_8 | 7989db897173cf939b90a2f286bd9324 | % function [Jacobian, varargout] = ...
% admComputeDerivComplex(diffFunction, seedMatrix, ...
% independents, dependents, fNargout, varargin)
%
% see also admDiffComplex.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright (C) 2014 Johannes Willkomm
% Copyright (C) 2010,2... |
github | microsoft/ADBench-master | partial_interp3.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/partial_interp3.m | 748 | utf_8 | 01667ad70a7ad5ec07017bf57921a8e0 | % function [partial z] = partial_interp3(xs,ys,zs, vs, xis,yis,zis,...)
% compute partial interp3(xs,ys,z,s vs, xis,yis,zis,...) w.r.t. to vs
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2013 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
function [pa... |
github | microsoft/ADBench-master | a_adimat_qr.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_adimat_qr.m | 16,621 | utf_8 | 03ce83aaf790617a36d70f7f729f6c63 | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | a_angle.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_angle.m | 378 | utf_8 | 4a55c0de198dffe1cfca3916adaa7e02 | % function a_x = a_angle(a_z, x)
%
% Compute adjoint of z = angle(x), where a_z is the adjoint of z.
%
% see also a_atan2
%
% Copyright (c) 2016 Johannes Willkomm
function a_x = a_angle(a_z, x)
t1 = real(x);
t2 = imag(x);
divi = (t2.^2 + t1.^2);
a_zr = real(a_z);
a_x = complex(-a_zr .* t2, -a_zr .* t1) ./ div... |
github | microsoft/ADBench-master | admMergeArgs2.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admMergeArgs2.m | 1,009 | utf_8 | 7ce54507996cec8a1194a5ba50dd42f9 | % function [dfargs] = admMergeArgs2(derivArgs2, derivArgs, funcArgs, independents)
%
% see also admDiffRev, admDiffVFor, admTransform, createFullGradients,
% createSeededGradients, admJacFor, admOptions, adimat_derivclass.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2010,2011,2014 Johannes Wi... |
github | microsoft/ADBench-master | partial_roots.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/partial_roots.m | 744 | utf_8 | 993a4267e46806c3cc5470d5e2cf32d5 | % function [partial r] = partial_roots(c)
% Compute partial derivative of r = roots(c). Also return the
% function result r.
%
% Computation according to ansatz from Martin Bücker using implicit
% function theorom.
%
% see also g_roots, a_roots
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2... |
github | microsoft/ADBench-master | adimat_adjdivr.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_adjdivr.m | 601 | utf_8 | 13255149ef5cb95fac8938188030f053 | % function adj = adimat_a_mrdivider(divident, divisor, adj)
%
% This determines the adjoint of divisor in expr = divident / divisor
% where adj is the adjoint of expr.
%
% see also adimat_a_mldividel
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2010-2014 Johannes Willkomm
%
function adj = adim... |
github | microsoft/ADBench-master | partial_trapz_X.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/partial_trapz_X.m | 1,044 | utf_8 | d92a73565d1251d297933846b2fa6c81 | % function z = partial_trapz_X(arg...)
%
% Compute partial derivative of trapz function w.r.t. X
%
% For calling sequence see trapz.
%
% see also trapz
%
% Copyright (C) 2015 Johannes Willkomm
function J = partial_trapz_X(a, b, c)
if nargin == 1
% trapz(Y)
J = [];
elseif nargin == 2
if isscalar(b)
... |
github | microsoft/ADBench-master | dpartial_sin.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/dpartial_sin.m | 445 | utf_8 | 67e38a7fb5ffe7dfcc0e2a5744fc434a | % function [dpartial y] = dpartial_sin(x)
%
% Compute partial derivative diagonal of y = sin(x).
%
% see also dpartial_exp.
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Fachgebiet Scientific Computing
% TU Darmstadt
function [dpartial y] = dpartial_si... |
github | microsoft/ADBench-master | admRemoveNonTransformOptions.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admRemoveNonTransformOptions.m | 1,748 | utf_8 | 705eb0b525d83586aa38e051e1d7b2cc | % function r = admRemoveNonTransformOptions(s)
%
% Remove fields from admOptions structure that do not influence
% transformation.
%
% see also admOptions, admCompareOptions, admCheckWhetherToTransformSource
%
% Copyright 2011 Johannes Willkomm, Institute for Scientific Computing
% RWTH Aachen Uni... |
github | microsoft/ADBench-master | adimat_mrdivide.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_mrdivide.m | 386 | utf_8 | cd842a83970dc66cfb881ec78b05f5a2 | % function z = adimat_mrdivide(A, B)
%
% Model MATLAB's / operator (mrdivide), for automatic differentiation.
%
% Copyright 2013 Johannes Willkomm, Fachgebiet Scientific Computing
% TU Darmstadt
function z = adimat_mrdivide(A, B)
if isscalar(B)
z = A ./ B;
else
z = linsolve(B', A')';
... |
github | microsoft/ADBench-master | dpartial_coth.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/dpartial_coth.m | 460 | utf_8 | 75e2ffd99307362493a04bf3bc2fdbd7 | % function [dpartial y] = dpartial_coth(x)
%
% Compute partial derivative diagonal of y = coth(x).
%
% see also dpartial_exp.
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Fachgebiet Scientific Computing
% TU Darmstadt
function [dpartial y] = dpartial_... |
github | microsoft/ADBench-master | a_horzcat.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_horzcat.m | 699 | utf_8 | 5c3a7cb341985e4f1e99862a4115f213 | % function adj = a_horzcat(adj, v1, v2, ..., vi)
%
% compute adjoint of horzcat(v1, v2, ..., vi, ..., vn) w.r.t. vi.
%
% see also a_vertcat, a_sum
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2011-2013 Johannes Willkomm, Institute for Scientific Computing
% RWTH Aachen Unive... |
github | microsoft/ADBench-master | a_besselk.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_besselk.m | 573 | utf_8 | 84fab448d59745ba43335f33d290019f | % function [adj] = a_besselk(adj, nu, x, opt?)
%
% compute the adjoint of x in z = besselk(nu, x, opt?), where adj is
% the adjoint of z.
%
% see also dpartial_besselk
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2013 Johannes Willkomm, Fachgebiet Scientific Computing
% TU D... |
github | microsoft/ADBench-master | tay_zerosv1.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/tay_zerosv1.m | 607 | utf_8 | 0733f0e0e5f3100de26e260aefc70d88 | % function varargout = a_zeros1(v)
%
% Create zero adjoints. This function creates zero derivative objects
% of the current derivative class selected with adimat_derivclass, by
% calling the function g_zeros. These adjoints are used in vector
% reverse mode.
%
% see also g_zeros, adimat_derivclass, adimat_adjoint
%
% T... |
github | microsoft/ADBench-master | g_cross3.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/g_cross3.m | 452 | utf_8 | e25f40b7e817f49908fdc3f9ad1b1ab6 | % function g_z = g_cross3(g_a, a, g_b, b, dim)
%
% Compute derivative of z = cross(a, b, dim).
%
% This file is part of the ADiMat runtime environment.
%
% Copyright (c) 2013 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
% Copyright (c) 2016 Johannes Willkomm
function g_z = g_... |
github | microsoft/ADBench-master | a_eig_01_11.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_eig_01_11.m | 614 | utf_8 | 649423679ba658b2599b82a43de60f32 | % function a_A = a_eig_01_11(a_D, x)
%
% Compute adjoint of A in [V D] = eig(A), given adjoint of D.
%
% see also a_eig_10_11, a_eig_11_11, a_eig_1_11
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
funct... |
github | microsoft/ADBench-master | a_adimat_sol_qr.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_adimat_sol_qr.m | 9,336 | utf_8 | 151b862d6e26ece35b3b36cbc838688e | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | d_adimat_prod1.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/d_adimat_prod1.m | 3,608 | utf_8 | d2511c6f058db0cda6d61e67291e988f | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | g_adimat_norm1.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/g_adimat_norm1.m | 2,307 | utf_8 | 71ba15b3ca525be6b95baf81823b110c | % Generated by ADiMat 0.6.0-4867
% Copyright 2009-2013 Johannes Willkomm, Fachgebiet Scientific Computing,
% TU Darmstadt, 64289 Darmstadt, Germany
% Copyright 2001-2008 Andre Vehreschild, Institute for Scientific Computing,
% RWTH Aachen University, 52056 Aachen, Germany.
% Visit us on the web at http://www.adimat.de
... |
github | microsoft/ADBench-master | a_polyval2.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_polyval2.m | 753 | utf_8 | 2cf97e5ddc25c22869b24bf368997090 | % function [a_x z] = a_polyval2(a_z, p, x)
%
% Compute adjoint of x in z = polyval(p, x), polynomial p(x), given
% the adjoint of z.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2012 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
function [a_x z] = a_p... |
github | microsoft/ADBench-master | dpartial_asinh.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/dpartial_asinh.m | 470 | utf_8 | d8337e1b05e5e5d8f30cc0ba48e32588 | % function [dpartial y] = dpartial_asinh(x)
%
% Compute partial derivative diagonal of y = asinh(x).
%
% see also dpartial_exp.
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Fachgebiet Scientific Computing
% TU Darmstadt
function [dpartial y] = dpartia... |
github | microsoft/ADBench-master | adimat_push_index.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_push_index.m | 614 | utf_8 | f9f5c699bfc5edf092a113c764db93b9 | % function r = adimat_push_index(arrayVar, index1, ...)
% Store values given by arrayVar(index1, ...) on the stack, if this
% indexing is possible. Otherwise the whole variables has to be
% saved, in order to restore it to previous size later.
%
% see also adimat_push, adimat_push_cell_index, adimat_pop_index
%
%... |
github | microsoft/ADBench-master | a_eigs_01_11.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_eigs_01_11.m | 415 | utf_8 | 1469ecdae1fcd243ae8e9e2d09510f6e | % function a_A = a_eigs_01_31(a_U, a_D, A, k?, sigma?)
%
% Compute adjoint of A in [V D] = eig(A), given adjoint of D.
%
% This file is part of the ADiMat runtime environment
%
% Copyright (C) 2015 Johannes Willkomm
function a_A = a_eigs_01_31(a_D, A, varargin)
[V D U] = adimat_lreigs(A, varargin{:});
a_A = V.' * c... |
github | microsoft/ADBench-master | adimat_stack_byte_size.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_stack_byte_size.m | 644 | utf_8 | f02cc20cc9e9afb78023d564f947d691 | % function r = adimat_stack_byte_size()
% return the size of the stack in bytes, if that measure is
% available, otherwise 0
%
% see also adimat_store, adimat_stack, adimat_clear_stack,
% adimat_size, adimat_file_size
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2010-2012 Johannes Willkomm, I... |
github | microsoft/ADBench-master | adimat_prepad.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_prepad.m | 274 | utf_8 | 00733e7d4c6d5aca92afe9062513df26 | % function r = adimat_prepad(a,l,c)
%
% Mimic the function prepad, which is available in Octave, but not in MATLAB
%
% (c) 2018 Johannes Willkomm
function r = adimat_prepad(a,l,c,dim)
psz = size(a);
psz(dim) = l - psz(dim);
pa = repmat(c, psz);
r = cat(dim, pa, a);
|
github | microsoft/ADBench-master | adimat_setup_stack.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_setup_stack.m | 1,917 | utf_8 | 548ea4b87c834a0cd811c0adb45fc471 | % function r = adimat_setup_stack(stackOptions)
%
% setup runtime environment for using stack given by
% stackOptions.stackName.
%
% see also adimat_stack, adimat_info_stack, adimat_stack_verbosity,
% adimat_stack_buffer_size, adimat_stack_num_buffers,
% adimat_stack_prefetch, adimat_stack_async_io_type,
% adimat_stack... |
github | microsoft/ADBench-master | partial_interp1_2.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/partial_interp1_2.m | 851 | utf_8 | 61a0a6c79599d63c6986efcf0a7107ed | % function [partial z] = partial_interp1_2(xs, ys, xis, varargin)
% compute partial interp1(xs, ys, xis, ...) w.r.t. to ys
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2013 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
function [partial z] = partial_... |
github | microsoft/ADBench-master | adimat_g_mldivide.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_g_mldivide.m | 794 | utf_8 | efdbd9d659d2c7383228041037e6e2b1 | % function [g_z z] = adimat_g_mldivide(g_a, a, g_b, b)
%
% Compute derivative of z = a \ b, matrix left division. Also return
% the function result z.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2013,2014 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt... |
github | microsoft/ADBench-master | a_eig_1_11.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_eig_1_11.m | 593 | utf_8 | 71b2c18a7903e1aaef2828e31892ecd9 | % function a_A = a_eig_1_11(a_l, A)
%
% Compute adjoint of A in l = eig(A), given adjoint of l.
%
% see also a_eig_10_11, a_eig_01_11, a_eig_11_11
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
function ... |
github | microsoft/ADBench-master | a_eig_10_11.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_eig_10_11.m | 631 | utf_8 | 4e2b784c3ace7e01745cc36efc60f990 | % function a_A = a_eig_10_11(a_V, x)
%
% Compute adjoint of A in [V D] = eig(A), given adjoint of V.
%
% see also a_eig_01_11, a_eig_11_11, a_eig_1_11
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
funct... |
github | microsoft/ADBench-master | adimat_g_pow.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_g_pow.m | 1,839 | utf_8 | a328f6529eda20b93970fc6010ec27ef | % function [g_z z] = adimat_g_pow(g_a, a, g_b, b)
%
% Compute derivative of z = a^b, matrix exponentiation. Also return
% the function result z.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2010-2012 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
% Cop... |
github | microsoft/ADBench-master | a_zeros.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_zeros.m | 799 | utf_8 | d85c57f96dc08d90b550d16a2f2dddc8 | % function varargout = a_zeros(varargin)
%
% Create zero adjoints. This function creates zero derivatives of
% class double, by calling the function zeros. These adjoints are used
% in scalar reverse mode.
%
% see also zeros, adimat_adjoint
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2011 Joha... |
github | microsoft/ADBench-master | d_adimat_mpower.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/d_adimat_mpower.m | 3,823 | utf_8 | b513251f25454bb023ec7c3f8428b26f | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | adimat_a_mpower.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_a_mpower.m | 1,876 | utf_8 | 20a23b00dfecea58fcd5eb53f0123c92 | % function [a_a, a_b, z] = adimat_a_mpower(a, b, a_z)
%
% Compute adjoints a_a and a_b of z = a^b, matrix exponentiation. Also
% return the function result z.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2010-2012,2014 Johannes Willkomm
%
function [a_a, a_b, z] = adimat_a_mpower(a, b, a_z)
z... |
github | microsoft/ADBench-master | adimat_stack_dir_name.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_stack_dir_name.m | 1,081 | utf_8 | 922463e50f600d888236c41171c4c31b | % function r = adimat_stack_dir_name(name?)
%
% get/set directory name for the stack data file. This functionality
% is implemented by setting the environment variable
% ADIMAT_STACK_DIR. This setting applies to all stack implementations
% with "file" in their name.
%
% adimat_stack_dir_name(level)
% - this sets the ... |
github | microsoft/ADBench-master | a_svd_110.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_svd_110.m | 2,263 | utf_8 | c5e6e0be313ce640c95ed7b06278cff2 | % function [a_A U S V] = a_svd_110(a_U, a_S, A)
%
% Compute adjoint of A in [ U S V ] = svd(A), where a_U and a_S are
% the adjoints of U and S, resp.
%
% Using the method from Mike Giles, "An extended collection of matrix
% derivative results for forward and reverse mode algorithmic
% differentiation", 2008, adding in... |
github | microsoft/ADBench-master | admGetFunc.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admGetFunc.m | 487 | utf_8 | 6969c808116097ca890108acb9d7ec5b | % function func = admGetFunc(nameOrHandle)
%
% Copyright 2011 Johannes Willkomm, Institute for Scientific Computing
% RWTH Aachen University
%
% This file is part of the ADiMat runtime environment.
%
function func = admGetFunc(nameOrHandle)
if isa(nameOrHandle, 'char')
func = str2func(nameO... |
github | microsoft/ADBench-master | a_adimat_chol.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_adimat_chol.m | 10,062 | utf_8 | 524be90d0c21fd746daf5034fb01ae4c | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | g_mk_householder_elim.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/g_mk_householder_elim.m | 2,441 | utf_8 | 29d703c0021062eb9ab4ff9018c260e3 | % Generated by ADiMat 0.6.0-4867
% Copyright 2009-2013 Johannes Willkomm, Fachgebiet Scientific Computing,
% TU Darmstadt, 64289 Darmstadt, Germany
% Copyright 2001-2008 Andre Vehreschild, Institute for Scientific Computing,
% RWTH Aachen University, 52056 Aachen, Germany.
% Visit us on the web at http://www.adimat.de
... |
github | microsoft/ADBench-master | a_svd_101.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_svd_101.m | 2,212 | utf_8 | a01655d65747416d210233d1e74e87e6 | % function [a_A U S V] = a_svd_111(a_U, a_S, a_V, A)
%
% Compute adjoint of A in [ U S V ] = svd(A), where a_U and a_V are
% the adjoints of U and V, resp.
%
% Using the method from Mike Giles, "An extended collection of matrix
% derivative results for forward and reverse mode algorithmic
% differentiation", 2008, addi... |
github | microsoft/ADBench-master | a_adimat_hess.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_adimat_hess.m | 21,387 | utf_8 | 90604a49b211970593f3e4bbb167be5e | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | partial_interp2.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/partial_interp2.m | 1,480 | utf_8 | 09b9f493377df99703916a93376b05b9 | % function [partial z] = partial_interp2(xs, ys, vs, xis, yis, ...)
% compute partial interp2(xs, ys, vs, xis, yis, ...) w.r.t. to vs
%
% See also:
% http://en.wikipedia.org/wiki/Bilinear_interpolation
% http://en.wikipedia.org/wiki/Bicubic_interpolation
%
% This file is part of the ADiMat runtime environment
%
% C... |
github | microsoft/ADBench-master | adimat_atan2.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_atan2.m | 560 | utf_8 | 6a3c00c9ec9cfe16c4c3e791390b3921 | % function t = adimat_atan2(y, x)
%
% ADiMat replacement function for atan2
%
% Copyright (C) 2015 Johannes Willkomm
function t = adimat_atan2(y, x)
xl0 = x < 0;
yl0 = y < 0;
xe0 = x == 0;
ye0 = y == 0;
yg0 = ~yl0 & ~ye0;
t = x + y;
t(~xe0) = atan(y(~xe0) ./ x(~xe0));
t(xe0 & yl0) = - atan(x(xe0 & yl0) ... |
github | microsoft/ADBench-master | admColorSeed.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admColorSeed.m | 1,226 | utf_8 | 7b2cf001e3b8be3e69e17b82a784c779 | % function [seed coloring] = admColorSeed(pattern, opts)
%
% Copyright 2011 Johannes Willkomm, Institute for Scientific Computing
% RWTH Aachen University
%
% This file is part of the ADiMat runtime environment.
%
function [seed coloring] = admColorSeed(pattern, opts)
if nargin < 2
opts = a... |
github | microsoft/ADBench-master | g_adimat_svd.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/g_adimat_svd.m | 7,586 | utf_8 | a95742bddb379eb830daf3575f045dda | % Generated by ADiMat 0.6.0-4867
% Copyright 2009-2013 Johannes Willkomm, Fachgebiet Scientific Computing,
% TU Darmstadt, 64289 Darmstadt, Germany
% Copyright 2001-2008 Andre Vehreschild, Institute for Scientific Computing,
% RWTH Aachen University, 52056 Aachen, Germany.
% Visit us on the web at http://www.adimat.de
... |
github | microsoft/ADBench-master | dpartial_cot.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/dpartial_cot.m | 451 | utf_8 | f803e2af968e09e41762d0f1c1775b22 | % function [dpartial y] = dpartial_cot(x)
%
% Compute partial derivative diagonal of y = cot(x).
%
% see also dpartial_exp.
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Fachgebiet Scientific Computing
% TU Darmstadt
function [dpartial y] = dpartial_co... |
github | microsoft/ADBench-master | g_bessely.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/g_bessely.m | 612 | utf_8 | 18d066da2018f0ab36009e5b1b97f758 | % function [z, g_z] = g_besselj(nu, g_x, x, opt?)
%
% compute the derivative of z in z = bessely(nu, x, opt?), where g_x
% is the derivative of x.
%
% see also dpartial_bessely
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2013 Johannes Willkomm, Fachgebiet Scientific Computing
% ... |
github | microsoft/ADBench-master | partial_cumtrapz_X.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/partial_cumtrapz_X.m | 1,422 | utf_8 | 613d65e6c1dc51a1e4de80ca6f7ab45c | % function z = partial_cumtrapz_X(arg...)
%
% Compute partial derivative of cumtrapz function w.r.t. X.
%
% For calling seequence see cumtrapz.
%
% see also cumtrapz
%
% Copyright (C) 2015 Johannes Willkomm
function J = partial_cumtrapz_X(a, b, c)
if nargin == 1
% cumtrapz(Y)
J = [];
elseif nargin == 2
... |
github | microsoft/ADBench-master | a_hypot2.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_hypot2.m | 462 | utf_8 | 513d024bc525a262856daa5188b9f0c4 | % function [a_y] = a_hypot(a_z, x, y)
%
% Compute adjoint of y in z = hypot(x, y), given the adjoint of z.
%
% see also a_zeros, a_mean
%
% Copyright 2012 Johannes Willkomm, Institute for Scientific Computing
% TU Darmstadt
function [a_y z] = a_hypot(a_z, x, y)
% t = x.^2 + y.^2
% z = sqrt(t... |
github | microsoft/ADBench-master | adimat_h_mrdivide.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_h_mrdivide.m | 587 | utf_8 | f59b42b79ce60aa5cb46fc7b7a3d9aaf | % function [h_z g_z z] = adimat_h_mrdivide(h_g_a, g_a, h_a, a, h_g_b, g_b, h_b, b)
%
% Compute first and second order derivative of z = a / b, matrix right
% division. Also return the function result z.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright (C) 2013,2014 Johannes Willkomm
%
function [h_... |
github | microsoft/ADBench-master | adimat_a_mldividel.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_a_mldividel.m | 1,251 | utf_8 | cc3d5a3084e69c36e5b7f83c49a9be3c | % function [a_a, z] = adimat_a_mldividel(a, b, a_z)
%
% Compute adjoint a_a of z = a \ b, matrix right division. Also return
% the function result z.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2012-2014 Johannes Willkomm
%
function [a_a, z] = adimat_a_mldividel(a, b, a_z)
[m n] = size(a);
... |
github | microsoft/ADBench-master | g_adimat_sol_tril.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/g_adimat_sol_tril.m | 3,641 | utf_8 | 5b0d5350097da9988e9e60cac8eb9241 | % Generated by ADiMat 0.6.0-4788
% Copyright 2009-2013 Johannes Willkomm, Fachgebiet Scientific Computing,
% TU Darmstadt, 64289 Darmstadt, Germany
% Copyright 2001-2008 Andre Vehreschild, Institute for Scientific Computing,
% RWTH Aachen University, 52056 Aachen, Germany.
% Visit us on the web at http://www.adimat.de
... |
github | microsoft/ADBench-master | admTotalNumel.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admTotalNumel.m | 890 | utf_8 | 91c899523d8cd1541e258b2d4f063f7f | % function nels = admTotalNumel(varargin)
%
% Computes sum of prod(size(.)) of each argument. It does not use
% numel, because this has a different meaning for derivative classes.
%
% Copyright 2014 Johannes Willkomm
% Copyright 2010,2011 Johannes Willkomm, Institute for Scientific Computing
% RWTH ... |
github | microsoft/ADBench-master | dpartial_asecd.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/dpartial_asecd.m | 480 | utf_8 | 7a40e553ba8a1f94f993146225336614 | % function [dpartial y] = dpartial_asecd(x)
%
% Compute partial derivative diagonal of y = asecd(x).
%
% see also dpartial_exp.
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Fachgebiet Scientific Computing
% TU Darmstadt
function [dpartial y] = dpartia... |
github | microsoft/ADBench-master | a_cat.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_cat.m | 770 | utf_8 | 4bd31205e99ae2931d4770a9978789f6 | % function adj = a_cat(adj, v1, v2, ..., vi)
%
% compute adjoint of cat(v1, v2, ..., vi, ..., vn) w.r.t. vi.
%
% see also a_vertcat, a_horzcat, a_sum
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2013 Johannes Willkomm, Institute for Scientific Computing
% RWTH Aachen Univers... |
github | microsoft/ADBench-master | g_adimat_onesided_jacobi.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/g_adimat_onesided_jacobi.m | 11,022 | utf_8 | fe901caa99a62f787fa72a54379b0315 | % Generated by ADiMat 0.6.0-4867
% Copyright 2009-2013 Johannes Willkomm, Fachgebiet Scientific Computing,
% TU Darmstadt, 64289 Darmstadt, Germany
% Copyright 2001-2008 Andre Vehreschild, Institute for Scientific Computing,
% RWTH Aachen University, 52056 Aachen, Germany.
% Visit us on the web at http://www.adimat.de
... |
github | microsoft/ADBench-master | a_adimat_sol_ata.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_adimat_sol_ata.m | 4,646 | utf_8 | f09e52ae4ee8e62349537ddd88680838 | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | a_struct.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_struct.m | 859 | utf_8 | 2ff67fa5e8d8a00f24905480a558452c | % function r = a_zeros_index(adjArrayVar, arrayVar, index1, ...)
% Zero adjoint of index expression after assignment. If the forward
% assignment changed the size of the array variable, then maybe
% resize the adjoint here, undoing the size change. If the size did
% not change then fill the indexed adjoint wit... |
github | microsoft/ADBench-master | adimat_d_svd.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_d_svd.m | 3,615 | utf_8 | 6b53cd7178258cf13432d5a2d3f44241 | % function [g_U U d_S S d_V V] = d_svd(d_A, A)
%
% Compute derivatives of U, S, V in [ U S V ] = svd(A).
%
% Using the method from Mike Giles, "An extended collection of matrix
% derivative results for forward and reverse mode algorithmic
% differentiation", 2008
%
% Copyright 2012,2013 Johannes Willkomm, Institute for... |
github | microsoft/ADBench-master | adimat_a_mldivide.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_a_mldivide.m | 1,355 | utf_8 | faf0940574613713eecf281079dae1c2 | % function [a_a, a_b, z] = adimat_a_mldivide(a, b, a_z)
%
% Compute adjoints a_a and a_b of z = a \ b, matrix left
% division. Also return the function result z.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2012-2014 Johannes Willkomm
%
function [a_a, a_b, z] = adimat_a_mldivide(a, b, a_z)
[... |
github | microsoft/ADBench-master | adimat_stack.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_stack.m | 5,344 | utf_8 | 74f64c90747d4a1bfe3d1dea3bfe2df3 | % function r = adimat_stack(name)
% get/set name of ADiMat's stack implementation.
% This functionality is implemented by modifying the path() such
% that different versions of the function adimat_store are used.
%
% adimat_stack('avail')
% adimat_stack('list')
% adimat_stack('?')
% - return list of available s... |
github | microsoft/ADBench-master | admDiffFor.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admDiffFor.m | 7,657 | utf_8 | 5af31eb9a240b71301f8091b11d7215e | % function [Jacobian, varargout] = admDiffFor(handle, seedMatrix, arg1, arg2, ..., argN, admOptions?)
%
% Compute first order derivatives of function given by handle,
% evaluated at the arguments arg1, arg2, ..., argN. This works by
% running a version of the function differentiated in forward mode
% (FM) of automa... |
github | microsoft/ADBench-master | adimat_g_cumtrapz.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_g_cumtrapz.m | 821 | utf_8 | ed34080d508c384f6fa0f40b157b5206 | % function [g_z z] = adimat_g_cumtrapz(g_a, a, g_b, b, c)
%
% Copyright (C) 2015 Johannes Willkomm
function [g_z z] = adimat_g_cumtrapz(g_a, a, g_b, b, c)
if nargin == 2
% cumtrapz(Y)
z = cumtrapz(a);
J_Y = partial_cumtrapz_Y(a);
g_z = J_Y * g_a(:);
elseif nargin == 4
z = cumtrapz(a, b);
J_Y... |
github | microsoft/ADBench-master | adimat_postpad.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_postpad.m | 275 | utf_8 | 75faf25e0c3465d7fba125ae8c700a7c | % function r = adimat_prepad(a,l,c)
%
% Mimic the function prepad, which is available in Octave, but not in MATLAB
%
% (c) 2018 Johannes Willkomm
function r = adimat_postpad(a,l,c,dim)
psz = size(a);
psz(dim) = l - psz(dim);
pa = repmat(c, psz);
r = cat(dim, a, pa);
|
github | microsoft/ADBench-master | dpartial_asech.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/dpartial_asech.m | 500 | utf_8 | ea33604a67aaa44ab792b59529f366a8 | % function [dpartial y] = dpartial_asech(x)
%
% Compute partial derivative diagonal of y = asech(x).
%
% see also dpartial_exp.
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Fachgebiet Scientific Computing
% TU Darmstadt
function [dpartial y] = dpartia... |
github | microsoft/ADBench-master | createSeededGradientsComplex.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/createSeededGradientsComplex.m | 2,086 | utf_8 | 77bb8519537ea7ad3fd856ca998f2122 | % function varargout = createSeededGradientsComplex(seedVector, varargin)
%
% see also createSeededGradientsFor
%
% This file is part of the ADiMat runtime environment, and belongs
% to the scalar_directderivs derivative "class".
%
% Copyright 2010,2011 Johannes Willkomm, Institute for Scientific Computing
% ... |
github | microsoft/ADBench-master | a_ipermute.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_ipermute.m | 228 | utf_8 | 3f06e81e7c3e15508ea6420ea50a9bc5 | % function adj = a_ipermute(adj, ind, a, b)
%
% Copyright (C) 2015 Johannes Willkomm
function adj = a_permute(adj, ind, a, b)
switch ind
case 1
adj = permute(adj, b);
otherwise
adj = a_zeros(0);
end
end
% $Id$
|
github | microsoft/ADBench-master | a_adimat_cumtrapz_i1.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_adimat_cumtrapz_i1.m | 24,599 | utf_8 | 7b3d7a7c16ac6f6a8eac3c4c99256dbf | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | dpartial_tan.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/dpartial_tan.m | 450 | utf_8 | bd4f89b339420a01444772ab065afc30 | % function [dpartial y] = dpartial_tan(x)
%
% Compute partial derivative diagonal of y = tan(x).
%
% see also dpartial_exp.
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Fachgebiet Scientific Computing
% TU Darmstadt
function [dpartial y] = dpartial_ta... |
github | microsoft/ADBench-master | adimat_sol_ata.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_sol_ata.m | 340 | utf_8 | d822612b7916817015375ba18200d12a | % function z = adimat_sol_qr(a, b)
%
% Solve linear system a*x = b for x by solving (a'*a)*x=a'*b which is
% always a square system, for automatic differentiation.
%
% Copyright 2013,2014 Johannes Willkomm
%
function z = adimat_sol_ata(a, b)
L = a';
z = linsolve(L*a, L*b);
% $Id: adimat_sol_ata.m 4813 2014-10-09 0... |
github | microsoft/ADBench-master | a_ifft.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_ifft.m | 2,362 | utf_8 | 69af30845f5709bb684dadfdc19116fd | % function adj = a_ifft(adj, ind, a, varargin)
%
% Compute adjoint of z = ifft(x, n?, dim?) and also of the forms z =
% ifft(..., method), were method is a string.
%
% see also a_fft
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2013 Johannes Willkomm, Institute for Scientific Computing
% ... |
github | microsoft/ADBench-master | admStackSize.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admStackSize.m | 1,227 | utf_8 | b67b1adac845e495a492324dbe4ee4d0 | % function [Jacobian, varargout] = admDiffRev(handle, seedMatrix, arg1, arg2, ..., argN, admOptions?)
%
% see also admDiffRev.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2012 Johannes Willkomm, Institute for Scientific Computing
% RWTH Aachen University
function [stack_by... |
github | microsoft/ADBench-master | admMatMultV1TTa.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admMatMultV1TTa.m | 451 | utf_8 | d50cd4c69ce3183054e40e01ebc599ad | % function Z = admMatMultV1TTa(A, B)
%
% see also admMatMultV1
%
% Copyright (C) 2014 Johannes Willkomm
%
function Z = admMatMultV1TTa(A, B)
% A is (k x m x n) tensor
% B is (k x n x p) tensor
szA = size(A);
szB = size(B);
Z = zeros(szA(1), szA(2), szB(3));
for k=1:szA(2)
T = bsxfun(@times, reshap... |
github | microsoft/ADBench-master | g_polyvalm.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/g_polyvalm.m | 661 | utf_8 | c88a1f0f23415912d03656b91f9a18e8 | % function [g_z z] = g_polyvalm(g_p, p, g_x, x)
%
% Compute derivative of z in z = polyval(p, x), matrix polynomial
% p(x), given the derivatives of p and x.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2012 Johannes Willkomm, Institute for Scientific Computing
% TU Darmsta... |
github | microsoft/ADBench-master | shiftedbsx.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/shiftedbsx.m | 431 | utf_8 | cc951c3cd826f48b1d08262872beb48e | % function res = shiftedbsx(v1, v2)
%
% Copyright (C) 2014 Johannes Willkomm
%
function v1 = shiftedbsx(handle, v1, v2)
n = 1;
sz1 = size(v1);
sz2 = size(v2);
while n <= length(sz1) && n <= length(sz2) && sz1(n) == 1 && sz2(n) == 1
n = n + 1;
end
n = n - 1;
if n > 0
v1 = shiftdim(v1, n);
v2 = ... |
github | microsoft/ADBench-master | g_mk_givens.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/g_mk_givens.m | 2,415 | utf_8 | 48973b891d93dcea2e50b8cda1482965 | % Generated by ADiMat 0.6.0-4867
% Copyright 2009-2013 Johannes Willkomm, Fachgebiet Scientific Computing,
% TU Darmstadt, 64289 Darmstadt, Germany
% Copyright 2001-2008 Andre Vehreschild, Institute for Scientific Computing,
% RWTH Aachen University, 52056 Aachen, Germany.
% Visit us on the web at http://www.adimat.de
... |
github | microsoft/ADBench-master | d_adimat_sol_qr.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/d_adimat_sol_qr.m | 4,130 | utf_8 | 6e18fbb3820785645c8480be4595f67f | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | d_adimat_pinv2.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/d_adimat_pinv2.m | 18,070 | utf_8 | 91cc6992f65d6daa49ccb936f3ea00ac | % Generated by ADiMat 0.6.0-4867
% © 2001-2008 Andre Vehreschild <vehreschild@sc.rwth-aachen.de>
% © 2009-2013 Johannes Willkomm <johannes.willkomm@sc.tu-darmstadt.de>
% RWTH Aachen University, 52056 Aachen, Germany
% TU Darmstadt, 64289 Darmstadt, Germany
% Visit us on the web at http://www.adimat.de/
% Report bugs to... |
github | microsoft/ADBench-master | adimat_h_mrdivide1.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_h_mrdivide1.m | 566 | utf_8 | e2975c75b068859d1b816099a0a0777a | % function [h_z g_z z] = adimat_h_mrdivide1(h_a, g_a, a, h_b, b)
%
% Compute first and second order derivative of z = a / b, matrix right
% division, with b constant. Also return the function result z.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright (C) 2013,2014 Johannes Willkomm
%
function [h_g... |
github | microsoft/ADBench-master | adimat_stack_buffer_size.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_stack_buffer_size.m | 1,073 | utf_8 | 4ac75510d07aed9f20bb1d780bb1beb8 | % function r = adimat_stack_buffer_size(sz?)
%
% get/set the size of the memory buffer used by the stack. This
% functionality is implemented by setting the environment variable
% ADIMAT_STACK_BUFFER_SIZE. This setting applies to all stack
% implementations with "file" in their name.
%
% adimat_stack_buffer_size(sz)
% ... |
github | microsoft/ADBench-master | adimat_neigs.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_neigs.m | 387 | utf_8 | 89a1506cc77706c2afb8223d465ee14b | % function [V D] = adimat_neigs(A, k?, sigma?, opts?)
%
% Return eigenvectors from eigs in a normalized form: first component
% is always positive.
%
% Copyright (C) 2015 Johannes Willkomm
function [V D] = adimat_neigs(A, varargin)
[V D] = eigs(A, varargin{:});
for k=1:size(V,2)
V(:,k) = sign(V(1,k)) .* V(:,k)... |
github | microsoft/ADBench-master | admDiffDD.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admDiffDD.m | 2,661 | utf_8 | cdd0992fa100118402f7b402bbe7f80b | % function [Jacobian, varargout] = admDiffDD(handle, seedMatrix, arg1, arg2, ..., argN, admOptions?)
%
% This function is deprecated and replaced by admDiffFD.
%
% see also admDiffFD, admDiffFor, admDiffRev.
%
% This file is part of the ADiMat runtime environment.
%
% Copyright 2010,2011 Johannes Willkomm, Institute fo... |
github | microsoft/ADBench-master | adimat_d_trapz.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/adimat_d_trapz.m | 898 | utf_8 | f12f966dddf93e8d6d89d14be88d2b27 | % function [d_z z] = adimat_d_trapz(d_a, a, d_b, b, c)
%
% Copyright (C) 2015 Johannes Willkomm
function [d_z z] = adimat_d_trapz(d_a, a, d_b, b, c)
if nargin == 2
% trapz(Y)
z = trapz(a);
J_Y = partial_trapz_Y(a);
d_z = admMatMultV1MTb(J_Y, d_a(:,:));
elseif nargin == 4
z = trapz(a, b);
J_Y... |
github | microsoft/ADBench-master | a_eigs_10_21.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/a_eigs_10_21.m | 672 | utf_8 | 3efb5d7b21189fd0c8e3ba6c04df6521 | % function a_A = a_eigs_10_31(a_V, A, k?, sigma?)
%
% Compute adjoint of A in [V D] = eig(A), given adjoint of V.
%
% This file is part of the ADiMat runtime environment
%
% Copyright (C) 2015 Johannes Willkomm
function a_A = a_eigs_10_31(a_U, A, varargin)
[V D U] = adimat_lreigs(A, varargin{:});
k = size(U,2);
I... |
github | microsoft/ADBench-master | dpartial_acoth.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/dpartial_acoth.m | 465 | utf_8 | 99301db031904b782aff07f294171a35 | % function [dpartial y] = dpartial_acoth(x)
%
% Compute partial derivative diagonal of y = acoth(x).
%
% see also dpartial_exp.
%
% This file is part of the ADiMat runtime environment
%
% Copyright 2012 Johannes Willkomm, Fachgebiet Scientific Computing
% TU Darmstadt
function [dpartial y] = dpartia... |
github | microsoft/ADBench-master | admCheckNArgsRev.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admCheckNArgsRev.m | 2,620 | utf_8 | 27520c9e9945107cc7859f74690af31e | % function r = admCheckNArgsRev(functionName, resultFunctionName, ...
% nFuncArgs, nActArgs, fNargout,
% nActResults, nFunctionHandles)
%
% Check whether the amount of function and derivative arguments is OK
% for differentiated function resultFunctionName.
%
% ... |
github | microsoft/ADBench-master | admHessianT1Rev.m | .m | ADBench-master/submodules/adimat-0.6.2-5288/runtime/admHessianT1Rev.m | 10,279 | utf_8 | 2a2a902dc974c23af2787c58e13ece53 | % function [Hessian, Jac, varargout] = ...
% admHessianT1Rev(handle, seedMatrix, arg1, arg2, ..., argN, admOptions?)
%
% Compute second order derivatives of function given by handle,
% evaluated at the arguments arg1, arg2, ..., argN and return the
% Hessian matrix, or the Hessian matrix times seedMatrix. This uses... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.