tebdol, branch HEAD

Update example outputs

2017-06-29T02:37:24Z

commit 6a027b9ea8b27be756ce2e2503bb0733ce687812 parent 59c8635f223f84a79ba25caa252b984a09350e20 Author: Miroslav Urbanek Date: Thu, 29 Jun 2017 04:37:24 +0200 Update example outputs Parameter maxdim is a list now.

Update README

2017-06-29T02:00:48Z

commit 59c8635f223f84a79ba25caa252b984a09350e20 parent 75590722e531f14a12991479a938916676926fe5 Author: Miroslav Urbanek Date: Thu, 29 Jun 2017 04:00:48 +0200 Update README Explain that the library has to be compiled first by running SBCL as a single process.

Remove tensor-normalize from inputs/revivals.lisp

2017-06-29T01:59:39Z

commit 75590722e531f14a12991479a938916676926fe5 parent 71847f41e7784bdfd84c7624d61d3ac45d2fbdb4 Author: Miroslav Urbanek Date: Thu, 29 Jun 2017 03:59:39 +0200 Remove tensor-normalize from inputs/revivals.lisp The function has been recently added to tensor.lisp.

Update the MBL example

2017-06-29T01:45:36Z

commit 71847f41e7784bdfd84c7624d61d3ac45d2fbdb4 parent 459aa52d5e0aab5bc5074e5f343ba7af46bc9b53 Author: Miroslav Urbanek Date: Thu, 29 Jun 2017 03:45:36 +0200 Update the MBL example Clean up the example and generate a new sample output.

Add a simulation of boson expansion

2017-06-29T00:51:46Z

commit 459aa52d5e0aab5bc5074e5f343ba7af46bc9b53 parent cbb1537f28a49b20d17d09180288240049f72b39 Author: Miroslav Urbanek Date: Thu, 29 Jun 2017 02:51:46 +0200 Add a simulation of boson expansion This calculation is similar to the calculation of many-body localization. However, there is no random potential.

Do not mask floating-point errors

2017-06-28T23:56:46Z

commit cbb1537f28a49b20d17d09180288240049f72b39 parent 040c8fb8c22a1883384f31a4076f9b81e7da64af Author: Miroslav Urbanek Date: Thu, 29 Jun 2017 01:56:46 +0200 Do not mask floating-point errors It is better to inform the user that an error occurred.

Add a simulation of many-body localization

2017-06-28T23:49:35Z

commit 040c8fb8c22a1883384f31a4076f9b81e7da64af parent 20d98c1504d1fc68c9c2f2e501001f65f0bd80ee Author: Miroslav Urbanek Date: Thu, 29 Jun 2017 01:49:35 +0200 Add a simulation of many-body localization The system of 2 particles in a 4 x 4 lattice is too small to observe a crossover from thermalization to localization. The calculation serves as a simple example.

Rename directory "states" to "inputs"

2017-06-28T21:35:01Z

commit 20d98c1504d1fc68c9c2f2e501001f65f0bd80ee parent a425029cc2f8da8b54478f2b3539d8b4926114a3 Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 23:35:01 +0200 Rename directory "states" to "inputs" Some input files are not states.

Add support for two-dimensional models

2017-06-28T17:43:34Z

commit a425029cc2f8da8b54478f2b3539d8b4926114a3 parent 43e837ff846bd38b7766877c64515bc8fa8efd9e Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 19:43:34 +0200 Add support for two-dimensional models Two-dimensional states are represented as tree tensor network states (TTNS). The time-evolution algorithm rearranges the network to apply a two-site operator to each pair of adjacent sites.

Improve index truncation

2017-06-28T14:02:06Z

commit 43e837ff846bd38b7766877c64515bc8fa8efd9e parent 93185eb0dc2356dce5638a94a564361cca852158 Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 16:02:06 +0200 Improve index truncation Control truncation with a list (EPSILON MIN MAX). Discard the smallest singular values whose squared and normalized sum is less than or equal EPSILON while keeping at least MIN and at most MAX of them.

Remove unused code

2017-06-28T13:30:32Z

commit 93185eb0dc2356dce5638a94a564361cca852158 parent 0de51d6170e4379061104c85481decb7219f7f97 Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 15:30:32 +0200 Remove unused code Clean up tiny bits of useless code.

Clean up LAPACK functions

2017-06-28T13:17:48Z

commit 0de51d6170e4379061104c85481decb7219f7f97 parent ea1b5401f5594b2ccdd875c860fd3859adc7da16 Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 15:17:48 +0200 Clean up LAPACK functions Fix definitions of LAPACK functions and types of arrays passed to them.

Add alternative SVD functions

2017-06-28T13:02:07Z

commit ea1b5401f5594b2ccdd875c860fd3859adc7da16 parent e07a73fb30c01b3040c0e12095e5192fc70b0cbe Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 15:02:07 +0200 Add alternative SVD functions Allow to select among the functions ZGESVD, ZGESDD, and ZGESVDX to calculate a matrix SVD. A default action is to call ZGESDD, and fall back to ZGESVD if it fails.

Raise an error if ZGESDD does not converge

2017-06-28T11:29:23Z

commit e07a73fb30c01b3040c0e12095e5192fc70b0cbe parent 2eb5beb3c8b51d6c7b88f6d88f68e502c7884a89 Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 13:29:23 +0200 Raise an error if ZGESDD does not converge ZGESDD returns its status in the output parameter INFO. Check its value and raise an error if ZGESDD does not finish successfully.

Pin array storage vectors instead of arrays

2017-06-28T09:45:21Z

commit 2eb5beb3c8b51d6c7b88f6d88f68e502c7884a89 parent 1f7921ed6268a66b5ba9b71cc096f723ccfa2cdb Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 11:45:21 +0200 Pin array storage vectors instead of arrays SBCL was crashing with errors "GC invariant lost" and "no scavenge function for object". Crashes began with the SBCL commit 1439c9f07ad8 ("In GC cards that are pointed to conservatively, wipe out unused words."). The bug was in not pinning vectors returned by storage-array-vector. It was present from the very first version on and only manifested itself due to the commit above.

Update the ZGESDD arguments

2017-06-28T07:24:36Z

commit 1f7921ed6268a66b5ba9b71cc096f723ccfa2cdb parent 16d0947435afaab8ddaf9d66078e108e0d96e483 Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 09:24:36 +0200 Update the ZGESDD arguments Required dimension of RWORK has changed in the LAPACK 3.6.1 documentation. Also call make-double-array instead of make-array to simplify the code.

Permute indices in decompositions as requested

2017-06-28T06:44:25Z

commit 16d0947435afaab8ddaf9d66078e108e0d96e483 parent 0d404b0e2d3d10a2fc552048e797d4a34fd2e422 Author: Miroslav Urbanek Date: Wed, 28 Jun 2017 08:44:25 +0200 Permute indices in decompositions as requested Callers expect that array-decomposition and tensor-decomposition return the first tensor with a specified order of indices. If the decomposition indices are left indices, but they are not in the correct order, permute them.

Change the order of fusion indices in examples

2016-08-28T19:34:12Z

commit 0d404b0e2d3d10a2fc552048e797d4a34fd2e422 parent 556a0755b6616ca2c95b124f74f3c97100abf5f6 Author: Miroslav Urbanek Date: Sun, 28 Aug 2016 21:34:12 +0200 Change the order of fusion indices in examples This does not change the results, because for fusion of two indices, changing their order is equivalent to using :reverse type. However, it makes the code correct.

Add support for distributed propagators

2015-12-02T03:40:13Z

commit 556a0755b6616ca2c95b124f74f3c97100abf5f6 parent e314eb37d13058deffafbe97dd725026146f00eb Author: Miroslav Urbanek Date: Wed, 2 Dec 2015 04:40:13 +0100 Add support for distributed propagators A MPO can contain a NIL if the respective tensor belongs to a different node.

Add functions for saving distributed MPS

2015-12-02T03:26:19Z

commit e314eb37d13058deffafbe97dd725026146f00eb parent 621b566fb4c305d4ab8dcc1f214ea5e16ace0502 Author: Miroslav Urbanek Date: Wed, 2 Dec 2015 04:26:19 +0100 Add functions for saving distributed MPS

Rename mps:load-state to mps:load-mps

2015-10-08T09:18:28Z

commit 621b566fb4c305d4ab8dcc1f214ea5e16ace0502 parent d3780098760abe02f519b1fcd0e2d4808694f32e Author: Miroslav Urbanek Date: Thu, 8 Oct 2015 11:18:28 +0200 Rename mps:load-state to mps:load-mps

Support arguments in run script

2015-10-06T11:40:51Z

commit d3780098760abe02f519b1fcd0e2d4808694f32e parent afaa4e3295d6a8cdd774b47796c4e52b16ef7b50 Author: Miroslav Urbanek Date: Tue, 6 Oct 2015 13:40:51 +0200 Support arguments in run script

Add function make-uniform-propagator

2015-10-06T11:34:06Z

commit afaa4e3295d6a8cdd774b47796c4e52b16ef7b50 parent 389e75d9f9ec6df21ba0f711bba3468404f503e1 Author: Miroslav Urbanek Date: Tue, 6 Oct 2015 13:34:06 +0200 Add function make-uniform-propagator This function creates a propagator for a matrix product operator with the same Hamiltonian at each site.

Import the initial code

2015-06-27T22:25:05Z

commit 389e75d9f9ec6df21ba0f711bba3468404f503e1 Author: Miroslav Urbanek Date: Sun, 28 Jun 2015 00:25:05 +0200 Import the initial code