Code

Add einsum, closes #124#363

Code

Merged

mratsim merged 36 commits intomratsim:mastermratsim/Arraymancer:masterfrom Vindaar:addEinsumVindaar/Arraymancer:addEinsumCopy head branch name to clipboardJun 30, 2019

+874-3

Commits

Commits on Jun 15, 2019

• WIP commit of einsum implementation

  Show description for 1fb6b18
  Vindaar

  committed
  1fb6b18

  Copy full SHA for 1fb6b18

• allow contraction of a single axis

  Vindaar

  committed
  35ed0a4

  Copy full SHA for 35ed0a4

• add explicit and implicit matrix vector multiplication

  Vindaar

  committed
  8a1238d

  Copy full SHA for 8a1238d

Commits on Jun 16, 2019

• WIP commit towards a better approach for mapping for idx idents to shapes

  Vindaar

  committed
  a6aee65

  Copy full SHA for a6aee65

Commits on Jun 17, 2019

• fix transposition by assigning indices in order of appearing in stmt

  Vindaar

  committed
  580accf

  Copy full SHA for 580accf

• WIP commit making all tests work

  Show description for b98e526
  Vindaar

  committed
  b98e526

  Copy full SHA for b98e526

• remove a bunch of dead code / debug echoes

  Vindaar

  committed
  583f933

  Copy full SHA for 583f933

• add reference to where tests come from

  Vindaar

  committed
  48b48be

  Copy full SHA for 48b48be

• only import tensor

  Vindaar

  committed
  9f778dd

  Copy full SHA for 9f778dd

• move einsum tests to tensor subdirectory

  Vindaar

  committed
  336eb01

  Copy full SHA for 336eb01

Commits on Jun 23, 2019

• add einsum test to test suite

  Vindaar

  committed
  4fb5202

  Copy full SHA for 4fb5202

• fix import path of arraymancer in einsum test

  Vindaar

  committed
  cbb4ab0

  Copy full SHA for cbb4ab0

• refactor out statement split

  Vindaar

  committed
  a928293

  Copy full SHA for a928293

• add a note about getTensors

  Vindaar

  committed
  18cf9b2

  Copy full SHA for 18cf9b2

• add check about number of einsum statement

  Vindaar

  committed
  694ca64

  Copy full SHA for 694ca64

• add a reference test file for tests designed to fail

  Vindaar

  committed
  b566d06

  Copy full SHA for b566d06

• add TensorIdx object to combine tensor and its indices

  Show description for 4f122dd
  Vindaar

  committed
  4f122dd

  Copy full SHA for 4f122dd

• remove unnecessary indexing variable in axes iterations

  Vindaar

  committed
  e15c7a1

  Copy full SHA for e15c7a1

• add enumerateIdx iterator to yield indices of seq[TensorIdx]

  Vindaar

  committed
  3297ad8

  Copy full SHA for 3297ad8

• only generate shapesContr variable, if contracting at least 1 axis

  Vindaar

  committed
  51681c2

  Copy full SHA for 51681c2

• refactor code further, simplify index / result shape mapping

  Vindaar

  committed
  dfac63c

  Copy full SHA for dfac63c

• use LHS explicit assignment as let stmt

  Show description for e491fd7
  Vindaar

  committed
  e491fd7

  Copy full SHA for e491fd7

• remove now useless findAxes

  Vindaar

  committed
  d48e3be

  Copy full SHA for d48e3be

• remove dead line in failed test file

  Vindaar

  committed
  69b660a

  Copy full SHA for 69b660a

Commits on Jun 26, 2019

• perform a type check of all tensors, use that type as result type

  Show description for 33bd3ea
  Vindaar

  committed
  33bd3ea

  Copy full SHA for 33bd3ea

• revert returning a let section for explicit einsum

  Show description for b472e11
  Vindaar

  committed
  b472e11

  Copy full SHA for b472e11

• create local contiguous, row major tensors for more eff iteration

  Show description for f4ad9a2
  Vindaar

  committed
  f4ad9a2

  Copy full SHA for f4ad9a2

• add example to docstring

  Vindaar

  committed
  966f128

  Copy full SHA for 966f128

• add some documentation to the top of the file

  Vindaar

  committed
  ac95c1c

  Copy full SHA for ac95c1c

• fix two links in the documentation

  Vindaar

  committed
  366f62c

  Copy full SHA for 366f62c

• reverse the order of idxIdentPairs for correct loop order

  Show description for 3153677
  Vindaar

  committed
  3153677

  Copy full SHA for 3153677

Commits on Jun 27, 2019

• use genSym for unique symbols in the macro block

  Show description for 57d20ad
  Vindaar

  committed
  57d20ad

  Copy full SHA for 57d20ad

• move type declarations to top of file, add short comment

  Vindaar

  committed
  5fb96f6

  Copy full SHA for 5fb96f6

• clean up sub type gen, use new type for contiguous tensors

  Vindaar

  committed
  1332276

  Copy full SHA for 1332276

• take into account tensor type for scalar result

  Vindaar

  committed
  a269b7c

  Copy full SHA for a269b7c

• comment about scalar result type, arbitrariness of LHS for explicit

  Vindaar

  committed
  08d2771

  Copy full SHA for 08d2771