dgasmith/opt_einsum

Issues

• Thoughts on making the numpy dependency optional?

  #203 opened 2 months ago by janeyx99
  9

• [REQUEST] Release new version

  #238 opened 2 months ago by EarlMilktea
  3

• Sub-optimal contraction path when using broadcasting

  #220 opened a year ago by pimdh
  1

• When I use opt_einsum optimizes torch.einum, the running time after optimization increases.

  #202 opened 2 months ago by edwin-zft
  6

• New release with Python 3.12 support?

  #221 opened 2 months ago by singingwolfboy
  2

• Opt Einsum v3.4.0

  #191 opened 2 months ago by dgasmith
  3

• contract_path with `optimize=bool` raises

  #219 opened 2 months ago by ev-br
  3

• Document state of PyTorch upstreaming in the README

  #231 opened 2 months ago by bmillwood
  3

• opt_einsum.contract is slower than raw numpy.einsum

  #234 opened 2 months ago by lin-cp
  4

• Github identifier format in docs/source/conf.py

  #230 opened 4 months ago by NGC2023
  1

• `contract_path` with ellipsis fails when `shapes=True`

  #235 opened 5 months ago by nova77
  1

• Alternative Greedy Algorithms

  #233 opened 6 months ago by dgasmith
  1

• Multiple GPUs

  #215 opened 6 months ago by alex-orca
  2

• `contract` failed to find required methods from `jax.numpy` backend

  #224 opened 6 months ago by zazbone
  4

• Setup.py says install requires Numpy >= 1.7

  #223 opened a year ago by chabazite
  1

• contract_expression inconsistent with contract

  #218 opened a year ago by SebastianSemper
  0

• Setting memory_limit in contract_path causes "ValueError: invalid subscript 'Ų' in einstein sum subscripts string, subscripts must be letters"

  #217 opened a year ago by rht
  7

• Suggestion: Multiple einsum steps

  #216 opened a year ago by davystrong
  2

• Add Aesara as a backend

  #213 opened 2 years ago by rlouf
  4

• Contigugous loss after opt-einsum that may affect the efficiency of subsequent operations

  #211 opened 2 years ago by Geositta2000
  8

• Return type of opt_einsum.contract -- linter errors

  #212 opened 2 years ago by danielkelshaw
  2

• 'out' option in 'contract' makes a copy

  #209 opened 2 years ago by wxj6000
  6

• Way to keep tensor Ids from being rearranged to cretae custom paths

  #206 opened 2 years ago by LionFrangoulis
  2

• Fink optimal path for caching constant intermediates

  #199 opened 2 years ago by fangzhangmnm
  3

• Zero dimension tensors (as zero entries) lead opt_einsum slower than einsum

  #198 opened 2 years ago by Geositta2000
  1

• allow `backend=None` as shorthand for 'auto'

  #193 opened 2 years ago by jcmgray
  1

• UnicodeEncodeError in contract_path

  #182 opened 2 years ago by ChienKaiMa
  0

• Could `opt_einsum` understand repeated inputs?

  #194 opened 2 years ago by romanngg
  0

• Slow contractions when scalars are present and output is large

  #189 opened 2 years ago by Geositta2000
  3

• Empty step in contractions with scalar by dp

  #187 opened 2 years ago by Geositta2000
  1

• Reuse path including a scalar constant in the contraction

  #186 opened 2 years ago by Geositta2000
  2

• Possibilities to utilize symmetry in opt_einsum

  #185 opened 3 years ago by Lancashire3001
  1

• error in Test opt_enisum

  #183 opened 3 years ago by jiaminglei-lei
  2

• [question]: compute contraction order and provide order at future call

  #184 opened 3 years ago by rohany
  2

• How to use tblis to improve TDOT?

  #180 opened 3 years ago by Lancashire3000
  7

• Customized cost functions for faster contractions

  #171 opened 3 years ago by jcmgray
  4

• Use timeit to check the performance

  #178 opened 3 years ago
  1

• Optimizing a sum of einsums?

  #179 opened 3 years ago by yaroslavvb
  2

• Customize library in binary contractions

  #177 opened 3 years ago by Lancashire3000
  3

• Bug: Dynamic_programming path solver may return illegal contraction path

  #172 opened 3 years ago by refraction-ray
  5

• Contraction route "ipx, iqx -> pq" probably slow and not multi-threading most time

  #157 opened 3 years ago by ajz34
  5

• Could opt_einsum be employed in PyTorch when the torchscript was used to compile the code?

  #156 opened 3 years ago by zhangylch
  8

• Taking strides into account

  #162 opened 3 years ago by danpovey
  2

• when used with pytorch, is it possible to apply gradient checkpointing?

  #168 opened 3 years ago by seongwook-ham
  1

• Is there anything like np.show_config() in opt_einsum

  #173 opened 3 years ago
  1

• Do you provide a way to compute the output shape of an einsum?

  #170 opened 3 years ago by mariogeiger
  2

• paths.ssa_to_linear() might be buggy

  #169 opened 3 years ago by springer13
  1

• Is opt_einsum.paths.optimal algorithm correct?

  #167 opened 3 years ago by igtrnt
  7

• reusing contraction generated from contract_path leads to 'ValueError: Number of einsum subscripts must be equal to the number of operands.'

  #164 opened 3 years ago
  2

• Question: optimal contractions and batch dimension

  #158 opened 3 years ago by Linux-cpp-lisp
  6