This is a pure-TF implementation of tf.math.cumulative_logsumexp; unlike
the built-in op, it supports XLA compilation. It uses a similar algorithmic
technique (parallel prefix sum) as the built-in op, so it has similar numerics
and asymptotic performace. However, this implemenentation currently has higher
overhead, so it is significantly slower on smaller inputs (n < 10000).
Args
x
the Tensor to sum over.
axis
int Tensor axis to sum over.
name
Python str name prefixed to Ops created by this function.
Default value: None (i.e., 'cumulative_logsumexp').
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Missing the information I need","missingTheInformationINeed","thumb-down"],["Too complicated / too many steps","tooComplicatedTooManySteps","thumb-down"],["Out of date","outOfDate","thumb-down"],["Samples / code issue","samplesCodeIssue","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2023-11-21 UTC."],[],[],null,["# tfp.math.log_cumsum_exp\n\n\u003cbr /\u003e\n\n|----------------------------------------------------------------------------------------------------------------------------------------|\n| [View source on GitHub](https://github.com/tensorflow/probability/blob/v0.23.0/tensorflow_probability/python/math/generic.py#L93-L119) |\n\nComputes log(cumsum(exp(x))). (deprecated) \n\n tfp.math.log_cumsum_exp(\n x, axis=-1, name=None\n )\n\n| **Deprecated:** THIS FUNCTION IS DEPRECATED. It will be removed after 2023-03-01. Instructions for updating: `log_cumsum_exp` is deprecated; Use [`tf.math.cumulative_logsumexp`](https://www.tensorflow.org/api_docs/python/tf/math/cumulative_logsumexp) instead.\n\nThis is a pure-TF implementation of [`tf.math.cumulative_logsumexp`](https://www.tensorflow.org/api_docs/python/tf/math/cumulative_logsumexp); unlike\nthe built-in op, it supports XLA compilation. It uses a similar algorithmic\ntechnique (parallel prefix sum) as the built-in op, so it has similar numerics\nand asymptotic performace. However, this implemenentation currently has higher\noverhead, so it is significantly slower on smaller inputs (`n \u003c 10000`).\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n| Args ---- ||\n|--------|---------------------------------------------------------------------------------------------------------------------|\n| `x` | the `Tensor` to sum over. |\n| `axis` | int `Tensor` axis to sum over. |\n| `name` | Python `str` name prefixed to Ops created by this function. Default value: `None` (i.e., `'cumulative_logsumexp'`). |\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n| Returns ------- ||\n|------------------------|------------------------------------|\n| `cumulative_logsumexp` | `Tensor` of the same shape as `x`. |\n\n\u003cbr /\u003e"]]
View source on GitHub