tensorstate.States

The States module contains functions that operate on state space.

compress_states(states)

Compress a state space tensors

This function quantizes neurons into firing (>0) or non-firing (<=0), then compresses the bits into uint8 values. Thus, if a layer has 8 neurons in it, then the output is a raw byte ranging in value from 0-255. This compresses the statespace by 32x relative to holding values as floats, or by 8x relative to holding values as boolean. This is an important consideration since state space is large and grows exponentially with the number of neurons in the layer.

Parameters:states ([numpy.ndarray]) – A 2d array of neuron outputs as numpy.float32 or np.bool_ values, where columns are a particular neuron’s value, and rows are states.
Returns:
A 2d array of uint8 values, where each value is the
compressed representation of the state.
Return type:numpy.ndarray
decompress_states(states, num_neurons)

Decompress states to numpy array of booleans

This functions takes a 2d numpy array of compressed neuron states and returns a boolean array of states, where each column of values represents the state of an individual neuron (firing=True, non-firing=False).

For example, take a neuron layer with 5 neurons. The compressed state will be represented by a single byte, and if all but the first neuron is firing then the bits will be set as follows:

'00011110'

To decompress this, the number of neurons needs to be input to know how many of the bits are actual neuron representations. When this state is decompressed, the numpy array will be:

[False, True, True, True, True]

Parameters:
  • states ([numpy.ndarray]) – A 2d array of compressed states. See compress_states function.
  • num_neurons ([int]) – The number of neurons in the layer.
Returns:

Boolean numpy array of neuron states
Return type:

decompressed_states ([np.ndarray])
sort_states(states, state_count)

Sort the states to place identical states next to each other

This function sorts the states stored in a 2d numpy.ndarray so that identical states are placed next to each other. To increase speed, the states are not actually sorted since moving data around in memory can be time consuming, and usually not useful. What is returned is a sorted index and the location of unique states in the sorted index.

Parameters:
  • states ([numpy.ndarray]) – A 2d array of compressed states. See compress_states function.
  • state_count ([int]) – The number of states (or number of rows to sort).
Returns:

Bin edges, or locations of unique states index ([np.ndarray]): Sorted index. This output can be used to actually

sort the input states by doing states[index]
Return type:

edges ([np.ndarray])