spires_reservoir_reset

Reset all neuron states to zero without reallocating the reservoir or modifying its weight matrices.

Signature

spires_status spires_reservoir_reset(spires_reservoir *r);

Parameters

Returns

spires_status — SPIRES_OK on success, or SPIRES_ERR_INVALID_ARG if r is NULL.

Example

/* Run a first time series */
spires_run(r, input_a, len_a);
spires_train_ridge(r, input_a, target_a, len_a, 1e-6);
 
/* Reset state before processing a new, independent series */
spires_reservoir_reset(r);
 
/* Run a second time series with a clean state */
double *out = spires_run(r, input_b, len_b);
free(out);

Notes

• What is reset. All neuron membrane potentials and any internal history buffers (e.g., the fractional derivative memory for FLIF_* models) are set to zero. This is equivalent to the state immediately after spires_reservoir_create.
• What is preserved. The internal weight matrix W, the input weight matrix W_in, and the readout weight matrix W_out are all retained. Trained output weights survive a reset.
• Use case. Call this between independent input sequences to ensure that residual state from one sequence does not contaminate the next. This is essential when evaluating per-sequence classification accuracy.
• Thread safety. Do not call this function on a reservoir that is concurrently being used by another thread.

See Also

• spires_reservoir_create — create a new reservoir.
• spires_reservoir_destroy — free a reservoir entirely.
• spires_step — advance the reservoir by one timestep.