def _save_values(self, result_vec, objects, offset_map):
"""Saves the values of the optimal primal/dual variables.
Parameters
----------
results_vec : array_like
A vector containing the variable values.
objects : list
The variables or constraints where the values will be stored.
offset_map : dict
A map of object id to offset in the results vector.
"""
if len(result_vec) > 0:
# Cast to desired matrix type.
result_vec = self._DENSE_INTF.const_to_matrix(result_vec)
for obj in objects:
rows, cols = obj.size
if obj.id in offset_map:
offset = offset_map[obj.id]
# Handle scalars
if (rows, cols) == (1,1):
value = intf.index(result_vec, (offset, 0))
else:
value = self._DENSE_INTF.zeros(rows, cols)
self._DENSE_INTF.block_add(value,
result_vec[offset:offset + rows*cols],
0, 0, rows, cols)
offset += rows*cols
else: # The variable was multiplied by zero.
value = self._DENSE_INTF.zeros(rows, cols)
obj.save_value(value)
def _save_values(self, result_vec, objects, offset_map):
"""Saves the values of the optimal primal/dual variables.
Parameters
----------
results_vec : array_like
A vector containing the variable values.
objects : list
The variables or constraints where the values will be stored.
offset_map : dict
A map of object id to offset in the results vector.
"""
if len(result_vec) > 0:
# Cast to desired matrix type.
result_vec = intf.DEFAULT_INTF.const_to_matrix(result_vec)
for obj in objects:
rows, cols = obj.size
if obj.id in offset_map:
offset = offset_map[obj.id]
# Handle scalars
if (rows, cols) == (1, 1):
value = intf.index(result_vec, (offset, 0))
else:
value = intf.DEFAULT_INTF.zeros(rows, cols)
intf.DEFAULT_INTF.block_add(value,
result_vec[offset:offset + rows*cols],
0, 0, rows, cols)
offset += rows*cols
else: # The variable was multiplied by zero.
value = intf.DEFAULT_INTF.zeros(rows, cols)
obj.save_value(value)
def index_coeffs(lin_op):
"""Returns the coefficients for INDEX linear op.
Parameters
----------
lin_op : LinOp
The index linear op.
Returns
-------
list
A list of (id, size, coefficient) tuples.
"""
# Special case if variable.
if lin_op.args[0].type is lo.VARIABLE:
return index_var(lin_op)
key = lin_op.data
coeffs = get_coefficients(lin_op.args[0])
new_coeffs = []
for id_, size, block in coeffs:
# Index/slice constants normally.
if id_ is lo.CONSTANT_ID:
size = lin_op.size
block = intf.index(block, key)
# Split into column blocks, slice column blocks list,
# then index each column block and merge.
else:
block = get_index_block(block, lin_op.args[0].size, key)
new_coeffs.append((id_, size, block))
return new_coeffs
def index_coeffs(lin_op):
"""Returns the coefficients for INDEX linear op.
Parameters
----------
lin_op : LinOp
The index linear op.
Returns
-------
list
A list of (id, size, coefficient) tuples.
"""
# Special case if variable.
if lin_op.args[0].type is lo.VARIABLE:
return index_var(lin_op)
key = lin_op.data
coeffs = get_coefficients(lin_op.args[0])
new_coeffs = []
for id_, size, block in coeffs:
# Index/slice constants normally.
if id_ is lo.CONSTANT_ID:
size = lin_op.size
block = intf.index(block, key)
# Split into column blocks, slice column blocks list,
# then index each column block and merge.
else:
block = get_index_block(block, lin_op.args[0].size, key)
new_coeffs.append((id_, size, block))
return new_coeffs
def get_index_block(block, idx_size, key):
"""Transforms a coefficient into an indexed coefficient.
Parameters
----------
block : matrix
The coefficient matrix.
idx_size : tuple
The dimensions of the indexed expression.
key : tuple
(row slice, column slice)
Returns
-------
The indexed/sliced coefficient matrix.
"""
rows, cols = idx_size
# Number of rows in each column block.
# and number of column blocks.
col_selection = range(cols)[key[1]]
# Split into column blocks.
col_blocks = get_col_blocks(rows, block, col_selection)
# Select rows from each remaining column block.
row_key = (key[0], slice(None, None, None))
# Short circuit for single column.
if len(col_blocks) == 1:
block = intf.index(col_blocks[0], row_key)
else:
indexed_blocks = []
for col_block in col_blocks:
idx_block = intf.index(col_block, row_key)
# Convert to sparse CSC matrix.
sp_intf = intf.DEFAULT_SPARSE_INTERFACE
idx_block = sp_intf.const_to_matrix(idx_block)
indexed_blocks.append(idx_block)
block = sp.vstack(indexed_blocks)
return block
def get_index_block(block, idx_size, key):
"""Transforms a coefficient into an indexed coefficient.
Parameters
----------
block : matrix
The coefficient matrix.
idx_size : tuple
The dimensions of the indexed expression.
key : tuple
(row slice, column slice)
Returns
-------
The indexed/sliced coefficient matrix.
"""
rows, cols = idx_size
# Number of rows in each column block.
# and number of column blocks.
col_selection = range(cols)[key[1]]
# Split into column blocks.
col_blocks = get_col_blocks(rows, block, col_selection)
# Select rows from each remaining column block.
row_key = (key[0], slice(None, None, None))
# Short circuit for single column.
if len(col_blocks) == 1:
block = intf.index(col_blocks[0], row_key)
else:
indexed_blocks = []
for col_block in col_blocks:
idx_block = intf.index(col_block, row_key)
# Convert to sparse CSC matrix.
sp_intf = intf.DEFAULT_SPARSE_INTERFACE
idx_block = sp_intf.const_to_matrix(idx_block)
indexed_blocks.append(idx_block)
block = sp.vstack(indexed_blocks)
return block
def get_col_blocks(rows, coeff, col_selection):
"""Selects column blocks from a matrix.
Parameters
----------
rows : int
The number of rows in the expression.
coeff : NumPy matrix or SciPy sparse matrix
The coefficient matrix to split.
col_selection : list
The indices of the columns to select.
Returns
-------
list
A list of column blocks from the coeff matrix.
"""
col_blocks = []
for col in col_selection:
key = (slice(col * rows, (col + 1) * rows, 1), slice(None, None, None))
block = intf.index(coeff, key)
col_blocks.append(block)
return col_blocks
def get_col_blocks(rows, coeff, col_selection):
"""Selects column blocks from a matrix.
Parameters
----------
rows : int
The number of rows in the expression.
coeff : NumPy matrix or SciPy sparse matrix
The coefficient matrix to split.
col_selection : list
The indices of the columns to select.
Returns
-------
list
A list of column blocks from the coeff matrix.
"""
col_blocks = []
for col in col_selection:
key = (slice(col * rows, (col + 1) * rows, 1), slice(None, None, None))
block = intf.index(coeff, key)
col_blocks.append(block)
return col_blocks
