Synthesizer produces error when mixing certain argument types

Required prerequisites

Consult the security policy. If reporting a security vulnerability, do not report the bug using this form. Use the process described in the policy to report the issue.
Make sure you've read the documentation. Your issue may be addressed there.
Search the issue tracker to verify that this hasn't already been reported. +1 or comment there if it has.
If possible, make a PR with a failing test to give us a starting point to work on!

Describe the bug

When running a test program, the synthesizer produces the following error:

terminate called after throwing an instance of 'std::bad_alloc'
  what():  std::bad_alloc
Aborted

Steps to reproduce the bug

Compile and run the following code using nvq++ --target nvqc path/to/file.cpp:

#include <ctime>
#include <cudaq.h>
#include <cudaq/algorithm.h>
#include <cudaq/optimizers.h>
#include <random>

struct kernel {
  auto operator()(const int n_qubits, const std::vector<double> &parameters) __qpu__ {
    cudaq::qvector qubits(n_qubits);
    for (size_t i = 0; i < n_qubits; i++)
      rx(parameters[i], qubits[i]);
  }
};

std::vector<double> initial_parameters(int n_parameters, int seed) {
  std::default_random_engine generator(seed);
  std::uniform_real_distribution<float> distribution(0.0, 1.0);
  std::vector<double> parameters(n_parameters);
  for (size_t i = 0; i < n_parameters; i++)
    parameters[i] = distribution(generator);
  return parameters;
}

int main() {

  const int n_qubits = 26;
  const int n_parameters =  n_qubits;
  std::vector<double> parameters = initial_parameters(n_parameters, 13);
  auto h = cudaq::spin::z(0);
  auto exp_val = cudaq::observe(kernel{}, h, n_qubits, parameters);
  printf("Expectation Value: %f \n", exp_val.expectation());

  return 0;
}

Expected behavior

The program should run without errors.

Is this a regression? If it is, put the last known working version (or commit) here.

Not a regression

Environment

CUDA Quantum version: 8dd93ad
Python version:
C++ compiler:
Operating system:

Suggestions

I believe the problem has something to do with offset not being correct in this line of code:

cuda-quantum/lib/Optimizer/Transforms/QuakeSynthesizer.cpp

Line 387 in 8dd93ad

char *ptrToSizeInBuffer = static_cast<char *>(args) + offset;

.

I think that offset was incremented by 4 to skip past the 32-bit value, but the vector is actually beginning at offset=8 instead of offset=4.

The synthesizer isn't taking into account the structure of the data buffer correctly.

I'll work on a fix.